Citation
An empirical analysis of open source software developers' motivation using expectancy-valence theory

Material Information

Title:
An empirical analysis of open source software developers' motivation using expectancy-valence theory
Creator:
Wu, Chorng-Guang
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
xv, 208 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Computer software developers -- Attitudes ( lcsh )
Open source software ( lcsh )
Motivation (Psychology) ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 177-208).
Thesis:
Computer science and information systems
General Note:
Department of Computer Science and Engineering
Statement of Responsibility:
by Chorng-Guang Wu.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
182721548 ( OCLC )
ocn182721548
Classification:
LD1193.E52 2007d W82 ( lcc )

Downloads

This item has the following downloads:


Full Text
AN EMPIRICAL ANALYSIS OF OPEN SOURCE SOFTWARE
DEVELOPERS MOTIVATION USING
EXPECTANCY-VALENCE THEORY
by
Chomg-Guang Wu
B.B.A., National Cheng Kung University, Taiwan, 1985
M.S., State University of New York at Binghamton, 1991
A thesis submitted to the
University of Colorado at Denver and Health Sciences Center
in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
Computer Science and Information Systems
2007


This thesis for the Doctor of Philosophy
degree by
Chomg-Guang Wu
has been approved
by
James H. Gerlach
Ronald V. Ramirez
7- lz-ol
Date


Wu, Chomg-Guang (Ph.D., Computer Science and Information Systems)
An Empirical Analysis of Open Source Software Developers Motivation
using Expectancy-Valence Theory
Thesis directed by Prof. James Gerlach
ABSTRACT
The purpose of this study was to investigate the motivations of
individuals that are willing to join open source communities and voluntarily
dedicate their effort and expertise for OSS development. Despite the
emergence of various studies on developers motivation, there is little research
that focuses on developers intentions to continue their participation and
expend their effort. Expectancy-valence theory, human resource theory, and
volunteerism were adapted to form two research models: a theoretical model
of OSS developers continuation, and an exploratory model of developers
effort.
The continuation model captures those factors that might significantly
influence OSS developers satisfaction with OSS development, their
commitment and intentions to continue participating in future open source
projects, and the causal relationships among these factors. The effort model


is developed to measure the associations between developers effort and
motivation, schedule availability, and expertise.
The findings regarding the motivation of developers showed that
volunteering can reflect multiple motivations: intrinsic motivation (i.e.,
helping) and extrinsic motivation (i.e., career advancement). A positive and
satisfactory experience leads to a positive attitude toward retention.
Moreover, volunteer developers commitment primarily depends on
motivation to help. These results, taken as a whole, suggest that OSS
developers are motivated both by the altruism of helping and the economic
incentive for career advancement.
Developers effort was primarily influenced by both their motivation to
help and peer recognition, which implies that the impact of intrinsic
motivators on effort is stronger than that of extrinsic motivators. Enjoyment
demonstrated a significantly negative effect on effort, suggesting that
developers with high levels of motivation on enjoyment did not spend more
time on open source projects than those with less motivation on enjoyment.
This study also provided empirical evidence that volunteers may vary
their motivation according to their actual experience and the commercial
viability of their project. The length of experience suggests a life cycle effect
on motivation; i.e., the longer developers stay, the higher levels of motivation
grow. In addition, the more commercially viable an open source project is,
the more likely developers are motivated. Finally, scheduling availability is
an important factor of predicting continuation and effort.


This abstract accurately represents the content of the candidates thesis. I
recommend its publication.


ACKNOWLEDGEMENT
I would like to acknowledge the members of my doctoral committee for
helping me make this dissertation possible. First of all, I would especially
like to express my sincerest gratitude and appreciation to my adviser,
Professor Jim Gerlach, for his mentorship and commitment. Throughout my
doctoral work, Jim gave me the freedom to explore on my own and at the
same time the guidance to recover when my steps faltered. He continually
encouraged me to develop independent thinking and research skills. His
patience and support helped me overcome many crisis situations and finish
this dissertation. I hope that one day I would become as good an advisor to
my students as Jim has been to me.
I would also like to thank Professor Clifford Young for assisting me with
the quantitative analysis. Professor Youngs insightful comments were
thought-provoking and they helped me improve my knowledge in the area of
structural equation modeling. I am very thankful to him for enforcing strict
validations for each research result.
I am also very grateful to Dr. Ramirez and Dr. Cios for reading this
dissertation and offering constructive comments. In addition, I owe a special
note of gratitude to Dr. Mannino for his generous help during my study at this
University.


Finally, I would like to thank my family for their life-long love and
support. Especially, my mother has been a constant source of concern,
support and strength all these years. I would like to express my deepest
gratitude to her.


TABLE OF CONTENTS
Figures........................................................xiii
Tables.........................................................xiv
CHAPTER
1. INTRODUCTION................................................1
Purpose of the Study....................................5
Statement of the Problem................................6
Research Questions......................................8
Readers Guide..........................................9
2. LITERATURE REVIEW OF OPEN SOURCE SOFTWARE..................13
Definition of Open Source Software.....................13
Evolution of Open Source Development...................15
Types of Open Source Software Licenses.................21
The BSD License....................................22
General Public License.............................23
Lesser General Public License......................24
Mozilla Public License.............................25
IBM Public License.................................26
Open Source Communities................................27
Open Source Participants...........................27
Structure of Open Source Communities...............28
Mode of Software Development.......................29
viii


Process of Submissions Acceptance...................31
Communities of Practice.............................32
Private-Collective Innovation.......................34
Prior Research on OSS Developers........................36
Studies on Motivation...............................36
Intrinsic Motivation............................37
Helping Behavior.......................... 38
Enjoyment..................................40
Peer Recognition...........................41
Extrinsic Motivation............................42
Enhancing Human Capital....................42
Career Advancement.........................43
Satisfying Personal Needs..................44
Effort and Performance..............................45
Effort..........................................45
Performance.....................................47
Example Models......................................48
Wu, Gerlach, and Young model (2007).............49
Roberts, Harm, and Slaughter model (2006).......50
Summary.................................................53
3. LITERATURE REVIEW OF VOLUNTEERISM............................55
Volunteerism............................................56
Motivation to Volunteer.................................58
ix


Reasons for Volunteering..............................58
Functional Approach...................................60
Paid v.s. non-Paid....................................63
Satisfaction...............................................65
Organizational Commitment..................................68
Retention and Turnover.....................................71
Retention.............................................71
Turnover..............................................73
Determinants of Volunteers Effort.........................75
Example Models.............................................78
Mowen and Sujan Model (2005)..........................78
Miller, Powell, and Seltzer model (1990)..............80
Dailey model (1986)...................................83
Empirical Findings.........................................84
Summary....................................................88
4. THEORETICAL PERSPECTIVES.......................................90
Motivation.................................................90
Expectancy and Valence................................93
Satisfaction...............................................95
Open Source Commitment.....................................97
Continuance Intention......................................98
Availability..............................................100
Expertise.................................................101
x


The Influential Factors of Motivation.......................102
Project Characteristics................................102
Years of OSS Experience................................103
Participation Status...................................103
Summary....................................................104
5. RESEARCH MODELS.................................................105
Continuation Model.........................................105
Effort Model...............................................114
6. METHODOLOGY.....................................................118
Identification of the Research Models......................118
Continuation Model.....................................120
Effort Model...........................................123
Instrumentation............................................125
Pilot Test.................................................127
Sampling and Data Collection...............................130
Data Analyses..............................................134
Reliability............................................135
Measurement Model Assessment...........................137
Continuation Model................................138
Effort Model......................................140
Structural Model Assessment............................143
Continuation Model................................143
Effort Model......................................146
XI


The Analysis of Influential Factors of Motivation.148
Project Characteristics.......................148
Years of OSS Experience.......................150
Participation Status..........................152
Limitations...........................................153
7. DISCUSSION AND CONCLUSIONS.................................157
Summary of Results....................................157
Implications..........................................160
Additional Research...................................164
Contributions.........................................165
Conclusions...........................................166
APPENDIX
A. HUMAN SUBJECTS APPROVAL...................................168
B. SURVEY ITEMS..............................................169
C. INFORMED CONSENT..........................................175
BIBLIOGRAPHY.......................................................177
xii


LIST OF FIGURES
Figure
1.1 Conceptual Diagram......................................................4
1.2 Readers Guide..........................................................9
2.1 Open Source Participants Continuation Model...........................49
2.2 OSS Developers Motivation and Performance Model.......................51
3.1 Trait plus Functional Motive Model.....................................79
3.2 Hospital Volunteers Turnover Model.....................................81
3.3 Volunteer Workers Organizational Commitment Model......................83
5.1 Research Model of OSS Developers Continuance Intention...............106
5.2 Research Model of OSS Developers Effort..............................114
6.1 Measurement Model of OSS Developers Continuation Model................121
6.2 Measurement Model of OSS Developers Effort Model......................124
6.3 Standardized LISREL Estimations of the Continuation Model.............145
6.4 Standardized LISREL Estimations of the Effort Model...................147
6.5 LISREL Estimations of Project Characteristics to Motivations..........149
6.6 LISREL Estimations of OSS Experience to Motivations...................151
xiii


LIST OF TABLES
Table
2.1 Important events in open source history.................................15
2.2 Characteristics of open source licenses.................................22
2.3 Major participants in open source communities...........................28
2.4 Major motivations of OSS developers.....................................37
3.1 Motives for volunteering.................................................59
3.2 Research findings on volunteerism.......................................85
5.1 Construct definitions for continuation model...........................107
5.2 Construct definitions for effort model.................................115
6.1 Types and number of parameters for the continuation model..............122
6.2 Types and number of parameters for the effort model....................125
6.3 Descriptive statistics and Cronbachs alpha for the pilot test.........129
6.4 Survey respondents position in OSS projects hosted by SourceForge.......132
6.5 Descriptive statistics of indicator variables for the continuation model.133
6.6 Descriptive statistics of indicator variables for the effort model.....134
6.7 Reliability measurements for constructs................................136
6.8 Factor loadings for the continuation model.............................139
6.9 AVE and correlation of constructs for the continuation model...........140
6.10 Factor loadings for the effort model...................................141
6.11 AVE and correlation of constructs for the effort model.................143
XIV


6.12 Testing for the equivalence of factor loadings across two groups.......152
6.13 Invariant mean tests of motivational constructs........................153
XV


CHAPTER 1
INTRODUCTION
Software practitioners and academic researchers are captivated by the
recent success of Open Source Software (OSS) in producing a number of
highly successful software products such as Linux, Apache, MySQL, Perl,
Sendmail, and Firefox. Traditionally, due to trade secret protection by the
computer vendors, commercial software products are sold only in binary form
and with no source code. Thus, the users have no rights to examine or
modify the code. In contrast to traditional software development, adopters of
OSS seek low cost solutions that they can modify or use as software
components to build new or extended applications (Johnson 2002; Lemer and
Tirole 2002; Appelbe 2003).
OSS is developed by a loosely organized community of developers
spread throughout the world and working over the Internet. Unlike a
traditional organizational structure, these communities have no/few paid staff
or management. However, they do provide their members with a variety of
resources to develop practical and helpful software products that have
occasionally displaced or significantly improved upon some commercial
products. Remarkably, OSS developers contribute to software projects
without necessarily being employed or recruited by any organization.
Although the recent trend is for companies that benefit from OSS to loan


employees to this effort, surveys have shown that the majority of OSS
developers contribute strictly voluntarily (Hars and Ou 2002; Lemer and
Tirole 2002; Wu et al. 2007).
The open source community is often described as a gift culture instead of
an exchange culture. Essentially, gift culture refers to helping behavior
which includes altruism and reciprocity. In lieu of tangible rewards, givers
receive psychological benefits such as the warm glow of sympathy or the
satisfaction of living up to a moral commitment (Rose-Ackerman 1998).
Others assert that additional intrinsic rewards such as boosting ones ego,
self-determination, and community identification, or even an anti-Microsoft
mentality, might provide sufficient intrinsic motivation to engage in OSS
development.
However, economists argue that it is very unlikely that intrinsic rewards
provide sufficient incentives to explain the enormous contributions in time and
effort that some participants donate to OSS since there are no obvious similar
patterns of behavior in most other areas of economic activity (Schmidt and
Schnitzer 2003). Many beneficiaries are well-to-do and could afford to pay
OSS participants for their efforts. Furthermore, altruism or reciprocity has
not played a major role in other industries, so the reason individuals in the
software industry are more altruistic or reciprocal than others would need to be
explained. In contrast, individual participants might contribute to an open
source project to increase human capital, develop career opportunities, and
obtain software that solves specific technical problems related to personal job
2


requirements (Hann et al. 2002; Hars and Ou 2002; Lemer and Tirole 2002; Ye
and Kishida 2003; Wu et al. 2007).
Exploratory empirical research on the motivations of OSS developers
demonstrates the validity of some of these beliefs. However, these
arguments have not been tested rigorously using existing human resource
theory and volunteerism. Moreover, there is little research investigating OSS
developers work attitudes (e.g., satisfaction and commitment) and intentions
to engage in future open source projects once they have already become
involved. The sustainability of an open source project would generally rely
on developers continual contributions. Thus, understanding OSS
developers intentions to continue their involvement in OSS projects is very
important for predicting the long-term viability of OSS development. These
aspects are presented conceptually in Figure 1.1.
Open source communities cannot exist or prosper without the
contributions of highly motivated developers who are willing to donate their
effort (i.e., time) and expertise. However, because these developers are often
volunteers, not traditional employees, it is impossible to solely rely on
employment relationships or employment contracts to manage these persons
(Roberts et al. 2006). Effort has been used in a number of previous OSS
studies (Hars and Ou 2002; Hertel et al. 2003; Lakhani and Wolf 2003) and
provides an appropriate proxy for developers contribution and interest in OSS
projects (Lakhani and Wolf 2003). The literature suggests that different
motivators for contributing to open source projects should have diverse
3


influences on developers efforts. With respect to the success of an OSS
project, it is imperative to understand whether all types of motivations affect
OSS developers effort equally or in the same way.
According to Olson et al. (1996), individuals persist longer and put more
effort on tasks in which they expect to succeed. Persons who believe that
their skills set is adequate for achieving success with a new venture are
motivated to exert the necessary effort (Douglas and Shepherd 2000; Shaver et
al. 2001). Therefore, it is also important to examine the relationship between
OSS developers expertise and the effort dedicated to open source projects.
Availability has been regarded as a significant determinant of
tumover/retention in the workplace or volunteer activities. According to the
4


volunteerism literature, many volunteers claimed that they had limited free
time dedicated to volunteer activities because of existing family and
professional commitments (Ralston and Rhoden 2005). Rotolo and Wilson
(2006) indicated that more flexible work schedules make it easier for
volunteers to commit to volunteer work. The more convenient the schedule,
the less likely the volunteer is to intend to or actually quit (Miller et al. 1990).
Thus, availability is examined to test whether it will influence OSS
developers effort and intention to continue their work. Likewise, the
arguments regarding the relationships of OSS developers continuance
intention with availability and the associations of effort with motivation,
expertise, and availability are conceptually illustrated in Figure 1.1.
Because monetary reward is lacking in the context of OSS development,
motivational processes underlying developers behavior are likely to differ
from those in traditional employment. Thus, this study also analyzes several
factors (i.e., project characteristics, the length of OSS experience, and
participation status) that may enhance developers motivation.
Purpose of the Study
This study seeks to investigate OSS volunteer developers motivation,
which in turn can provide a better understanding of what motivates those
developers to continue making OSS contributions and to expand their effort.
An additional purpose is to explore the impact of project characteristics
(commercial viability), years of OSS experience, and participation status
5


(volunteer or paid) on motivation among OSS developers. The following
four aspects are relevant to the investigation of these issues.
First, with respect to continuance intention, the research attempts to
measure developers motivation, to identify the antecedents of developers
satisfaction with participation in open source projects and commitment to OSS
development, and to analyze the relationships of these factors and availability
with developers intentions to continue their involvement in future projects.
Second, to evaluate developers efforts, motivation, expertise, and
availability for OSS development are taken into consideration. How these
factors influence developers effort will be considered and explored.
Third, project commercial viability and years of OSS experience are
examined, since they may influence developers motivation based upon
volunteerism theory.
Fourth, theoretical arguments underpinning this study are based on
expectancy-valence theory, human resource theory, and volunteerism.
Statement of the Problem
According to the literature, each individual has attitudes on many
different topics, such as politics, education, and work. Work attitudes refer to
a collection of feelings, beliefs, and thoughts about how to behave with respect
to the jobs people currently hold and the organizations thereof (George and
Jones 1996, p. 66). Theoretically, there are two major components of work
related attitudes: satisfaction and organizational commitment.
6


The extant open source literature regarding individual OSS developers
focuses on the analysis of their motivation to get involved in open source
projects. There is a shortage of empirical investigation concerning their
attitudes toward OSS development after participating in open source projects
and their efforts associated with motivation, availability, and expertise. Are
OSS developers satisfied with OSS development? To what extent are the
developers committed to OSS development? Will OSS developers continue
to contribute to future open source projects even if they are dissatisfied with
the experience? What are the relationships between motivators, satisfaction,
commitment, availability, and continuance intention? Which types of
motivations are likely to generate more (or less) effort? What impact will
availability and expertise have on developers effort? These questions need
to be investigated with the open source context.
In this study, there are two research models: continuation model and
effort model. By applying the expectancy-value theory, the continuation
model captures those factors that might significantly influence OSS
developers satisfaction with OSS development, their commitment and
intentions to continue participating in future open source projects, and the
causal relationships among these factors. The effort model is developed to
measure the associations between developers effort and motivation, schedule
availability, and expertise.
7


Research Questions
Question 1: Does commitment to OSS development significantly influence
OSS developers continuance intention?
Question 2: Does satisfaction with OSS development significantly influence
developers commitment to OSS development and/or continuance
intention?
Question 3: What motivators will have a significant effect on OSS developers
satisfaction, commitment, and continuance intentions?
Question 4: What is the impact of future schedule availability on OSS
developers continuance intention?
Question 5: What motivators will have a significant effect on OSS developers
effort?
Question 6: What is the impact of current schedule availability on OSS
developers effort?
Question 7: What is the impact of expertise on OSS developers effort?
Question 8: Does project commercial viability influence OSS developers
motivation?
Question 9: Does the length of OSS experience influence developers
motivation?
Question 10: Does volunteer developers motivation differ from that of paid
developers?
8


Readers Guide
Part I: Literature Review
Figure 1.2 Readers Guide
9


The dissertation is divided into four major parts, shown in Figure 1.2.
Part I refers to an understanding of the open source software movement and
OSS developers as well as the behavior of volunteers. In this part, I present
the literature review on open source software (Chapter 2) and volunteerism
(Chapter 3).
Chapter 2 presents OSS definition, history, licenses, community structure,
developers motivation and effort as well as performance, two example
research models related to developers motivation, and a summary of
empirical findings. These aspects highlight the prevalent issues related to
extant OSS research.
Chapter 3 examines existing research on volunteerism. It covers
volunteers motivation, work attitudes (satisfaction and organizational
commitment), retention as well as turnover, effort, three research models
associated with volunteer behavior, empirical findings, and it concludes with a
summary. These perspectives are valuable in understanding the issues raised
by OSS developers volunteer behavior and they emphasize aspects that are
important in volunteers work motivation. This chapter also raises theoretical
implications for the exploration of paid versus non-paid volunteers and the
issues regarding volunteer effort.
In Part II, I introduce the base theories related to this study and depict
how the theories are associated with OSS developers behaviors (Chapter 4).
In addition, I describe the development of two research models: continuation
and effort for this study based on the ground theories (Chapter 5).
10


Based upon the expectancy-value theory, human resource theory, and
volunteerism, Chapter 4 presents extant research on traditional paid employees
and volunteer workers, and it explicates how these studies can be applied to
analyzing OSS developers. The topics include the expectancy-value theory
on work motivation, the human resource studies on work attitudes and
behavior (i.e., satisfaction, commitment and continuation), schedule
availability from volunteerism, expertise from human resource management,
and the influential factors of motivation from volunteerism. The issues and
theoretical arguments in Chapter 4 were chosen in order to illuminate different
aspects of the base theories.
Chapter 5 demonstrates two research models: the continuation model and
the effort model. It explains the proposed models and the interdependence
among these constructs in detail. Moreover, hypotheses for each model are
proposed for testing the theoretical arguments in this study.
In Part III, I describe the research design based upon quantitative
methods used to measure the research models (Chapter 6).
Chapter 6 exhibits the research methodology used in this study. This
chapter contains the model identification, the survey developed for collecting
information from existing OSS developers, and data analysis. The processes
of instrument design, pilot test for the instrument, sampling, and data
collection are all discussed. The data analysis refers to how to measure the
reliability of constructs and validate the research models. Furthermore,
hypothesis testing and the limitations of this research design are spelled out.
11


In Part IV, I summarize the research findings, provide several suggestions,
and conclude this study (Chapter 7). Chapter 7 presents the discussion
regarding the outcomes generated from the empirical testing of research
models and the implications for academia and practitioners. This chapter
also discusses the contributions of this studyand several issues thereinto
future research on OSS development.
12


CHAPTER 2
LITERATURE REVIEW OF OPEN SOURCE SOFTWARE
The review of related literature on OSS is comprised of two phases.
First, this treatise contains a definition of OSS, a brief history of open source
development, software licenses, open source community structure, and several
structural/procedural issues related to OSS development, such as development
mode, acceptance process, communities of practice, and private-collective
innovation. Second, prior studies of OSS developers are reviewed, including
studies on developers motivation for participation in OSS projects, their effort
as well as performance in relation to OSS development, and certain research
models developed to empirically test the relationship among motivation and
developers behavior.
Definition of Open Source Software
Significantly different from traditional closed source software whose
source code is protected as a commercial secret by the software vendor, OSS
means that the source code written in any programming language should be
freely available to any person who requests it. According to Open Source
Definition1, the major implication of open source software primarily lies in
the condition that anyone should be able to unrestrictedly access computer
1 http://www.opensource.org, Open Source Definition
13


program code, modify it and redistribute the modifications at zero or nominal
cost. This allows any individual to make software improvements and fixes.
OSS has also been called free software, and both terms have been used
interchangeably. Actually, free in free software mainly emphasizes the
free(dom) to modify the programs source code, rather than at no cost
(Appelbe, 2003). Just as used by the Free Software Foundation, free
essentially refers to the freedom to access, modify, and redistribute the
softwares source code. Therefore, this potential ambiguity in the meaning of
the term free software has led to open source software becoming the more
common use term. On the other hand, open source software is also
distinguished from freeware and shareware since freeware is distributed for
free but users do not get access to the source code and have no right to modify
or extend the software and shareware is often offered for free for a trial period
(Schmidt and Schnitzer, 2003).
In summary, open source software has three main features (1) the
source code must be available to anyone, (2) anyone may modify the source
code or derive new works from it, and (3) there are no restrictions on
redistributing the software; that is, anyone may give away or charge the
software for a nominal price, and companies are free to package and sell
products containing OSS, and consumers who buy these products are free to
redistribute that OSS at no cost if they desire so.
14


Evolution of Open Source Development
The open source phenomena have lasted for a half century. Table 2.1
demonstrates the major events in the history of open source.
Table 2.1 Important events in open source history
Era Event
1950s to Early 80s Software source code is distributed without restrictions in some scientific computing communities (e.g. ACMs algorithm section) and user groups among computer vendors such as IBM, DEC. The very first version of Unix and C Language were introduced, and their source code was freely available. BSD (Berkeley Software Distribution) Unix was freely distributed. Sendmail was originally developed by Eric Allman, a computer science student at UC-Berkeley, in the late 70s.
Early 1980s to Late 80s Ricard Stallman, a programmer at the MIT artificial intelligence laboratory, established the Free Software Foundation in 1983, and published GNU2 (a recursive acronym for GNU's Not Unix) manifesto forming GPL (General Public License) to seek for making software freely available. Perl (Practical Extraction and Reporting Language) was created in 1987 by Larry Wall, a programmer at Burroughs (a computer mainframe manufacturer now part of Unisys). The source code of Minix, a version of Unix for PC and Mac, was released in 1987 by its developer, Andrew Tanebaum.
2 GNU is a computer operating system composed entirely of free software. Its name is a
recursive acronym for GNUs Not Unix, which was chosen because its design is Unix-like,
but differs from Unix by being free software and by not containing any Unix code. GNU
was founded by Richard Stallman and was the original focus of the Free Software
Foundation (FSF).
15


Table 2.1 (Cont.)
Era Event
Early 1990s to Late 90s Linux, a Unix variant based on Minix, was created in 1991 by Linus Torvalds, a Finnish computer science student at University of Helsinki. FreeBSD 1.0 was released in 1993. Debian Linux, a new Linux distribution, was created by Ian Murdock in 1993. The development of Apache began in 1994. Marc Ewing founded Red Hat, the leading Linux distributor, in 1994. Star Office, the equivalent of Microsofts Office suite, was originally developed in mid 90s by SUN Microsystems, and its introduction represented that OSS development started to cover desktop applications. Debian Free Software Guidelines developed in 1995, were adopted in early 1997 by a number of OSS developers and subsequently became known as Open Source Definition. Netscape in 1998 gave away the source code of its Communicator 5.0, which eventually became an open source project.
2000 to Today December 11 2000, IBM Chairman and CEO Lou Gerstner announced that the company would invest $1 billion in Linux development for promoting the diffusion of its e-business platforms. Open Office, which evolved from Star Office is now 99.9% Word and PowerPoint compatible.
Source: Barahona et al. 1999; Hars and Ou 2002; Lemer and Tirole 2002; Appelbe 2003
Open source softwares origins can be dated from the earliest days of
software development in the 1950s. Then, software source code developed
by some scientific programming communities such as academic/research
institutions was frequently and mutually given away to other similar
organizations without any restrictions. As a result, this kind of benevolent
behavior let those software beneficiaries also contribute their own
16


improvements back to the whole community just as scientists publish or give
away their research results so that other scientists can build on their results to
further continual innovation (Barahona et al, 1999; Comerford, 1999;
Dempsey et al, 2002). According to Appelbe (2003), there is an implicit
honor code in science, which says that all research results and related
developments should be shared with the scientific community scientists view
colleagues who are not willing to share their results and work openly with
deep suspicion. Thus as early as the late 1950s, scientific software was being
freely distributed.
During the 1960s and 1970s, when proprietary software flourished all
over the world, two major software products emerged as the most important
milestone of OSS development, i.e. Unix and C Language. Unix, an
operating system capable of running on multiple platforms, and C, a
programming language originally developed at AT&Ts Bell Labs in the early
1970s, were available for distribution in academia and research labs at zero or
nominal cost with source code, i.e. open source. From then on, many of the
sites where Unix and C Language were installed made further innovations,
which were in turn shared with others (Lemer and Tirole, 2002). For
instance, by using this open source, the University of California at Berkeley
enhanced Unix in the late 1970s, and consequently produced the BSD
(Berkeley Software Distribution) of Unix whose source code were freely
distributed and widely used in academia.
17


Since the early 1980s, more and more computer vendors had adapted
BSD Unix to run on their machines, which led to many of the early
commercial versions of Unix that were based on BSD. Due to the growth of
Unixs commercial success, AT&T then addressed a legal problem concerning
its intellectual property rights related to Unix. This issue evolved as a
controversial topic at that moment. Those maintaining BSD Unix had to
redevelop the operating system from scratch to prevent from violating AT&Ts
intellectual property right. Nonetheless, a free version of BSD Unix was
eventually released in the late 1980s and it is still freely available in source
code form open source.
Another important event in the history of open source occurred in the mid
1980s. Richard Stallman, a programmer from MITs Artificial Intelligence
Lab, founded the Free Software Foundation (FSF) in 1984 since he was
extremely dissatisfied with the trend for more and more software to be
proprietary and strongly believed in the importance of sharing source code
without limitation.3 Through the efforts of Stallman and a group of
volunteers around the world, these open source software contributors not only
initialized the GNU (a recursive acronym for Gnus Not Unix) project to
implement a totally free version of the Unix operating system but built a
variety of free versions of software utilities and development tools as a
preliminary to developing the operating system itself. According to
Raymond (2001), for more than a decade after its founding, FSF would largely
3 http://www.gnu.org
18


define the public ideology of the hacker culture, and Stallman himself would
be the only credible claimant to leadership of the tribe. Apparently,
Stallmans endeavor has been regarded as an open source software crusade
because he has constantly advocated the importance of freely available source
code.
Throughout the 1990s, with the prevalent usage of the Internet and the
World Wide Web, the development of open source software stepped into a new
era. Especially, the presence of Linux, a new Unix variant with an amalgam
of Linus and Unix, brought open source activity unprecedented prosperity.
In 1991, Linus Torvalds, a Finnish computer science student at Helsinki
University, started creating an open source Unix-type kernel from scratch, and
afterward encouraged others contributions in a sequence of postings to online
bulletin boards for free implementation of the software. Because of previous
BSD Unixs legal problems, many people interested in running Unix on their
personal computers turned to Linux. Consequently, the community
dramatically spread, and Linux soon became one of the most popular and
leading open source software projects. The key to Linuxs success has two
important factors: (1) Torvalds opened it up for community improvement
and development, and the software community jumped on the bandwagon; (2)
the rise of the Internet provided software developers with a very expedient tool
to communicate, collaborate, and distribute software (Appelbe, 2003).
In addition to operating systems, open source development of application
software was also launched in the mid 1990s. One of the most well-known
19


products was Star Office, the comparable of Microsofts (MS) Office suite
developed by SUN Microsystems in the mid 1990s. Although Star Offices
functionalities could not compete with those of MS Office and live up to its
original expectation of taking up some of the office desktop market, its
derivative Open Office about 99.9% Word and PowerPoint compatible had
been able to catch up with the majority of MS Offices features by the early
2000s.4
With respect to databases, the performance of mySQL, an open source
database management system, is close to that of commercial closed source
products such as Oracle and DB2. In recent times, even on the enterprise
resource planning side, an open source product Compiere ERP/CRM has
also emerged as a strong competitor to proprietary software such as Peoplesoflt.
Actually, more and more open source products have taken dominant positions
in their relevant areas. For instance, Apache takes the leading position in the
web server market (Hann et al. 2002; Lemer and Tirole 2002; Appelbe 2003).
Sendmail, an open source e-mail transfer agent, handles about three quarters
of all Internet e-mail traffic around the world (Lemer and Tirole 2002). Due
to the growth in popularity of open source, a number of websites such as
SourceForge.net, Collab.net and bkbits.net are devoted to making it easier to
find, download, install and maintain a wide variety of open source software
(Appelbe 2003).
4 http://www.openofFice.org
20


To date, there has been almost no arena of software for which there is not
an open source alternative to proprietary software (Appelbe 2003). Moreover,
many governments worldwide have also begun legislative intervention to
foster the open source movement and to spread the usage of open source
software in public administration and educational institutions (Schmidt and
Schnitzer 2002; Appelbe 2003).
Types of Open Source Software Licenses
Although the source code of OSS is freely available, open source
programs are commonly distributed under very precise licensing agreements
generally named Copyleft licensing scheme. Copyleft licensing was
created for ideological purpose by Richard Stallman and the Free Software
Foundation (GNU project, 2000b). The main feature of copyleft is that once
a source code is licensed by the initial developer, the subsequent code based
on the original must also be licensed in the same way. Copyleft is devised
for linking the programmer and his/her contribution permanently together.
According to Mustonen (2003), copyleft creates an environment where
talented programmers have an incentive to signal their abilities via the open
source community. Currently, there are two organizations Open Source
Initiative (OSI) and Free Software Foundation (FSF), in charge of maintaining
a series of copyleft licenses. Among these licenses, Berkeley Software
Distribution (BSD)-style license, FSFs GNU General Public License (GPL)
and the Lesser GPL (LGPL) are the most important and popular. On the
21


other hand, the emergence of the commercially oriented licenses such as
Mozilla Public License (MPL) and IBM Public License (IPL) has also been
quite influential. Table 2.2 shows the various characteristics of these open
source licenses. The followings are their detailed descriptions.
Table 2.2 Characteristics of open source licenses
Licenses Does it impact derived works? Can it be closed?
BSD (Berkeley Software Distribution) License No Yes
GPL (General Public License) Yes No
LGPL (GNU Lesser GPL) No No
MPL (Mozilla Public License) No Yes
IPL (IBM Public License) No Yes
Source: Gacek and Arief 2004
The BSD License
The Berkeley Software Distribution (BSD) Licenses represent a family of
permissive free software licenses. The original was used for the Berkeley
Software Distribution, a Unix-like operating system for which the license is
named. The original owners of BSD were the Regents of the University of
California because BSD was first written at the University of California,
Berkeley.
The first version of the license was revised, and the resulting licenses are
more properly called modified BSD licenses. Permissive licenses,
sometimes with important differences pertaining to license compatibility, are
referred to as BSD-style licenses. Several BSD-like licenses, including the
22


New BSD license, have been vetted by the OSI as meeting their definition of
open source.
The BSD Licenses impose very few limitations on revising or
redistributing source code compared to other free software licenses such as the
GNU GPL or even the default restrictions provided by copyright, putting it
relatively closer to the public domain. The licenses permit a software vendor
to base its products on open source software, add its own custom
enhancements, and then sell the final products as commercial software where
the source code needs not to be open source. The MIT license and Apache
Software License approved by the OSI are two well-known examples of such
licenses.
The BSD licenses have been referred to as copy-center, as a comparison
to standard copyright and copyleft free software. Many commercial software
companies find this type of license very attractive since they can incorporate
the software that falls under this license into their products and then sell the
packages without turning them into open source.
General Public License
The GNU General Public License (GPL) is one of the most common and
stringent copyleft licenses. Its main idea is to keep software free in
perpetuity. All source code that incorporates GPL source code legally
becomes open source code itself. According to Stallman (1996), anyone who
lawfully obtains a program covered by the GPL automatically inherits the full
23


rights to use, copy, modify, or distribute the source code in any manner desired
but subject only to the terms of GPL itself.
Based upon the GPL, any modification to the software must be
subsequently distributed under the terms of the license itself. In other words,
a person who enhances software in terms of the GPL must contribute the
enhancements back to the open source project/community. Thus, many
people regard the GPL as being anti-commercial due to its rigid restriction.
However, the GPL does not prohibit charging a positive price for the source
code covered by the license as long as the code is not turned into closed
source.
Lesser General Public License
Based upon the nature of the GNU GPL, all source code consisting of it
must be released under the GPL. A modified license, the Lesser GPL (LGPL,
formerly the GNU Library General Public License) emerged as GPL had been
proved impractical. The LGPL, published by the FSF, was developed as a
compromise between the GPL and the BSD. The LGPL places copyleft
restrictions on the code itself but does not apply these limitations to other
software that merely links with the code.5
Although the LGPL imposes fewer restrictions than the GPL, there are
some other restraints. Essentially, it must be possible for the software to be
linked with a newer version of the LGPL-covered code. The most common
5 http://www.gnu.org/licenses/lgpl.html
24


method for doing so is to use a suitable shared library mechanism for linking.
Alternatively, a statically linked library is allowed if either source code or
linkable object files are provided. Overall, the major difference between the
LGPL and the GPL lies in the fact that LGPL is intended for use with software
libraries and it may be linked with proprietary code (Fitzgerald 2006).
Mozilla Public License
The Mozilla Public License (MPL) was developed by Mitchell Baker
when she worked as a lawyer at Netscape Communications Corporation and at
the Mozilla Foundation.6 It is often regarded as being the middle-ground
between the strictness of the GNU GPL and the tolerance of the BSD License.
It is not used anywhere near as widely as either the GPL or the BSD license,
but its flexibility and thoughtful drafting mean that it is becoming more
popular.
The MPL is commercially oriented. The license is regarded as a weak
copyleft, although source code copied or changed under the MPL must stay
under the MPL. Unlike strong copyleft licenses, the code under the MPL
may be combined in a program with proprietary files which would otherwise
be derivative works of the MPL code.7 For example, Netscape 6 and later
releases were proprietary versions of the Mozilla Application Suite.
Additionally, the MPL is the license for the Mozillas families, such as
Application Suite, Mozilla Firefox, Mozilla Thunderbird and other Mozilla
6 http://www.opensource.org/licenses/mozillal. 1 .php
7 http://www.mozilla.org/MPL/MPL-l. 1 .html
25


software. The MPL has been adapted by others as a license for their software,
most notably Sun Microsystems, as the Common Development and
Distribution License for OpenSolaris, the open source version of the Solaris 10
operating system.
IBM Public License
The IBM Public License (IPL) is a free and open source software license
used by IBM. It is approved by the OSI and the FSF. Unlike GNU GPL,
IPL places the liability on the publisher or distributor of the licensed program.
This is to facilitate commercial use of programs, without placing the
contributor in risk of liability. Its proponents say it has a clearer definition of
whos responsible for the program than the GPL has.
The IPL is incompatible with the GPL because it contains restrictions
which are not in the GPL. According to the FSF, it requires certain patent
licenses be given that the GPL does not require.8 It differs from the GPL in
the handling of patents, as IPL terminates the license upon patent disputes.
This license has also been criticized because of provisions in section 4 which
require commercial distributors of code covered by this license to indemnify
all upstream originators for legal costs relating to lawsuits brought about by
users of the software. It has been argued that this exposes small distributors
(e.g., Linux distributions that happen to sell CDs) to unbounded legal costs,
possibly arising from vexatious claims.
8 http://www.gnu.org/philosophy/license-list.html
26


Open Source Communities
Since OSS grants contributors and users the right to freely read and
modify the source code, their activities associated with OSS development
become communities of practice. Generally, the community members
interact with each other for knowledge sharing and collaboration in pursuit of
solutions to certain software problems. Thus, OSS development is
impossible to be successful if there is no accompanied community that
provides the stage for contributors and users to collaborate with each other.
The open source community of practice is described as follows.
Open Source Participants
Based on project type, an open source participant can take on different
roles in different projects. According to Nakakoji et al. (2002), members of
an open source community assume certain roles by themselves according to
their personal interest in the project rather than being assigned by someone
else. Table 2.3 shows the roles on which OSS participants might take in their
proj ect/community.
27


Table 2.3 Major participants in open source communities
Role Duty
Project Leader The person who initiates the project and is responsible for the vision and overall direction of the project.
Core Member The persons who are responsible for guiding and coordinating the development of an open source project. Usually, core members are those who have been involved with the project for a relative long time and have made significant contributions to the development and evolution of the system. In particular, if a single project manager quits, the core members would form a council to take the responsibility of directing the project development.
Active Developer The persons who regularly contribute new features and fix bugs. Typically, they are one of the major development forces of open source projects.
Peripheral Developer The persons who occasionally contribute new functionality or features to the existing system. Their contribution is irregular, and the period of involvement is short and sporadic.
Bug Fixer The persons who fix bugs either discovered by these participants or reported by other members in the community.
Bug Reporter The persons who discover and report bugs. They do not fix the bugs themselves, and may not read source code either. They assume the same role as testers of the traditional software development model.
Reader The persons who are active users of the system. They not only use the system, but also try to understand how the system works via reading the source code.
Passive User The persons who only use the system in the same way as most users of commercially closed source software.
Source: Nakakoji et al. 2002; Ye and Kishida 2003
Structure of Open Source Communities
The structure of most OSS development teams is demonstrated as having
a hierarchical or onion-like form (Moon and Sprouli 2000; Nakakoji et al.
28


2002). According to Nakakoji et al (2002), at the core is a project manager
surrounded by a few core members and then more active developers and
peripheral developers respectively. Surrounding the developers are bug
fixers followed by bug reporters. Readers are located between bug reporters
and passive users. At the outermost layer of the onion are passive users.
In essence, the core of an open source community is kept as small as
possible since it would be difficult to control if the core group were large. In
addition, the roles closer to the core have a greater scale of influence. A
project managers action would have more impact than that of a core member
which in turn has a more significant influence than an active developer and the
like. Users have the least influence, but they still can make certain
contributions to the community since they use the latest product releases and
usually contribute bug reports or feature requests (but not code). Actually,
non-developers may also contribute by writing documentation or translating
the system. Generally, every open source community has its own features
based upon the project type and the number of its members. The structure
differs in the percentage that each role in the community changes (Ye and
Kishida 2003).
Mode of Software Development
OSS development has several particular features driven by its loosely
organized community and numerous highly distributed developers. The most
significant characteristic is modularization. The success of an open source
29


project heavily depends upon the ability to break the project into distinct
components (Lemer and Tirole 2002). Modular designs benefits are a direct
result that it supports increased understanding during design and concurrent
allocation of work during implementation (Gacek and Arief 2004). A
well-defined interface and modularized source code are a prerequisite for
effective remote collaboration, since OSS development is globally distributed
(Bollinger et al. 1999).
The second feature of OSS development lies in its architecture. A
computing systems software architecture represents its structure and
comprises its components, the components externally visible properties and
their relationships (Bass et al. 1998). OSS systems architecture might be
available or not. An unintentionally unavailable software architecture
suggests that the structure exists in some peoples minds only (Gacek and
Arief 2004).
Documentation and testing are two important tasks for software
development. Good documentation enables users to accurately use the
software and understand how to modify the software. Testing can give users
and/or developers confidence that the software would work as expected.
However, these two aspects are often ignored or changed broadly during OSS
development. OSS developers tend to be more interested in coding than
documenting or testing (Raymond 2001; Gacek and Arief 2004). This is
probably because OSS tries to replace the formal testing process with many
30


eyeballs approach to fixing bugs and also developers usually consider that the
addition of comments to the source code is sufficient documentation.
Numerous open source projects implement some types of configuration
management in order to facilitate concurrent software development and
controlled evolution. They do this by utilizing Concurrent Versions System
(CVS) or other ad hoc web-based supports. Therefore, this has also become
one of OSS important developments features.
Process of Submissions Acceptance
In the settings of OSS development, source code or patches are usually
submitted from a variety of sources. According to Gacek and Arief (2004),
open source projects often post the areas for which they want to receive
submissions, and often they might receive multiple concurrent submissions
addressing the exact same problem. Therefore, projects usually have
specified processes for accepting submissions and handling multiple
concurrent submissions.
Generally, an open source project accepts participants contributes
through three steps: (1) choosing the work area; (2) making decision for
approving the submissions; (3) disseminating the approved submission (Gacek
and Arief 2004). First, code or patches are written by the developer and are
submitted to the specific work area which addresses the projects
corresponding problems. Second, the decision to accept the submissions is
based on four aspects: (1) quality goals; (2) acceptance criteria; (3) decision
31


groups cognitive abilities; and (4) the projects social structure. Third, a
project might passively disseminate the accepted contributions through
newsgroups or comments in the code itself, or actively distribute the
information via email or mailing lists.
With respect to the decision making process, quality goals and acceptance
criteria extensively differ from project to project and from one community to
another. Likewise, the decision making for approving the submissions varies
among projects and potentially within projects since it highly relies on the
decision groups cognitive abilities to identify appropriate solutions. In
addition, the social structure of an open source project/community also plays
an important role in decision making. In a hierarchical structure, different
groups of members evaluate different submissions, or some members might
have greater power to make the final decision. If the structure is monolithic,
the decision group might use consensus or majority vote to decide whether the
submissions can be accepted.
Communities of Practice
OSS communities possess the essential qualities that Wenger and Snyder
(2000) identified as communities of practice, which are formed by people,
mostly practitioners, who are informally bound together by shared expertise
and passion for a joint enterprise. The term communities of practice, first
coined by Lave and Wenger (1991), refers to a theory that builds on learning
as social participation (Wenger 1998). According to Wenger (1998),
32


communities of practice are formed by people who engage in a process of
collective learning in a shared domain of human endeavor: a tribe learning to
survive, a band of artists seeking new forms of expression, a group of
engineers working on similar problems, a clique of pupils defining their
identity in the school, a network of surgeons exploring novel techniques, or a
gathering of first-time managers helping each other cope.
Wenger and Synder (2000) suggests that communities of practice can
drive strategy, generate new lines of business, solve problems, promote the
spread of best practices, develop peoples professional skills, and help
companies recruit and retain talent, since people in communities of practice
share their experiences and knowledge in free-flowing, creative ways that
foster new approaches to problems. Moreover, communities of practice can
function in virtual and distributed environments, since communities generally
are concerned with motivation, are self-generating, are self-selecting, are not
necessarily co-located, and have a common set of interests motivated to a
pattern of work not directed to them (Hildreth et al. 2000). Therefore, a
number of OSS researchers have introduced the idea of communities of
practice to study OSS communities.
Communities of practice emerge in companies that thrive on knowledge
(Wenger and Snyder 2000). The concept of communities of practice has
been regarded as setting the stage for effective knowledge sharing. Thus, it
is appealing to use the concept to describe and to analyze knowledge
communities whose purpose is knowledge creation and communication. As
33


knowledge communities are located in the virtual such as open source
communities, the investigation of communities of practice may help further
understand OSS communities.
Private-Collective Innovation
The practices of OSS developers and communities present a novel and
successful alternative to conventional innovation models. According to von
Hippel and von Krogh (2003), OSS development is an exemplar of a
compound private-collective model of innovation that contains elements of
both the private investment and the collective action models. In the
private-collective model, participants in OSS projects use their own resources
to privately invest in creating novel software code. These innovators could
then claim proprietary rights over their code, but instead they choose to freely
reveal it as a public good. Clearly, the net result of this behavior appears to
offer society the best of both worlds new knowledge is created by private
funding and then offered freely to all (von Hippel and von Krogh 2003).
In the private investment model, innovation is supported by private
investment and that private returns can be appropriated from such investments
(Demsetz 1967). Innovating contributors should only freely reveal their
innovations when the costs of free revealing are less than the benefits. It has
been argued that such conditions can hold in many fields, including OSS
(Harhoff et al. 2003). In contrast, the collective action model applies to the
provision of public goods ranging from provision of a public bridge to
34


provision of OSS (von Hippel and von Krogh 2003). It requires that
contributors of innovation relinquish control of knowledge they have
developed for a project and make it a public good by unconditionally
supplying it to a common pool. This requirement enables collective action
projects to avoid the social loss problem associated with the restricted access
to knowledge of the private investment model.
OSS projects display with respect to the assumptions about incentives
embedded in the private investment and the collective action models of
innovation is that contributions to OSS development are not pure public goods
since they have significant private elements even after the contribution has
been freely revealed (von Hippel and von Krogh 2003). More specifically,
the private-collective model of innovation occupies the middle ground
between private investment and collective action models. In the setting of
OSS development, contributors must engage in problem solving to create
novel code. When they freely reveal this code to the project, it becomes a
public good. However, the problem-solving process and effort used to
produce the code have other important outputs as well, such as learning and
enjoyment, and a sense of ownership and control over their work product. In
other words, programmers contribute freely to the provision of a public good
because they gamer private benefits from doing so.
Therefore, the emerging phenomenon of OSS development obviously
does not undermine either the private investment or the collective action
models of innovation. However, it does make clear the utility of combining
35


both into a private-collective incentive model that can more effectively
address the interlinked private and collective incentive structures observable in
that field, and perhaps elsewhere as well (von Hippel and von Krogh 2003).
Prior Research on OSS Developers
Open source development involves lots of developers spread all over the
world. Generally, developers from different geographic areas work
asynchronously on a large variety of open source projects. Moreover, the
majority of developers are volunteers who donate their time, energy, and
knowledge to the projects without any financial compensation. According to
Mockus et al. (2000), there are potentially very large numbers of people
engaged in open source software development, and often hundreds or
thousands of volunteers contributing to the exact same project. Therefore,
this raises an intriguing and widely debated issue: what motivates open source
developers to contribute their effort? This topic has been explored from a
variety of theoretical viewpoints such as social psychology, organizational
behavior, economics, and so on. Related studies and empirical findings on
OSS developers motivation and effort are described as follows.
Studies on Motivation
The OSS literature suggests that OSS developers show both intrinsic and
extrinsic motivations (Bitzer et al. 2004). OSS developers motivations are
considered to be related in complex ways rather than independent, and
36


different motivations have an impact on developers participation in different
ways (Roberts et al. 2006). The list of all possible motivators is too
extensive to include all motivators in one study. Therefore, this research
focuses on those motivators that appear to have the strongest theoretical
arguments. The various theoretical viewpoints reviewed here are
summarized in Table 2.4.
Table 2.4 Major motivations of OSS developers
Motivation Descriptions Type
Helping Aiding others to increase the welfare of others Intrinsic
Enjoyment Enjoying writing programs Intrinsic
Peer Recognition Receiving recognition from others within the community Intrinsic
Enhancing Human Capital Accumulating skills and knowledge via adult learning Extrinsic
Career Advancement Demonstrating capacities and skillfulness to signal potential employers Extrinsic
Satisfying Personal Needs Acquiring software patches or components for personal use or solving job related technical problems Extrinsic
Intrinsic Motivation
Intrinsic motivation is defined by Deci (1975) as the doing of an activity
for its inherent satisfactions rather than for some separable consequence.
Based upon Ryan and Deci (2000), an individual when intrinsically motivated
is moved to act for the fun or challenge entailed rather than because of
external prods, pressures, or rewards. Decis and Ryans theory has been
37


regularly applied to explain why open source developers voluntarily contribute
their efforts. Empirical analyses show that intrinsic motivations are essential
in determining the participation of individual programmers in open source
projects (David et al. 2003). Generally, from the perspective of intrinsic
incentives, open source developers rank helping others, enjoyment to program,
and peer recognition among the most important reasons to take part in open
source movement (Bates et al. 2002; Ghosh et al. 2002; Hars and Ou 2002;
Lakhani and von Hippel 2003).
Helping behavior. Open source community is a gift culture that is
motivated by altruism and reciprocity (Raymond 2001; Hars and Ou 2002;
Bonaccorsi and Rossi 2003). Altruism, understood as doing something for
someone else at some cost to oneself, is contrasted with selfishness ...
altruism is a natural part of human nature that it is not just found in a few rare
people that it has evolutionary value and is exhibited in some manner by
everyone (Ozinga 1999). People often behave altruistically and pro-socially,
contributing to the welfare of others without apparent compensation (Schwartz
1970). Altruism exists when people derive intrinsic enjoyment from helping
others without expecting anything in return (Krebs 1975; Smith 1981). OSS
developers voluntary behavior has been regarded as an altruism-driven
incentive since they are willing to proactively contribute their work with no
reimbursement.
Giving away goods and services for free allows individuals to make and
maintain social links and entails the duty to reciprocate (Mauss 1959).
38


Reciprocity has been highlighted as a benefit for individuals to engage in
social exchange (Blau 1964). Wasko and Faraj (2000) indicate that people
who share knowledge in online communities believe in reciprocity since they
feel that sharing knowledge to help others not only feels good but that
everyone is better off when knowledge is shared. Moreover, people who
regularly help others in virtual communities seem to receive help more quickly
when they ask for it (Rheingold 2000). This suggests that such a personal
benefit is more likely to accrue to individuals who actively participate and
help others in open source communities (von Hippel and von Krogh 2003).
Rheingold (1994) points out that information-sharing based on a hunger
for intellectual companionship is initially found most commonly among
professionals who work more or less on their own, e.g., journalists, freelance
artists and designers, programmers, etc. Open source community is
organized by numerous devoted programmers who are both producers and
users. The creation and maintenance of social relations in such a community
are not regulated by the possession or exchange of money or commodities but
the economy of gift exchange (Bergquist and Ljungberg 2001). Literature
suggests that OSS developers voluntarily make contributions because they
would like to lend a hand to other members and simultaneously give
something back to those who have ever assisted them. In a gift cultures
setting, given the abundance of resources, social status is determined not by
what you have but what you give away (Raymond 2001). This phenomenon
is most likely to happen when the exchange is not in favor of well-known
39


individuals but of a community of unknown subjects (Bonaccorsi and Rossi
2003), such is the case of the OSS community.
Enjoyment. According to Csikszentmihalyi (1975), enjoyment-based
motivation refers to a satisfying flow of activity from which the enjoyment can
be derived, such as playing sports or collecting stamps for pleasure.
Csikszentmihalyi provided the concept of a state of flow to explicate the
phenomenon. Based upon the theory, enjoyment is maximized in a state of
flow. Flow states arise when a persons skill matches the challenge of a task.
Moreover, there is an optimal zone of activity in which flow is maximized.
Enjoyable activities are found to provide feelings of creative discovery, a
challenge overcome and a difficulty resolved (Csikszentmihalyi, 1975).
Following the theory of flow states, Lakhani and Wolf (2003) suggest that
OSS developers may be seeking flow states by selecting projects that match
their skill levels with task difficulty, a choice that may not be available in their
regular jobs, when they participate in an open source project. Additionally,
the literature in electronic networks of practice has shown that individuals are
motivated intrinsically to contribute knowledge to others because engaging in
intellectual pursuits and solving problems is challenging or fun (Wasko and
Faraj 2000). Lemer and Tirole (2002) point out the programmer compares
how enjoyable the mission set by his/her employer and the open source
alternative are. A cool open source project may be more fun than a routine
task. Thus, enjoyment has been regarded as one of the most important
incentives that motivate OSS developers to participate in open source projects.
40


Peer Recognition. Peer recognition has been viewed as one type of
signaling incentive. Lemer and Tirole (2002) suggest that programmers
participating in open source projects based on this motivation may also want
to signal their abilities to different subjects although they may shun future
monetary rewards. According to Hars and Ou (2002), peer recognition stems
from the desire for fame and esteem which is associated with future returns.
Following Maslow (1970), All people in our society (with a few pathological
exceptions) have a need or desire for ... reputation or prestige, status, fame
and glory, dominance, recognition, attention, importance, dignity, or
appreciation. Hence, it is reasonable that programmers get involved in OSS
development simply because they expect the acquisition of recognition from
their counterparts.
Essentially, OSS development is quite similar to the publication of
scientific research. In scientific societies, sharing results enables researchers
both to improve their results through feedback from other members of the
scientific community and to gain recognition and hence prestige for their work
(Banaccorsi and Rossi, 2003). In the settings of OSS development, while
releasing the source code to the communities, OSS developers can not only
obtain other members fast feedback helping them refine their work but also
gain recognition from those who are using their contributions and probably
bring to themselves future employers notice. As a result, successful
contributors of open source projects will have potential opportunities to
advance their career because of the effect of recognition.
41


Extrinsic Motivation
Ryan and Deci (2000) define extrinsic motivation as a construct that
pertains whenever an activity is done in order to attain some separable
outcome. They describe that extrinsic motivation refers to behavior where
the reason for doing it is something other than interest in the activity itself
(Deci and Ryan, 1985). Intriguingly, economics appears to have become the
most dominant theory used to explain open source developers extrinsic
motivation. According to economic literature, a programmer is expected to
participate in an open source project only if a benefit can be derived from
engaging in the activity (Lemer and Tirole 2002). The major benefits from
participation mainly consist of enhancing human capital, advancing career
opportunities, and satisfying personal needs for software.
Enhancing human capital. A number of economists and OSS
researchers regard enhancing human capital as one of the important incentives
that motivates open source participants to engage in open source projects.
From the standpoint of labor economics, human capital as a determinant of
productivity refers to personal skills, capabilities, and knowledge (Mankiw
2004). Human capital is the economists term for the accumulation of
investments in people such as education and on-the-job training. The most
important type of human capital is education ... education represents an
expenditure of resources at one point in time to raise productivity in the
future ... Human capital includes the skills accumulated in ... on-the-job
training for adults in the labor force (Mankiw 2004, p. 412, p. 542).
42


OSS development provides an excellent learning environment for
participants who seek opportunities to enhance their human capital.
Developers who start an open source project are generally master
programmers and their works are the products of fine craftsmanship, providing
examples of excellent programming for less skilled developers (Ye and
Kishida 2003). In addition, the freedom to choose tasks not only enable OSS
developers to select the learning experiences that meet their needs and
interests, but also enable those entry-level programmers like college students
to participate in realistic projects at a very early stage (Hars and Ou 2002).
Thus, those open source participants can develop skills, such as programming,
that are going to be useful when they enter the labor market (Bonaccorsi and
Rossi 2003).
Career advancement. Participation in OSS development has the
potential to advance ones career in two ways. The first is the opportunity for
OSS participants to demonstrate (advertise) their capabilities and skillfulness.
The larger the contribution of an individual to open source projects, the more
likely it is that the commercial software vendors will recognize the value of
the individual (Hars and Ou, 2002).
OSS development is an opportunity for participants to signal potential
employers. According to economic signaling theory, signaling refers to
actions taken by an informed party to reveal private information to an
uninformed party (Mankiw 2004). Signaling within an open source context
would be particularly valuable if employers have difficulty assessing potential
43


employees. Instead, employers could use success at developing OSS as an
indicator of ability.
Second, participants may also use OSS involvement to acquire access to
venture capital or to launch a new business endeavor. The most well known
examples are the founders of Linux, Sun, and Red Hat (Lemer and Tirole
2002). However, the majority of OSS developers are more likely to use open
source communities for job advancement or acquiring shares in commercial
open source-based companies than accessing venture capital.
Satisfying personal needs. Numerous open source projects were
initiated by the need that the founders had for specialized software or software
patches since they had met certain technical problems in their day-to-day work
(Hars and Ou 2002; Lemer and Tirole 2002). For example, one of the origins
of the free software movement was Stallmans inability to improve a printer
program because its owner, Xerox, refused to release the source code (Lemer
and Tirole 2002).9 During the development of the Apache web server a
similar problem was encountered when the original developers struggled to
find patches for revising the server software.
Definitely, many OSS projects take shape because the people promoting
them have looked in vain for a program to perform a particular function.
They arise to satisfy a work-related demand for which there is no
corresponding supply, in short to fill an unfilled market (Bonaccorsi and
Rossi 2003). Thus, participants in OSS efforts may directly benefit from the
9 Richard Stallman founded the Free Software Foundation in 1984. Then, he was a
programmer from MITs Artificial Intelligence Lab.
44


software and software improvements they develop because they have a
personal or job related use for them (von Hippel 1988; Raymond 2001).
In addition, a major advantage of OSS is the freedom to incorporate the
software into other products or to produce add-on products that enhance the
features of OSS. In many cases, the ability to modify OSS to satisfy exact
personal or business requirements or knowing how to incorporate OSS into
other products is valuable. Developing OSS is one way of acquiring the
knowledge needed to produce new products that utilize OSS.
Effort and Performance
For the sake of the sustainability of an OSS project, an understanding of
OSS developers effort and performance is an important issue on OSS
development. According to Christen et al. (2006), effort is an input to work
while performance is an output from this effort. In the setting of OSS
development, there has been several research concerning developers effort
and performance.
Effort
Effort defined as the number of hours per week spent on an open source
project has been used in a number of previous OSS studies (Hars and Ou 2002;
Hertel et al. 2003; Lakhani and Wolf 2003) and provides an appropriate proxy
for developers contribution and interest in OSS projects (Lakhani and Wolf
2003). With respect to the success of an OSS project, it is imperative to
45


understand whether all types of motivations affect OSS developers effort
equally or in the same way. Some motivations may strongly affect the effort
whereas others may not be as salient.
Academic theorizing on individual motivations for participating in OSS
projects has posited that external motivational factors in the form of extrinsic
benefits (e.g., career advancement) are the main drivers of effort (Lakhani and
Wolf 2003). In contrast, Hars and Ou (2002) argued that open source
programmers with intrinsic motivations will spend more time and effort in
open source projects since intrinsically motivations have been suggested to be
associated with most effortful behaviors (Sheldon and Elliot 1997).
Therefore, there is no consensus in the OSS literature as to which motivation
has the most dominant impact on individual OSS contributors effort.
Hars and Ou (2002) performed a simple correlation analysis, finding a
weak correlation between effort and altruism and strong correlations between
effort and both building human capital and signaling employers. However,
Lakhani and Wolf (2003) argued that the most significant and pervasive
determinant of effort dedicated to OSS projects was enjoyment-related
intrinsic motivation in the form of a sense of creativity, followed by extrinsic
motivation in form of payment. Moreover, they revealed that paid
contributors would spend more hours per week than volunteer contributors on
OSS projects, which implied financial subsidy is substantial to OSS
development. Contrary to experimental findings on the negative impact of
extrinsic rewards on intrinsic motivations (Deci et al. 1999), Lankhanis and
46


Wolfs results suggested that both being paid and feeling creative on OSS
projects do not have a significant negative impact on developers effort.
Performance
One of the tenets of OSS projects is the frequent provision of feedback to
contributors (Moon and Sproull 2000). In a number of well known open
source communities like Apache Software Foundation, continued contribution
is rewarded with a change in performance ranking (Roberts et al. 2006).
They periodically evaluate the actual contributions of their members and
assign each member a certain performance ranking. These rankings are
based on merit and reflect the contributors level of participation in the OSS
community. Advancement within the meritocracy recognizes individuals
commitment and contributions to the OSS projects (Fielding 1999).
In essence, performance is broadly defined as an aggregate construct of
effort, skill, and outcomes (Walker et al. 1977; Behrman and Perreault 1984;
Lusch and Serpkenci 1990). It refers to an evaluation of the results of an
individuals behavior usually by someone other than the individual, and
involves determining how well or poorly an individual has accomplished a
task (Kanfer 1990). Experimental research in psychology also suggests that
performance is the outcome of an evaluation by others of an individuals
behavior and this behavior is often manifested by individuals task output
(Mitchell and Daniels 2003).
47


Theoretically, motivation has an important influence on performance, and
it has been deemed as an antecedent of performance. According to Roberts
et al. (2006), motivations vary across individuals and combine with
individuals knowledge, skills, and abilities to produce task relevant behaviors.
These behaviors contribute to individual performance. Studies have shown
that motivation focuses attention on particular task elements and produces
effort as people work harder when they are motivated (Roberts et al. 2006).
In the OSS context, Roberts et al. (2006) argued that motivations
influence OSS developers participation in OSS projects as exemplified by the
level of their contributions to the source code. Over time, contributors
participation is evaluated by the OSS community. This performance
evaluation may lead to a rise in a contributors rank within the community,
which can, in turn, act as feedback to influence the future motivations of
contributors.
Example Models
A number of researchers have explored motivational processes and
performance of persons who voluntarily engage in OSS development. A
rather intuitive approach to understand the motives of OSS developers is built
on expectancy-valence theory that explains motivational processes in
individual developers mindset. Another method relevant for the
understanding of developers motivation and performance stems from
48


systematic approaches provided by a theoretical model from intrinsic and
extrinsic motivation. The two relevant studies are described as follows.
Wu, Gerlach, and Young model (2007)
Figure 2.1 Open Source Participants Continuation Model
Wu, Gerlach, and Young (2007) conducted a field survey of 148 OSS
participants to identify salient determinants of open source participants
intention to continue making OSS contributions. Toward this goal, the
concepts from expectancy-valence theory were adapted to build a theoretical
model of OSS participants continuance intention shown in Figure 2.1.
49


According to the findings, satisfaction with participating in open source
projects was the strongest influence on OSS participants intentions to
participate in future OSS projects, followed by their motivation on enhancing
human capital and job related problem-solving. These associations
demonstrated that satisfactory experience and recognition of benefits from
solving job related technical problems and increasing human capital influence
OSS participants continuance behavior. The results asserted that satisfaction
is the key to continuance, participants may subconsciously pursue instrumental
behavior because of the requirement for specific software functionalities, and
the theory that learning by doing is of essential importance to the success of
OSS projects and the sustainability of OSS communities.
Roberts, Harm, and Slaughter model (2006)
To understand what motivates OSS developers to participate in OSS
development, Roberts et al. (2006) conducted a study on the developers of the
Apache projects by revealing how different motivations were interrelated, how
these motivations influenced participation leading to performance, and how
past performance influenced subsequent motivations. Drawing on theories of
intrinsic and extrinsic motivation, Roberts et al. developed a theoretical model
relating the motivations, participation, and performance of OSS developers.
The model is shown in Figure 2.2. They evaluated the model using survey
and archival data collected from a longitudinal field study of software
developers in the Apache projects.
50


Figure 2.2 OSS Developers Motivation and Performance Model
The theoretical framework for Roberts et al.s study leveraged the general
model of motivation and performance in organizational and social psychology.
In their model, extrinsic motivations include being paid to contribute,
enhancing status/career opportunities, and use value (i.e., the desire to fix a
bug or solve a problem of immediate relevance to the contributor), while
intrinsic motivations refer to needs for competence, control, and autonomy.
Performance is related to promotion to a higher rank within the Apache
hierarchy, which is awarded after one or more cycles of contribution followed
by a positive peer review and is consequently an acknowledgement of an
individuals substantive contributions to the project.
51


This study argued that motivations varied across individuals and
combined with individuals knowledge, skills and abilities to produce
task-relevant behaviors and influence the level of OSS developers
contribution to the source code. In addition, motivation has an important
influence on performance because it focuses attention on particular task
elements and produces effort as people work harder when they are motivated.
This performance evaluation may lead to an increase in a contributors rank
within the community.
Roberts et al.s results revealed several important findings. First, OSS
developers motivations were not independent but rather were related in
complex ways. Being paid to contribute to Apache projects was positively
related to developers status motivations but negatively related to their use
value motivations. Moreover, Roberts et al. found no evidence of diminished
intrinsic motivation in the presence of extrinsic motivations; rather, status
motivations enhance intrinsic motivations. Second, different motivations
differentially influenced participation. Developers paid participation and
status motivations led to above average contribution levels, but use value
motivations led to below average contribution levels, and intrinsic motivations
did not significantly impact average contribution levels. Third, developers
contribution levels positively influenced their performance rankings. Finally,
past performance rankings enhanced developers subsequent status
motivations.
52


Summary
OSS development affords a particularly rich context in which to examine
individual motivations. Exploratory empirical research has been done
pertaining to the motivations of OSS developers. For years, OSS researchers
have identified a variety of motivators and suggest that developers motivation
may contain various sources and, simultaneously, different motivations could
differentially relate to participation in open source projects.
OSS developers motivations have been classified as intrinsic or extrinsic.
Intrinsic motivators include helping, enjoyment, and peer recognition, while
extrinsic motivators include enhancing human capital, career advancement,
and personal requirements for the software. The empirical findings from
extant research on developers help us understand the relationships between
motivations and developers behavior as well as related issues associated with
OSS development, such as effort and performance. For example,
participation in open source projects may be based on multiple incentives;
developers may vary their effort according to different motivators; motivations
influence developers performance; and developers satisfaction level
influences their continuance intention.
Although extant research has examined OSS developers motivations,
behavior, and the relationship of motivation to effort and performance, no
research heretofore has investigated OSS contributors motivations as a whole.
In other words, these studies do not take into consideration the influence of
each individual motivator. Moreover, many arguments have not been
53


empirically and rigorously measured with ground theories such as
expectancy-valence theory, human resource theory, and volunteerism. Based
on different motivators, what are developers expectations and attitudes in
association with OSS development after participating in open source projects?
Further, what impact will an individual motivator have on developers effort?
Clearly, additional research is needed to address these issues.
54


CHAPTER 3
REVIEW LITERATURE OF VOLUNTEERISM
Research on OSS developers suggests that open source communities
cannot exist or prosper without the contributions of highly motivated
developers who voluntarily donate their time and effort to the community.
Since the majority of OSS developers are strictly volunteers rather than
traditional paid employees, it is not possible to solely rely on employment
relationships or employment contracts to manage these persons (Roberts et al.
2006). Therefore, an understanding of volunteers motivations and their
behavior is considered an important first step in exploring how to motivate
OSS developers and how to direct, sustain, and influence their volunteer
behavior.
Volunteering is routinely identified as work that is unpaid and taken on
freely (Mutchler et al. 2003). For many individuals, volunteering may
provide rewards that paid employment does not, e.g., self-fulfillment through
challenging and interesting activities (Miller 1985). However, seeking more
time and more work from a volunteer is usually not a welcome request
(Galindo-Kuhn and Guzley 2001). Because conventional rewards are
non-existent in volunteer work, one must turn to intrinsic rewards instead
(Gidron 1984). Therefore, paid work is infused into volunteer activities
when there is an economic and social necessity for it (Gidron 1983).
55


In the context of OSS development, a number of open source
communities allow companies benefiting from OSS to involve their paid
employees in open source projects. Alternately, communities may
financially support their own volunteer developers to devote their full attention
to the tasks assigned by the projects such as Debian and GNOME (GNU
Network Object Model Environment). The combination of volunteer
workers with paid employees may have a certain impact on volunteers
mindset and behavior which, in turn, may influence the progression of their
activities. For example, the influence of paid employees whose job is
GNOME development has been instrumental in the success of the project,
while Debian has been suffering serious delay. In this chapter, volunteer
motivation, work attitudes, and other related issues such as turnover and effort
are elaborated.
Volunteerism
One important manifestation of human helpfulness is volunteerism (Clary
et al. 1998). Volunteerism is conceptually defined as an ongoing activity
aimed at improving the well-being of others (Omoto and Snyder 1995). It is
an activity that can be found all over the world (Curtis et al. 1992). Every
year, millions of people share significant amount of their time and talents as
volunteers to helping others, and the number is on the rise (Clary et al. 1998;
Skoglund 2006).
56


The study of volunteerism is one element in the large effort to study
helping behavior (Mowen and Sujan 2005). Specially, it has long captivated
researchers of social behavior that an individual would make significant
personal sacrifices for another person particularly when that person is a
stranger (e.g., Latane and Darley 1970; Staub 1978; Piliavin et al. 1981;
Eisenberg 1986; Batson 1991; Schroeder et al. 1995). Existing volunteerism
literature speaks largely to varieties of helping somewhat different from
volunteerism, focusing on helping in contexts where a potential helper is faced
with an unexpected need for help, calling for an immediate decision to act and
an opportunity to provide one and only one relatively brief act of help (Benson
et al. 1980; Bar-Tal 1984; Piliavin and Chamg 1990). Clary et al. (1998)
suggested that factors revealed by research on the helping that occurs in these
kinds of contexts, sometimes referred to as spontaneous helping, may have the
important impact on volunteerism.
Volunteerism seems to be a rather different kind of helping, a kind that is
prototypic of planned helping, which often calls for considerably more
planning, sorting out of priorities, and matching of personal capabilities and
interests with type of intervention (Benson et al. 1980; Clary et al. 1998).
Therefore, based upon Clary et al. (1998), volunteers should have three
characteristics (1) they often actively seek out opportunities to help others,
(2) they may deliberate for considerable amounts of time about whether to
volunteer, the extent of their involvement, and the degree to which particular
activities fit with their own personal needs, (3) they may make a commitment
57


to an ongoing helping relationship that may extend over a considerable period
of time and that may entail considerable personal costs of time, energy, and
opportunity.
Motivation to Volunteer
It is essential to understand volunteers initial motivation to volunteer and
their expectations of the volunteering experience, since recruitment and
selection of volunteers is a costly process and it can be critical to the future
use of volunteers in an agency (Cnaan and Goldberg-Glen 1991; Ralston and
Rhoden 2005). Although Gidron (1984) argued that the motives that initially
influence people to volunteer may differ from those that influence their
decision to continue to volunteer, it is extremely important to understand the
initial motivation of those who remain as volunteers for a long run (Cnaan and
Goldberg-Glen 1991). Moreover, based upon the definition and
characteristics of volunteerism, Clary et al. (1998) indicated that it may be
productive to adopt a motivational perspective and to inquire about the
motivations that may dispose individuals to seek out volunteer opportunities,
to commit themselves to voluntary helping, and to sustain their involvement in
volunteerism over extended periods of time.
Reasons for Volunteering
Motivation is a difficult concept in general, because it is subconsciously
constructed and it is neither systematic nor consistent (Cnaan and
58


Goldberg-Glen 1991). Gillespie and King (1985) suggested that we will
never know the answer if we do not ask people what motivates them to
volunteer. According to Tschirhart et al. (2001), volunteering can reflect
multiple motivations (including altruistic, instrumental, social, self-esteem,
and other goals) that can be collapsed into one or more factors or dimensions.
Cnaan and Goldberg-Glen (1991) summarized twenty eight significant
motives for volunteering in human services on the basis of motivation to
volunteer literature. Those motivations are listed in Table 3.1.
Table 3.1 Motives for volunteering
No. Reason
1 It is Gods expectation that people will help each other.
2 I adhere to the agencys specific goals.
3 If I did not volunteer there would be no one to carry out this volunteer work.
4 I did not have anything else to do with my time.
5 I was lonely.
6 I have more free time (i.e., kids have left home, retired, widowed, divorced).
7 I wanted to gain some practical experience toward paid employment (or new career).
8 I wanted to broaden my horizons.
9 Being involved with this agency is considered prestigious.
10 Volunteering for others makes me feel better about myself.
11 Volunteering in this agency provides challenging activities.
12 Most people in my community volunteer.
13 Helping people in need improves my attitude regarding my own life situation.
14 Volunteering creates a better society.
15 My employer-school expect their employees-students to provide volunteer community service.
16 Volunteering is an opportunity to change social injustices.
17 Volunteering is an opportunity to develop relationships with others.
18 Volunteering is an opportunity to work with different age groups.
59


Table 3.1 (Cont.)
No. Reason
19 Volunteering is an opportunity to do something worthwhile.
20 Volunteering is an opportunity to return good fortune.
21 A relative of friend is/was a client of this agency.
22 I have past experience providing similar service.
23 I am able to relate better to the patients/residents situation because of my own similar experience.
24 This volunteering gives me an opportunity to vary my weekly activities.
25 Previous contact with professionals in this agency.
26 Volunteering for this agency enables it to provide more care for less money.
27 Its a way to continue a family tradition of helping in need.
28 This is an excellent educational experience.
Source: Cnaan and Goldberg-Glen 1991
Functional Approach
Based upon Clary et al. (1998), the fundamental concerns of motivational
inquiry with understanding the processes that move people to voluntary action
are precisely the concerns engaged by the questions: Why do people
volunteer? and What sustains voluntary helping? A useful approach to
answering these questions begins with the premise that volunteering serves
different functions for different people, i.e., functional analysis (Houle et al.
2005). The functional approach is explicitly concerned with the reasons and
the purposes, the plans and the goals, that underlie and generate psychological
phenomena, that is, the personal and social functions being served by an
individuals thoughts, feelings, and actions (Snyder 1993).
60


Houle et al. (2005) suggested that the functional approach may help to
discover underlying motivations of volunteering since it implicates the
importance of matching volunteer motivations to the benefits that
volunteerism provides. According to a functional analysis of volunteerism,
people engaging in similar acts may have different underlying motivations for
doing so (Houle et al. 2005). Clary et al. (1998) adopted the strategy of
functional analysis to analyze volunteer motivations. They catalogued six
functions of volunteerism: values, understanding, social, career, protective,
and enhancement. The six functions are described as follows.
The values function refers to concerns for the welfare of others, and
contributions to society (Clary et al. 1998). The function has been likened to
altruism (Clary and Miller 1986), the value-expressive attitude function (Katz
1960), and the quality of expressiveness (Smith et al. 1956). An empirical
evidence for the values function indicated that over 70% of the respondents
endorsed to help others as a reason for volunteering (Anderson and Moore
1978).
The understanding function served by volunteering involves the
opportunity for volunteerism to permit new learning experiences and the
chance to exercise knowledge, skills, and abilities that might otherwise go
unpracticed (Clary et al. 1998). This function is related to Katzs (1960)
knowledge function and Smith et al.s (1956) object appraisal function. In
support of this function, Gidron (1978) found that young volunteers (high
61


school and college students) tended to view their volunteer work as a learning
and self-development experience.
A third function served by volunteering is the social function in which an
individual volunteers due to strong normative or social pressure, or to get
along with others in his or her reference group (Clary et al. 1998; Haule et al.
2005). Conceptually, this function is similar to Smith et al.s (1956) social
adjustive function and Francies (1983) need to respond to the expectations of
others. Piliavin et al. (1984) found evidence for the social function in their
investigation of motives for donating blood. They found that some
individuals donate blood due to external, social motives.
A fourth function that may be served by volunteering is concerned with
increasing ones job prospects and enhancing ones career that may be
obtained from participation in volunteer work (Clary et al. 1998). For
example, Beale (1984) suggested encouraging students to volunteer as the
experiences may serve as steppingstones to employment. Jenner (1982)
indicated that some Junior League volunteers perceived volunteering to be a
means of preparing for a new career or of maintaining career-relevant skills.
A fifth function that may be served by volunteering is the protective
function in which an individual volunteer tries to reduce feelings of guilt about
being more fortunate than others, or to escape from ones own problems
(Clary et al. 1998). Thus function could be likened to Katzs (1960)
ego-defensive function, Smith et al.s (1956) extemalization function, and
Francies (1983) need to express feelings of social responsibility. Schwartz
62


(1970) found support for the protective function in his study of volunteering to
be a bone marrow donor. His results showed that individuals had a greater
level of commitment to volunteer when the salience of personal responsibility
for others was high.
A sixth function is the enhancement function in which volunteerism
serves to enhance an individual volunteers self-esteem, self-confidence, and
self-improvement (Clary et al. 1998; Houle et al. 2005). Results of related
studies have found support for the esteem function. For example, volunteers
working in mental hospitals showed an increase in self-acceptance as a
consequence of their volunteer participation (Holzberg et la. 1964; King et al.
1970).
Based upon the approach of functional analysis, volunteerism may serve
more than one motive for an individual and different motivations may be
served within a group of volunteers performing the same activity (Houle et al.
2005). Moreover, in the same individual, different motives may be primarily
engaged by different volunteer activities.
Paid v.s. non-Paid
Volunteer work in the human services has many similarities to paid work
(Gidron 1983). According to Galindo-Kuhn and Guzley (2001), both paid
and unpaid workers interact with the organization and other people in the
organization, and they also have certain expectations about what their
participation will provide for them. Whether paid or unpaid, workers are
63


provided with a job that they are expected to do to the best of their abilities.
It involves a situation where there is a job to be done, the job can utilize ones
skills and creativity, ones efforts can bear fruit in the form of results or
achievements, and one can be recognized for it (Gidron 1983). Nonetheless,
volunteer work is somewhat different from paid work.
The volunteerism literature (e.g., Gidron 1983, 1984; Cnaan and
Goldberg-Glen 1991; Galindo-Kuhn and Guzley 2001) makes a distinction in
motivations between volunteers and paid labor. The dissimilarities are
described as follows:
First, volunteer work is by definition an act of free will; individuals
engage in it and discontinue it at will. In contrast, in paid work, the pay
element represents for most people the necessity to work, the fact that one has
to work in order to live. It is related to ones survival. This notion is
well-grounded in most cultures. It means a different form of relationship to
the workplace, a different form of compliance (Gidron 1983).
Second, volunteer work per se is usually not related to an occupational
career in the field in which one works (Gidron 1983). If a volunteer decides
to develop a career in the field where he or she volunteers, the volunteer
usually crosses the lines and joins the paid work force (Gidron 1983).
Third, volunteerism involves volition. Implicit in the very word
volunteer is the root concept of volition (Ellis and Noyes 1990). There is
volitional nature to unpaid work that goes above and beyond any economic or
social necessity; volunteers choose to engage in unpaid work simply because
64


that is how they choose to spend their leisure time. In contrast, paid work is
subtly coercive in its origin.
Fourth, volunteering and going to work represent vastly different
psychological approaches to organizational participation (Pearce 1983). The
primarily expressive orientation of volunteer work can best be described in
terms of whether there is a goal of social responsibility (Ellis and Noyes 1990).
People engage in volunteer work because they want to help others (Schram
1985; Cnaan and Goldberg-Glen 1991; McSweeney and Alexander 1996).
This orientation can also be found among some paid workers, but paid work is
predominantly instrumental in orientation (Galindo-Kuhn and Guzley 2001).
The primary emphasis for most paid workers, even paid work in the public and
nonprofit agencies, is towards responsibility for self and self-benefit (Mirvis
and Hackett 1983; Pearce 1983; Vinokur-Kaplan et al. 1994).
Fifth, the perceived value of reward that is obtained by volunteers is quite
different from by paid workers. Volunteers attach a stronger sense of reward
value to the incidental outcomes of the work experience such as friendships
(Mirvis and Hackett 1983; Pearce 1983; Vinokur-Kaplan et al. 1994). The
traditional work reward of material compensation is of greater value to the
paid worker.
Satisfaction
According to Gidron (1984), what originally motivated the volunteer to
get involved may not be sufficient to sustain their involvement in the long
65


term, unless they feel satisfied with their volunteering experience.
Satisfaction and psychic benefits of various kinds are not just a by-product for
those engaged in volunteer work, but are expected by volunteers (Smith 1981).
Volunteer work is perceived as an exchange between the volunteer and his/her
work situation (Sharp 1978; Kemper 1980), whereby time and effort are
exchanged for satisfactions and psychic rewards to the individual (Qureshi et
al. 1979).
The descriptive and prescriptive literature on volunteering suggests that
volunteers should be satisfied on their job in order to persevere with it (Naylor
1967). Satisfaction increases the likelihood of predicting retention-related
outcomes, namely turnover potential (Galindo-Kuhn and Guzley 2001).
Individuals who believed that they had satisfied their initial motivations for
volunteering were more likely to intend to continue to volunteer than those
who did not believe that they had received benefits that matched their initial
motivations (Clary et al., 1998). People continue to volunteer because they
enjoy what they are getting from the experience; they value the rewards they
are receiving and want to maintain and expand them (Gidron 1983, 1984).
What constitutes satisfaction from volunteer work? What is it in the job
that volunteers find satisfying? Volunteer satisfaction is founded in a link
between motivations, expectations and actual experience (Ralston and Rhoden
2005). However, little research exists on volunteer satisfaction, since for
many years volunteer work was perceived as a purely altruistic act in which
the volunteer is treated as a benevolent person who sacrifices something of
66


oneself in order to give to others (Gidron 1983). This perception of volunteer
work creates the bias against viewing volunteer work as a satisfying endeavor
can cause volunteers to hesitate to discuss their real feeling about their work
freely and openly (Gidron 1983).
By conducting a multiple regression analysis, Gidron (1983) identified
twelve factors that have the strongest effect on overall volunteer satisfaction.
The twelve variables are work itself, task achievement, task convenience,
stress factors, family (outside support), supervisor-instrumental (information),
supervisor-expressive (emotional support), professionals (staff relationships),
perceived social acceptance of volunteer work, client, recognition, and other
volunteers. Gidrons (1983) work is reinforced by Cnnan and
Goldberg-Glen (1991) and Galindo-Kuhn and Guzley (2001) who, in an
attempt to predict volunteers intention to remain, correlate wider motivational
factors with more prescriptive volunteer expectation and satisfaction factors:
quality, clarity and adequacy of communication and job-related information;
feedback and recognition; suitability, convenience and autonomy of work
assignment; importance of their role, benefits to others, fulfillment of intended
contribution; quality of training, emotional and organizational support; social
aspects and group integration (Galindo-Kuhn and Guzley 2001).
Volunteers satisfaction increased the more they became involved with
the organization and took on additional responsibilities (Miles et al. 1998).
Satisfying the volunteer and meeting their task-related expectations,
particularly at the beginning of the volunteer-organization relationship, is
67


more important than satisfying the needs of organization if volunteer attrition
is to be avoided. Volunteers who enjoy their work, who think their work
makes a difference and who believe their efforts are of value are more likely to
persevere (Ralston and Rhoden 2005).
Organizational Commitment
Organizational commitment is defined as the identification with a
particular organization, willingness to exert considerable effort on behalf of
the organization, and the desire to maintain membership in the organization
(Porter et al. 1974; Mowday et al. 1982; Igbaria et al. 1991; Igbaria and
Greenhaus 1992; Igbaria and Guimaraes 1993; Thatcher et al. 2002-3). This
affectively-oriented definition for the basis of commitment recognizes the
importance of emotional attachment to the organization (Dailey 1986).
Organizational research on volunteers has shown that volunteers are important
human resources relative to the productivity of human services organizations
(Johnson 1981; Gamm and Kassab 1983). If volunteers are well managed
then they are likely to continue their contributions and influence others to
become volunteers (Dailey 1986).
In reality, research on volunteers organizational commitment is sketchy.
Related studies have most often been concerned with profiling the personality
characteristics of the volunteer for various volunteer activities (Dailey, 1986).
Pearce (1983) reported that volunteers tended to have higher socioeconomic
status than non-volunteers and they were less likely to leave their
68


organizations than paid employees working in comparable organizations.
Werner (1976) found individuals with moderate levels of independence and
nonconformity, plus a willingness to reject authoritarian attitudes to be more
likely to volunteer. These studies confirm findings in the organizational
behavior literature since researchers routinely describe willingness to stay with
the organization as a facet of the organizational commitment construct (Porter
etal. 1974).
Organizational commitment has been regarded as a useful index to
predict future levels of volunteer involvement (Cuskelly 1995). It is formed
by subsequent attitudes of job satisfaction (Bateman and Strasser 1984).
According to Gamm and Kassab (1983), substantial proportions of decisions
to volunteer and exert sustained effort on behalf of the organization are
functions of higher order need strength. Thus, opportunities to satisfy these
needs are prominent precursors for sustained job satisfaction among volunteer
workers (Dailey 1986). Additionally, the more focused attitudes of job
satisfaction and job involvement can be partially sustained through the
positive impact of organizational commitment (Dailey 1986).
Job satisfaction has a well documented history of being related to
organizational commitment (Hall and Schneider 1972; Porter et al. 1974;
Stevens et al. 1978). Most researchers cast job satisfaction as a predictor of
organizational commitment (Hall and Schneider 1972; Stevens et al. 1978;
Bateman and Strasser 1984). However, Steers and Mowday (1981) argued
that commitment involves a wider individual perspective regarding the entire
69


organization while job satisfaction centers on the persons reactions to specific
facets of the job (e.g., co-workers, pay, promotions, the work itself, and
supervision). He indicated that job satisfaction is very volatile depending on
rapid changes in the environment of work while organizational commitment
develops more slowly but consistently over time. Research has shown that
both job satisfaction and organizational commitment have central roles in the
job scope and turnover literature and they have been shown to be related to
intention to stay and actual turnover behavior (Mowday et al. 1982; Bateman
and Strasser 1984).
Job involvement is an important predictor of organizational commitment
(Dailey 1986). It is a work attitude which develops at the level of person-job
interaction while organizational commitment forms as an affective attachment
to the organization. Job involvement is conceptually defined by Rabinowitz
et al. (1977) as the degree to which employees identify with their jobs, the
degree to which employees actively participate in their jobs, and the degree to
which they believe that their jobs are important determinants of their
self-worth. Job involvement focuses on the affective relationship between
the person and work, and thus it is understandable that job involvement is
viewed as an individual-situation construct (Dailey 1986). Empirical
findings in the literature of individual differences and job scope indicate that
job involvement is a logical choice for predicting organizational commitment
among volunteer workers (Dailey 1986).
70


Retention and Turnover
Volunteer retention, in its simplest form, is making volunteers feel good
about their assignment and themselves (Lynch 2000). If the volunteer
experience makes the volunteers feel good, then they will continue to want to
volunteer (McCurley and Lynch 1996). Retention has an evil twin: turnover
or the number of volunteers who leave the organization that have to be
replaced (Skoglund 2006). Retention and turnover are important variables to
volunteer program managers because they present serious problems for
organizations that depend on volunteers to execute their mission statement
(Skoglund 2006). Flowers and Hughes (1973) argued that the concepts of
retention and turnover are inseparable since any effort to reduce turnover must
take into account the reasons people stay and the reasons people leave.
Retention
Retention is the ability to keep volunteers involved (Jamison 2003). It
is operationalized as active volunteers who have maintained the same level of
service they started volunteering at the agency (Jamison 2003). This term
refers to the number of volunteers who successfully complete their initial
commitment to agencies, including those who renew and continue serving at
the agency (Connors 1995). McCurley and Lynch (1996) suggested that
studies of volunteer retention have determined that the first 6 months of
volunteers experience is critical toward their retention, as the greatest losses
of volunteers occurs during this period.
71


According to Skoglund (2006), volunteers start their service in a
honeymoon stage, which is composed of euphoria, self-congratulation, and
eagerness to give of themselves. Upon gaining some experience, volunteers
regress to a post-honeymoon blues phase. The idealism motivating their
initial endeavor has now dissipated. This regression may occur when
volunteers realize they are not able to accomplish what they had initially
anticipated or when they realize that the organization does not represent the
values or issues they originally thought. Such realization put a damper on an
individuals initial motivating forces, and it merely becomes a matter of time
before the volunteer steps out of the role of servitude. Volunteers also
require more attention at anniversaries at the end of large projects, or at the
completion of an agreed term of participation (McCurley and Lynch 1996).
Based upon Skoglund (2006), several things contribute to a positive
volunteer experience, which in turn increases retention and reduces the risk of
turnover. First, retention of volunteers is accomplished through the
development of feelings of importance and belonging to a particular agency
(Murk and Stephan 1991). Stams and Wymer (2001) argued that volunteers
will be satisfied with volunteering if they have the chance to develop
friendship, share experiences, communicate with others, and develop support
groups.
Volunteers will feel positive about their experience if they have an
opportunity to cultivate their role identity (Grube and Piliavin 2000). Role
identity is defined as ones concept of the self that corresponds to the social
72


roles held by the individual (Grube and Piliavin 2000). A volunteer should
perceive his or her role as important to the overall success of the organization.
When this occurs, self-esteem should be increased, thereby fostering
commitment to the individuals role identity as volunteer. In essence, general
role identity as a volunteer will predict volunteer role performance (Skoglund
2006).
Finally, a positive volunteer experience can also be achieved when the
volunteer experience new learning opportunities with the potential for
personal or professional growth (Stams and Wymer 2001). Stams and
Wymer (2001) indicated that one of the most frequent motivations for
discontinuing volunteer service is inadequate training. Training for
volunteers is overlooked when agencies view their volunteers as employed
professionals who are just giving of their spare time. In contrast, that
someone is a professional does not mean that person does not need training in
volunteer work (Logue 2001). Training not only helps volunteers work better,
but also helps to motivate them to donate time (Skoglund 2006). There is no
money to pay volunteers so we always have to work on motivation ...
experience is the best way for volunteers to learn and stay motivated (Logue
2001).
Turnover
The term turnover is operationalized as inactive volunteers who had a
prior affiliation with their agencies but have decreased the number of hours
73


per week (or month) or who have completely stopped volunteering at the
agency (Jamison 2003). In deciding whether to quit a job, paid and volunteer
employees undoubtedly differ in the extent to which they weigh various
concerns (Miller et al. 1990). Volunteer turnover is to be expected in
volunteer organizations and creates opportunities for organizational change,
but high rates of turnover can hinder the capacity of organizations to deliver
the quality or range of services and programs clients come to expect (Razzak
2001). Volunteers may leave agencies because they relocate to another area;
because they become full-time employees, students, or parents; or because
they experience health or transportation difficulties (Jamison 2003).
Miller et al. (1990) suggested that turnover among volunteers, as with
paid employees, can have some benefits. The departure of some volunteers
may create opportunities for others to move into doing work that they prefer,
and new volunteers who are brought in to fill vacancies may be better
performers or have new ideas that result in positive changes for the
organization (Dalton and Todor 1979; Staw 1980; McEvoy and Cascio 1987;
Watts and White 1988). However, high turnover rates are critical when there
is a need for volunteers with special skills or intensive training; volunteer
responsibilities that require long-term commitments; the clients served by
volunteers are disrupted by the absence of volunteers; and there is a shortage
of qualified volunteers (Stams and Wymer 2001).
Due to the nature of the organizations that tend to rely on volunteer work,
turnover can have very adverse effects on service delivery and financial
74


resources (Miller et al. 1990). For examples, the resignation of a volunteer
tutor can seriously disrupt learning in a literacy program, and the resignation
of a Big Brother or Big Sister can, contrary to program goals, contribute
to a child feeling abandoned and unwanted. Moreover, many organizations
rely on the services of volunteers in part because they are understaffed and
strapped for funds; constantly having to recruit and train new volunteers can
be a severe financial drain.
Unlike most paid employees, volunteers are free to resign without first
locating alternative employment (Miller et al. 1990). Therefore, those
individuals who manage and oversee volunteer programs face a twofold
challenge: orienting, training, and monitoring volunteers as well as retaining
these volunteers (Forsyth 1999). An effective orientation and training of a
programs volunteers will engage participants in a way that results in
volunteers willingness to participate in the agencys volunteer program for a
significant period of time. However, if there is a breakdown in one of these
functions, it is not long before an organizations volunteer program starts to
flounder (Skoglund 2006).
Determinants of Volunteers Effort
Volunteers provide organizations with many vital resources in the form of
expertise, skills, knowledge, and labor. Volunteer work is regarded as an
expenditure of time and effort on behalf of an organization or another
individual (Mutchler et al. 2003). According to Dorsch et al. (2002), one
75


factor that determines the value of these volunteer resources is the amount of
effort volunteers expend while engaged in their voluntary activity. Dorsch et
al. (2002) conducted a survey on several volunteer organizations (e.g., sports,
culture, recreation), and found that four important factors were the most useful
for predicting volunteer effort. These factors described as follows are role
acceptance, role clarity, specific aspects of satisfaction, and role efficacy.
Dorsch et al. (2002) suggested that the most important factor influencing
volunteer effort was role acceptance. When people volunteer with an
organization, they are either recruited to fill a particular role or are assigned to
a role the organization needs to fill. In both instances, volunteers must
accept the responsibilities of the role they occupy. Survey suggested that
volunteers who accept the responsibilities associated with their role are more
likely to expend greater effort to carry out their tasks than those who do not
accept their responsibilities.
Based upon Dorsch et al. (2002), role clarity was the single strongest
contributor to role acceptance, and also was important in determining
volunteer effort. Volunteers need to have a clear picture of what they will be
doing, where they are doing it, and how it fits into the bigger picture. If
volunteers are not certain about the specific nature of their roles, it is very
difficult for them to accept those roles and work hard on the tasks associated
with them. Role clarity was influenced by two most important factors:
satisfaction with the organizations performance and role efficacy. A
volunteers level of satisfaction with how the organization is performing (e.g.,
76


meeting goals) has a direct impact on his or her perception of how clearly
defined the volunteer role is in the organization. Organizations with clearly
defined goals tend to have clearly defined roles for their volunteers.
Moreover, it tends to be easier to determine whether the organization is
performing well if the organization has clearly defined goals, which results in
greater satisfaction among volunteers.
Satisfaction with social service was a moderate predictor of role
acceptance, and thus contributes indirectly to volunteer effort. Volunteers
who are satisfied that they are being of service to others tend to put more
effort into their activity. This factor must be seen in light of volunteer
motives. Dorsch et al. (2002) suggested that two most important motives for
volunteering are to help the community and other people. If the most
important motivation of volunteers is to help and their perception is that they
are doing so, then they will be much more accepting of the roles they have
been assigned. This type of satisfaction also contributes to role clarity, when
volunteers are satisfied that they are helping their community and other people
they believe that their role has some meaning in the larger societal context.
Volunteers roles not only must be clearly stated and individuals perceive
that they are contributing to the welfare of others, but volunteers must feel
confident that they have the ability to carry out their assigned tasks. Role
efficacy was not as strong a predictor of role acceptance as role clarity, but it
was still a factor influencing volunteer effort. Moreover, role efficacy also
77


contributes to role clarity, so it cannot be overlooked when considering how
much effort volunteers will expend.
Generally, organizations that want to achieve high levels of effort from
their volunteers must ensure that volunteers accept their roles. This means
ensuring that volunteers should have clearly defined roles, understand these
roles, feel a sense of confidence in their ability to fulfill their roles, are
satisfied with the extent to which they perceive themselves as helping their
community and others, and are satisfied with the organizations overall
performance.
Example Models
Empirical studies on volunteers have been presented in the literature
concerning the psychological mechanisms underlying individual differences in
volunteers behaviors and motives, and the nature and extent of their
participation in volunteer activities. Several models have been empirically
tested to validate volunteerism theories. The followings are example models.
Mowen and Sujan Model (2005)
Mowen and Sujan (2005) conducted a three-phase study to explore the
antecedents of volunteer behavior within a hierarchical model of motivation
and personality shown in Figure 3.1. The goal of their study was to identify
individual difference variables predictive of a set of volunteer behaviors and
the relation between a functional motive approach and a trait approach for
78


predicting volunteer behavior. The functional motive approach seeks to
identify the reasons and purpose that motivate a person to engage in behavior.
In contrast, the trait approach was used to find enduring dispositions that
influence behavior. The model was measured with a confirmatory factor
analysis and the data were collected from the surveys of 600 members of a
consumer panel run by a marketing company, 138 students at a Midwestern
university and 630 members of an organization promoting volunteerism at the
community and state levels.
Compound
Traits
Situational
Trait
Functional
Motives
Outcomes
Figure 3.1 Trait plus Functional Motive Model
79


In the first phase study, Mowen and Sujan developed measures of
altruism and volunteer orientation and combined them with four compound
traits (need for activity, need for learning, altruism, and present time focus) to
predict a set of volunteer behaviors. In study 2, they examined the use of
functional motives to predict volunteer behaviors. The functional motives
developed and tested by Clary et al. (1998) included learning, help others,
make friends, self-enhance, career, self-protection. In study 3, Mowen and
Sujan used an overarching model of individual difference regarding
motivation and personality, developed by Mowen (2000), to examine the
relationships of functional motives with compound and situation traits.
Mowen and Sujans study had several intriguing results. First, volunteer
orientation and the functional motive of helping others were positively related
to volunteer behaviors. Second, the situational trait of volunteer orientation
was a significant predictor of volunteer behaviors and all functional motives.
Third, altruism was a significant predictor of volunteer orientation. Fourth,
the self-enhancement functional motive was negatively related to volunteer
behaviors.
Miller, Powell, and Seltzer model (1990)
By conducting a survey of 158 active volunteer workers at two hospitals,
Miller et al. (1990) examined the causal sequencing of attitude, personal
situations, and behavioral intentions as determinants of turnover among
hospital volunteers. They intended to investigate whether turnover can be
80


explained by factors analogous to those included in models of turnover among
paid employees. Therefore, Miller et al. developed a research model to study
the influences of three general types of presumed determinants of turnover:
attitudes of volunteers toward their volunteer assignments and organizations
(i.e., job satisfaction, organizational commitment, satisfaction with the work
itself), personal situation factors (convenience of schedule, experience)
believed to be particularly relevant for volunteers, and intentions to quit the
volunteer assignment. Additionally, age was added to the model to be tested
in order to examine its role as a determinant of turnover, since studies with
paid employees have frequently found that age is inversely related to turnover.
Miller et al.s model depicted in Figure 3.2 was assessed by structural equation
modeling.
Figure 3.2 Hospital Volunteers Turnover Model
81


Miller et al.s analysis focused on understanding volunteer turnover at the
individual level and examined psychological predictors of turnover. They
argued that volunteers are a critical human resource for not-for-profit
organization (NFPs) because they are frequently used in lieu of paid
employees. As with paid employees, turnover among volunteers can be an
important problem and major cost factor (Miller et al. 1990). Prevailing
models of turnover among paid employees have proposed that attitudes and
personal situations affect behavioral intentions to quit and that, in turn,
behavioral intentions affect turnover (Miller et al. 1979; Mobley et al. 1978,
1979; Steers and Mowday 1981). Miller et al. (1990) argued that turnover
among volunteers may result from the same general influences and
decision-making processes as turnover among paid employees.
According to Miller et als findings, attitudinal factors and the volunteers
personal experience influenced turnover indirectly with intention to leave
acting as a mediating factor. Of the variables originally designed as potential
predictors, only convenience of schedule had a direct effect on turnover. In
addition, the results revealed that age played an important role in determining
turnover for volunteers, much as it did for paid employees. Miller et al.
concluded that these findings were consistent with several studies of turnover
among paid employees and of behavioral domains other than turnover.
82


Dailey model (1986)
Figure 3.3 Volunteer Workers Organizational Commitment Model
To measure a proposed model of organizational commitment for
volunteer workers, Dailey (1986) administered a survey of 138 campaign
workers who volunteered to work as solicitors of contributions and
coordinators of fundraising activities for a major national charitable
organization. Dailey intended to examine the personality, job characteristics,
and attitudinal antecedents of organizational commitment for volunteers, since
prior research did not shed light on the psychological processes which underlie
organizational commitment among volunteer workers. The proposed model
shown in Figure 3.3 was empirically tested with multiple regression analysis.
Daileys studies including job scope constructs, individual differences,
job satisfaction, and organizational commitment had been cross-sectional in
design. Dailey did not conduct causal tests of hypothesis directionality, since
83


the importance of the constructs in the literature of individual attitudes and
behavior in the work place was doubtless. However, this assertion cannot be
applied with equal confidence to the scant literature of commitment for
volunteer workers. Therefore, given the central theme of the organizational
commitment construct in the organizational behavior literature, it was
reasonable to enhance understanding of the construct and its theoretical
antecedents for volunteer (Dailey 1986). Daileys research model shown in
Figure 3.3 presented a synthesis of the literature for prior studys constructs,
and it was exploratory and descriptive in nature.
According to Dailey, the importance of job characteristics and perceptual
judgments about job design variables cannot be underestimated for volunteer
workers. The findings indicated that the variables used to study work
attitudes and organizational commitment for paid employees were associated
with the outcomes for volunteers. Additionally, Dailey suggested that
researchers studying volunteerism should incorporate task characteristics into
their research and they need to recognize there is a wide range of behaviors
and attitudes that materialize and drive volunteer activity well after the
decisions to join and donate energy and time have been made.
Empirical Findings
After an extensive review of volunteerism, we can find researchers in a
variety of disciplines have investigated volunteerism or related constructs.
Most of these studies were based either on a conceptual analysis including
84


summary reports of a specific project or on an empirical analysis of certain
theories. The empirical findings are summarized in Table 3.2, and the
highlighted arguments represent they will be brought forward to examining the
motivation of OSS developers and their behavior in this study.
Table 3.2 Research findings on volunteerism
Topic Findings
Motivation * Volunteering can reflect multiple motivations (including altruistic, instrumental, social, self-esteem, and other goals) that can be collapsed into one or more factors or dimensions. * Some theorists have considered two basic motivations for volunteering: to satisfy self-regarding or instrumental interests and to satisfy other-regarding or altruistic interests. * Volunteers act not from a single motive or a category of motives but from a combination of motives that can be described overall as a rewarding experience. * Motivations to volunteer include anticipated benefits of the activity for other individuals and groups as well as perceived benefits for the individual engaged in the activity. * Volunteerism can be the key to gaining higher levels of specific knowledge, skills and experience, which may enhance existing ones and/or prove useful in the future. * The pleasure of new-found knowledge, seeking out new challenges and the opportunity to develop new skills and explore career options were important motivators and satisflers. * The opportunity to mix with people with a similar interest and to share their views was also seen as an important motivator and satisfier. * A positive volunteer experience can also be achieved when volunteer experiences new learning opportunities with the potential for personal or professional growth.
85


Full Text

PAGE 1

AN EMPIRICAL ANALYSIS OF OPEN SOURCE SOFTWARE DEVELOPERS MOTIVATION USING EXPECTANCY-VALENCE THEORY by Chomg-Guang Wu B.B.A., National Cheng Kung University, Taiwan 1985 M .S., State University ofNew York at Binghamton 1991 A thesis submitted to the University of Colorado at Denver and Health Sciences Center in partial fulfillment of the requirements for the degree of Doctor of Philosophy Computer Science and Information Systems 2007

PAGE 2

This thesis for the Doctor of Philosophy degree by Chomg-Guang Wu has been approved by James H. Gerlach Ronald V. Ramirez )l3-o7 Date

PAGE 3

Wu, Chorng-Guang (Ph.D. Computer Science and Information Systems) An Empirical Analysis of Open Source Software Developers' Motivation using ExpectancyValence Theory Thesis directed by Prof. James Gerlach ABSTRACT The purpose of this study was to investigate the motivations of individuals that are willing to join open source communities and voluntarily dedicate their effort and expertise for OSS development. Despite the emergence of various studies on developers' motivation there is little research that focuses on developers' intentions to continue their participation and expend their effort. Expectancy-valence theory human resource theory, and volunteerism were adapted to form two research models: a theoretical model of OSS developers' continuation, and an exploratory model of developers' effort. The continuation model captures those factors that might significantly influence OSS developers' satisfaction with OSS development, their commitment and intentions to continue participating in future open source projects, and the causal relationships among these factors. The effort model

PAGE 4

is developed to measure the associations between developers' effort and motivation, schedule availability, and expertise. The findings regarding the motivation of developers showed that volunteering can reflect multiple motivations: intrinsic motivation (i.e., helping) and extrinsic motivation (i.e., career advancement) A positive and satisfactory experience leads to a positive attitude toward retention Moreover, volunteer developers' commitment primarily depends on motivation to help. These results, taken as a whole, suggest that OSS developers are motivated both by the altruism of helping and the economic incentive for career advancement. Developers' effort was primarily influenced by both their motivation to help and peer recognition, which implies that the impact of intrinsic motivators on effort is stronger than that of extrinsic motivators. Enjoyment demonstrated a significantly negative effect on effort, suggesting that developers with high levels of motivation on enjoyment did not spend more time on open source projects than those with less motivation on enjoyment. This study also provided empirical evidence that volunteers may vary their motivation according to their actual experience and the commercial viability of their project. The length of experience suggests a life cycle effect on motivation; i.e., the longer developers stay, the higher levels of motivation grow. In addition, the more commercially viable an open source project is, the more likely developers are motivated. Finally, scheduling availability is an important factor of predicting continuation and effort.

PAGE 5

This abstract accurately represents the content of the candidate's thesis. I recommend its publication Signed

PAGE 6

ACKNOWLEDGEMENT I would like to acknowledge the members of my doctoral committee for helping me make this dissertation possible. First of all, I would especially like to express my sincerest gratitude and appreciation to my adviser, Professor Jim Gerlach, for his mentorship and commitment. Throughout my doctoral work, Jim gave me the freedom to explore on my own and at the same time the guidance to recover when my steps faltered. He continually encouraged me to develop independent thinking and research skills His patience and support helped me overcome many crisis situations and finish this dissertation I hope that one day I would become as good an advisor to my students as Jim has been to me. I would also like to thank Professor Clifford Young for assisting me with the quantitative analysis. Professor Young s insightful comments were thought-provoking and they helped me improve my knowledge in the area of structural equation modeling. I am very thankful to him for enforcing strict validations for each research result. I am also very grateful to Dr. Ramirez and Dr. Cios for reading this dissertation and offering constructive comments. In addition, I owe a special note of gratitude to Dr. Mannino for his generous help during my study at this University.

PAGE 7

Finally, I would like to thank my family for their life-long love and support. Especially, my mother has been a constant source of concern, support and strength all these years. I would like to express my deepest gratitude to her

PAGE 8

TABLE OF CONTENTS Figures ............. .... ...... ..... ..... ..... ............ ............. . .... ..... .......... ........ .... . ...... xiii Tables .. ... ........ .......... . ....... ........ .... ........ .... .... . . ........... . .... ............... .... . ... xiv CHAPTER 1. INTRODUCTION ..... .... .............................. . ............. ........ ........... ....... 1 Purpose of the Study . ....... . .... ...... .... ........ ...... ....... .......... ...... .... 5 Statement ofthe Problem .... .... ........ .......... ..... . ....... .... . ............... 6 Research Questions ................... . .... ................ ....... ........ .... ........ 8 Reader s Guide . . .... ......... . .... .... .... . ........ . ............ ............. ..... 9 2. LITERATURE REVIEW OF OPEN SOURCE SOFTWARE ... . . .... ... 13 Definition of Open Source Software ..... ...... ........ ..... ...... .... ......... 13 Evolution of Open Source Development.. ......... . . .... .... .... . . ...... 15 Types of Open Source Software Licenses ....... . ...... . ............ ...... 21 The BSD License ..... .... . ............................. .... . ............. ... .. 22 General Public License ...... ........... . . ...... ....... . .... ........ ....... 23 Lesser General Public License .... ........ ..... ... .......... ........... ... 24 Mozilla Public License ............. ....... ... . ........ . ... .. ... . .......... 25 IBM Public License ............. ..... ..... ........ . ............ ..... ...... . ..... 26 Open Source Communities .... .......... .... . .... ......... ................... .... 27 Open Source Participants ..... ....... ...... ..... ...... . . ......... ..... . .... 27 Structure of Open Source Communities .... . ........ .... . ..... ...... 28 Mode of Software Development ... ...... . .... ..... ........... ........... 29 vii i

PAGE 9

Process of Submissions Acceptance . . .... . ...... ... ...... .... . . ... .31 Communities of Practice ........................................................ 3 2 Private-Collective Innovation ...... ...... ....... ........ ........ ............ 34 Prior Research on OSS Developers .... .... .......... ...... .......... .... ........ .36 Studies on Motivation ............................................................ 36 Intrinsic Motivation ......... ....... ..... .............. ..... ............... 37 Helping Behavior ............................... ...... : ............ 38 Enjo y ment .... . ....... . .... ..... ...... .......... . .... .......... . 40 Peer Recognition ........ ............ . ........ ........... .... ...... 41 Extrinsic Motivation ...... ............. ....... ........ ......... ......... .42 Enhancing Human Capital .... .... ............................ .42 Career Advancement ............. .... ............. ............... 43 Satisfying Personal Needs ....................... ...... .... .... .44 Effort and Performance .... ...... ............... ........................ ...... .45 Effort ....... ............. .......... ............................ .... ..... ........ 45 Performance .. ..... . .... ..... ......... ......... ..... ......... ... ............. 47 Example Models .................. ... .............. ........... .................. .... 48 Wu, Gerlach, and Young model (2007) .. .......... .... ........ .49 Roberts Hann, and Slaughter model (2006) .......... ...... .. 50 Summary ...... . ..... ... . ...... .... . ...... . .......... . ..... ...... . ...... ... . .... ... ...... 53 3 LITERATURE REVIEW OF VOLUNTEERISM ...... ...... .... ................. 55 Volunteerism ... ...... . . . ...... .......... ...... ................. .............. . ......... 56 Motivation to Volunteer .......... .... ....... ......... ..... ...... ................. ...... 58 IX

PAGE 10

Reasons for Volunteering ........ ..... .... ,. ..... .............................. 58 Functional Approach ........ ................ .... ........ ........ ........ .... ...... 60 Paid v.s. non Paid ...................... ........ .......... ........... .... ........... 63 Satisfaction ............ ......... . ........... . .... .... . .... ........ ... . .... ...... . ...... 65 Organizational Commitment ... . ...... . ....... .. ............ ....... ... . ........ 68 Retention and Tumover .... .................... .... ........ .... .... ........ ....... ...... 71 Retention ..................... ..... ....... . .... . ...... ... .... ... ...... . ......... 71 Tumover ... ...... ....... ... ...... .... . .... ......... .... .... ... .... ... ..... ... ..... 73 Determinants of Volunteers' Effort ........ .......... ...... .... .................... 75 Example Models ......... ......... ....................................... .... ............... 78 Mowen and Sujan Model (2005) ................ ...................... ..... 78 Miller Powell, and Seltzer model ( 1990) ........ ........ .............. 80 Dailey model (1986) .... ...... ........ ........ .......... ............ ............ .. 83 Empirical Findings .... .... ...... . . .... ....... . .... ... .... ........... .... . ....... ... 84 Summary ......... . .... .... ....... ... ............ .... .... ...... . .................. . .... ...... 88 4 THEORETICAL PERSPECTIVES ......... ......... .......... ........ .... ... .... ....... 90 Motivation ....... . .... . ......... . ......... . . ...... .... ... . ... .... ........... ......... 90 Expectancy and Valence .... . . ...... ...... ... .......... ...... ............ ... 93 Satisfaction ...... .... .... ...... ........ .... . ....... . . ... .... ......... .. ........... .... 95 Open Source Commitment.. ...... ... ........ ............. .... ..... .... ................ 97 Continuance Intention ... ...... ... .. ..... .... ........................ ..... ......... ....... 98 Availability ............. .... ..... .. .......... ..................... .................... .... ...... 1 00 Expertise .......... .... ........................... ........................................ ....... 1 0 1 X

PAGE 11

The Influential Factors ofMotivation .... ................... ........ ............. 102 Project Characteristics ....... ........ .................. ......... ...... ...... ... 1 02 Years of OSS Experience ....................................................... 03 Participation Status ................................................................ 1 03 Summary ... ...... ...... ..... ............. ...... ........ ...... ........... ......... ..... .... ... 1 04 5 RESEARCH MODELS ..................... .... ...................... .............. ........ 105 Continuation Model ....................................................................... 1 05 Effort Model. .... . ..... ...................... .... .... .............. . .... ......... ...... 114 6 METHODOLOGY ................................................................................ 118 Identification of the Research Models . ................ ...... ............ . ..... 118 Continuation Mode1 ............. .... ... .... .......... ............... .............. 120 Effort Model. ................................. ................... ................ ... 123 Instrumentation ............ ........... ............... ............................ ....... . 125 Pilot Test ....... ........ ........ .......... ....... ............... .......... ............. ......... 12 7 Sampling and Data Collection ................. ........... ......... .................. 130 Data Analyses ................................................................................. 134 Reliability ............................................................................... 13 5 Measurement Model Assessment.. .......... ................. ........... ... 13 7 Continuation Mode1 ....... ...... . ...... ................................. 138 Effort Model. .................................................................. 140 Structural Model Assessment ................................................. 143 Continuation Model . ................................ .......... ....... .... 143 Effort Model. ............... ...... ............... .... ... ................ .... ... 146 xi

PAGE 12

The Analysis of Influential Factors of Motivation .... . .... .... ... 148 Project Characteristics ....... ......... ...... ....... . ........... . .... 148 Years of OSS Experience ....... .... ...... ..... . ... ... ..... . .... ... 150 Participation Status ..... ........... ... .... ....... ..... .......... ..... ... 152 Limitations ... ....... ... .. ..... ..... ...... ... . ... .... . ... ...... . ... .... .... . .... .... . 153 7 DISCUSSION AND CONCLUSIONS ........ . .... ........ . .... ... ......... . . ... ..... 157 S urn mary of Results .................. .... . . ...................... ............. . .... .... 15 7 Implications .... ................... ... .... .... . .... . ...... . ...... ... ... .... ...... .... 160 Additional Research . .................. ...... . ................. . ...... ...... ........ 164 Contributions .......... .... .... . .... ...... . .... . ............. . ........ ... ....... .... ... 165 Conclusions ...... .... .... ...... .... ......... . . ........... ........ .... ... . ...... ...... 166 APPENDIX A. HUMAN SUBJECTS APPROVAL ..... .... .... ...... ...... . . ...... ......... ..... ....... 168 B. SURVEY ITEMS ...... ........... ......... ...... . ........ ......... ..... . . ....... ...... ....... . 169 C. INFORMED CONSENT .......... .... . .......... .. ...... ... . .... ... .... ......... .......... 175 BIBLIOGRAPHY .... .... ...... ..... ...... ........... ......... ...... ......... . .... .... ......... . ..... ... ........ 177 Xll

PAGE 13

LIST OF FIGURES Figure 1.1 Conceptual Diagram ... . .............. . ....... ..... .... ......... .......... ......... ......... . ..... 4 1.2 Reader's Guide ........... ......... ........... . .... . .... ........... . ............. ......... . ....... ... 9 2 1 Open Source Participants' Continuation Model ....... . ...... ............... .... ......... .49 2.2 OSS Developers' Motivation and Performance Model... ........ ... . .................. 51 3.1 Trait plus Functional Motive Model .... . . .... ................ .... ........................... 79 3.2 Hospital Volunteers Turnover Model . . .......... ...... .......... ......... ................... 81 3.3 Volunteer Workers Organizational Commitment Model . ............... . ........... 83 5.1 Research Model of OSS Developers' Continuance Intention ........ .............. 1 06 5.2 Research Model of OSS Developers' Effort ... .............. ........ ..... ......... . .... ... 114 6 1 Measurement Model of OSS Developers Continuation Model .... ....... .... .... . 121 6.2 Measurement Model ofOSS Developers Effort Model... ........ . . ......... ....... l24 6.3 Standardized LISREL Estimations of the Continuation Model.. ....... .... . ..... 145 6.4 Standardized LISREL Estimations of the Effort Model ... . ........ ........ .. .... ... 147 6.5 LISREL Estimations of Project Characteristics to Motivations ............ . .... 149 6 6 LISREL Estimations of OSS Experience to Motivations .......... .... ............ .... 151 xiii

PAGE 14

LIST OF TABLES Table 2.1 Important events in open source history ..................................................... .15 2.2 Characteristics of open source licenses . .... .......... .... .... ............................... 22 2.3 Major participants in open source communities ............. ..... ........................ 28 2.4 Major motivations ofOSS developers ........ ............ .......... ..... ............ ....... 37 3.1 Motives for volunteering .............. . ....... ............................. ......... ........ ....... 59 3 2 Research findings on volunteerism .............................................................. 85 5 .I Construct definitions for continuation model .............................. ....... .... ... 1 07 5.2 Construct definitions for effort model .................... ...... .... ....... .... ...... ....... .115 6.1 Types and number of parameters for the continuation model... .... ...... .... ..... 122 6.2 Types and number of parameters for the effort model.. ..... . . ........... ......... . l25 6.3 Descriptive statistics and Cronbach 's alpha for the pilot test ...................... 129 6.4 Survey respondents' position in OSS projects hosted by SourceForge ....... 132 6.5 Descriptive statistics of indicator variables for the continuation model... ... 133 6 6 Descriptive statistics of indicator variables for the effort model ............ .... 134 6. 7 Reliability measurements for constructs .............................................. ....... 136 6 8 Factor loadings for the continuation model .............. ... ............ ................... 139 6. 9 AVE and correlation of constructs for the continuation model... ...... ........ ... 140 6 I 0 Factor loadings for the effort model ............. ....... . . ......... .............. ............. 141 6.11 AVE and correlation of constructs for the effort model.. ............... ......... ..... 143 XIV

PAGE 15

6.12 Testing for the equivalence of factor loadings across two groups ....... . .... . 152 6.13 Invariant mean tests of motivational constructs ........ . ....... ... .............. . ...... 153 XV

PAGE 16

CHAPTER 1 INTRODUCTION Software practitioners and academic researchers are captivated by the recent success of Open Source Software (OSS) in producing a number of highly successful software products such as Linux, Apache, MySQL, Perl, Sendmail, and Firefox. Traditionally, due to trade secret protection by the computer vendors, commercial software products are sold only in binary form and with no source code. Thus, the users have no rights to examine or modify the code. In contrast to traditional software development, adopters of OSS seek low cost solutions that they can modify or use as software components to build new or extended applications (Johnson 2002; Lerner and Tirole 2002; Appelbe 2003). OSS is developed by a loosely organized community of developers spread throughout the world and working over the Internet. Unlike a traditional organizational structure, these communities have no/few paid staff or management. However, they do provide their members with a variety of resources to develop practical and helpful software products that have occasionally displaced or significantly improved upon some commercial products. Remarkably, OSS developers contribute to software projects without necessarily being employed or recruited by any organization. Although the recent trend is for companies that benefit from OSS to loan

PAGE 17

employees to this effort, surveys have shown that the majority ofOSS developers contribute strictly voluntarily (Hars and Ou 2002; Lerner and Tirole 2002; Wu et al. 2007). The open source community is often described as a gift culture instead of an exchange culture Essentially, gift culture refers to helping behavior which includes altruism and reciprocity. In lieu of tangible rewards, givers receive psychological benefits such as the "warm glow" of sympathy or the satisfaction ofliving up to a moral commitment (Rose-Ackerman 1998). Others assert that additional intrinsic rewards such as boosting one's ego, self-determination, and community identification, or even an anti-Microsoft mentality, might provide sufficient intrinsic motivation to engage in OSS development. However, economists argue that it is very unlikely that intrinsic rewards provide sufficient incentives to explain the enormous contributions in time and effort that some participants donate to OSS since there are no obvious similar patterns of behavior in most other areas of economic activity (Schmidt and Schnitzer 2003) Many beneficiaries are well-to-do and could afford to pay OSS participants for their efforts Furthermore, altruism or reciprocity has not played a major role in other industries, so the reason individuals in the software industry are more altruistic or reciprocal than others would need to be explained In contrast, individual participants might contribute to an open source project to increase human capital, develop career opportunities, and obtain software that solves specific technical problems related to personal job 2

PAGE 18

requirements (Hann et al. 2002; Hars and Ou 2002; Lerner and Tirole 2002; Ye and Kishida 2003; Wu et al. 2007). Exploratory empirical research on the motivations of OSS developers demonstrates the validity of some of these beliefs. However, these arguments have not been tested rigorously using existing human resource theory and volunteerism. Moreover, there is little research investigating OSS developers' work attitudes (e.g., satisfaction and commitment) and intentions to engage in future open source projects once they have already become involved. The sustainability of an open source project would generally rely on developers' continual contributions. Thus, understanding OSS developers' intentions to continue their involvement in OSS projects is very important for predicting the long-term viability of OSS development. These aspects are presented conceptually in Figure 1.1. Open source communities cannot exist or prosper without the contributions of highly motivated developers who are willing to donate their effort (i.e time) and expertise. However, because these developers are often volunteers, not traditional employees, it is impossible to solely rely on employment relationships or employment contracts to manage these persons (Roberts et a!. 2006). Effort has been used in a number of previous OSS studies (Hars and Ou 2002; Hertel et al. 2003; Lakhani and Wolf2003) and provides an appropriate proxy for developers' contribution and interest in OSS projects (Lakhani and Wolf 2003 ). The literature suggests that different motivators for contributing to open source projects should have diverse 3

PAGE 19

influences on developers efforts. With respect to the success of an OSS project it is imperative to understand whether all types of motivations affect OSS developers' effort equally or in the same way Developers Availability Expertrse "' I --:.; I O (t.<.f .) -Effort ) J 1 How s the --[? proJect? Motivators Figure 1.1 Conceptual Diagram According to Olson et al. (1996), individuals persist longer and put more effort on tasks in which they expect to succeed Persons who believe that their skills set is adequate for achieving success with a new venture are motivated to exert the necessary effort (Douglas and Shepherd 2000 ; Shaver et al. 2001 ). Therefore, it is also important to examine the relationship between OSS developers' expertise and the effort dedicated to open source projects Availability has been regarded as a significant determinant of turnover / retention in the workplace or volunteer activities According to the 4

PAGE 20

volunteerism literature, many volunteers claimed that they had limited free time dedicated to volunteer activities because of existing family and professional commitments (Ralston and Rhoden 2005). Rotolo and Wilson (2006) indicated that more flexible work schedules make it easier for volunteers to commit to volunteer work. The more convenient the schedule, the less likely the volunteer is to intend to or actually quit (Miller et al. 1990) Thus, availability is examined to test whether it will influence OSS developers' effort and intention to continue their work. Likewise, the arguments regarding the relationships ofOSS developers' continuance intention with availability and the associations of effort with motivation expertise and availability are conceptually illustrated in Figure 1.1. Because monetary reward is lacking in the context of OSS development motivational processes underlying developers' behavior are likely to differ from those in traditional employment. Thus, this study also analyzes several factors (i.e project characteristics the length ofOSS experience and participation status) that may enhance developers' motivation. Purpose of the Study This study seeks to investigate OSS volunteer developers motivation, which in tum can provide a better understanding of what motivates those developers to continue making OSS contributions and to expand their effort. An additional purpose is to explore the impact of project characteristics (commercial viability), years of OSS experience, and participation status 5

PAGE 21

(volunteer or paid) on motivation among OSS developers. The following four aspects are relevant to the investigation of these issues. First, with respect to continuance intention, the research attempts to measure developers' motivation, to identify the antecedents of developers' satisfaction with participation in open source projects and commitment to OSS development, and to analyze the relationships of these factors and availability with developers' intentions to continue their involvement in future projects. Second, to evaluate developers' efforts, motivation, expertise, and availability for OSS development are taken into consideration. How these factors influence developers' effort will be considered and explored. Third, project commercial viability and years of OSS experience are examined, since they may influence developers' motivation based upon volunteerism theory. Fourth, theoretical arguments underpinning this study are based on expectancy-valence theory, human resource theory, and volunteerism. Statement of the Problem According to the literature, each individual has attitudes on many different topics, such as politics, education, and work. Work attitudes refer to a collection of feelings, beliefs, and thoughts about how to behave with respect to the jobs people currently hold and the organizations thereof (George and Jones 1996, p. 66). Theoretically, there are two major components of work related attitudes: satisfaction and organizational commitment. 6

PAGE 22

The extant open source literature regarding individual OSS developers focuses on the analysis of their motivation to get involved in open source projects There is a shortage of empirical investigation concerning their attitudes toward OSS development after participating in open source projects and their efforts associated with motivation, availability, and expertise. Are OSS developers satisfied with OSS development? To what extent are the developers committed to OSS development? Will OSS developers continue to contribute to future open source projects even if they are dissatisfied with the experience? What are the relationships between motivators, satisfaction commitment, availability, and continuance intention? Which types of motivations are likely to generate more (or less) effort? What impact will availability and expertise have on developers' effort? These questions need to be investigated with the open source context. In this study, there are two research models : continuation model and effort model. By applying the expectancy-value theory, the continuation model captures those factors that might significantly influence OSS developers' satisfaction with OSS development, their commitment and intentions to continue participating in future open source projects, and the causal relationships among these factors. The effort model is developed to measure the associations between developers' effort and motivation, schedule availability, and expertise. 7

PAGE 23

Research Questions Question 1: Does commitment to OSS development significantly influence OSS developers' continuance intention? Question 2: Does satisfaction with OSS development significantly influence developers' commitment to OSS development and/or continuance intention? Question 3: What motivators will have a significant effect on OSS developers' satisfaction, commitment, and continuance intentions? Question 4: What is the impact of future schedule availability on OSS developers' continuance intention? Question 5: What motivators will have a significant effect on OSS developers' effort? Question 6: What is the impact of current schedule availability on OSS developers' effort? Question 7: What is the impact of expertise on OSS developers' effort? Question 8: Does project commercial viability influence OSS developers' motivation? Question 9: Does the length ofOSS experience influence developers' motivation? Question 10: Does volunteer developers' motivation differ from that of paid developers? 8

PAGE 24

Reader's Guide Part I: Literature Review (Chapter 2) (Chapter 3) Perspectives on oss Perspectives on Volunteerism Part II: Theory (Chapter 4) (Chapter 5) Theoretical Arguments (EVT, HR, VL, OSS) Models Hypotheses Part III: Research Methodology (Chapter 6) Design Measurement Part IV: Research Results (Chapter 7) Findings Implications Contributions Future Work Figure 1 2 Reader's Guide 9

PAGE 25

The dissertation is divided into four major parts, shown in Figure 1.2. Part I refers to an understanding of the open source software movement and OSS developers as well as the behavior of volunteers. In this part, I present the literature review on open source software (Chapter 2) and volunteerism (Chapter 3). Chapter 2 presents OSS definition, history, licenses, community structure, developers' motivation and effort as well as performance, two example research models related to developers' motivation, and a summary of empirical findings. These aspects highlight the prevalent issues related to extant OSS research. Chapter 3 examines existing research on volunteerism. It covers volunteers' motivation, work attitudes (satisfaction and organizational commitment), retention as well as turnover, effort, three research models associated with volunteer behavior, empirical findings, and it concludes with a summary. These perspectives are valuable in understanding the issues raised by OSS developers' volunteer behavior and they emphasize aspects that are important in volunteers' work motivation. This chapter also raises theoretical implications for the exploration of paid versus non-paid volunteers and the issues regarding volunteer effort. In Part II, I introduce the base theories related to this study and depict how the theories are associated with OSS developers' behaviors (Chapter 4). In addition, I describe the development of two research models: continuation and effort for this study based on the ground theories (Chapter 5) 10

PAGE 26

Based upon the expectancy-value theory, human resource theory, and volunteerism, Chapter 4 presents extant research on traditional paid employees and volunteer workers, and it explicates how these studies can be applied to analyzing OSS developers. The topics include the expectancy-value theory on work motivation, the human resource studies on work attitudes and behavior (i.e., satisfaction, commitment and continuation), schedule availability from volunteerism, expertise from human resource management, and the influential factors of motivation from volunteerism. The issues and theoretical arguments in Chapter 4 were chosen in order to illuminate different aspects of the base theories. Chapter 5 demonstrates two research models: the continuation model and the effort model. It explains the proposed models and the interdependence among these constructs in detail. Moreover, hypotheses for each model are proposed for testing the theoretical arguments in this study. In Part III, I describe the research design based upon quantitative methods used to measure the research models (Chapter 6). Chapter 6 exhibits the research methodology used in this study. This chapter contains the model identification, the survey developed for collecting information from existing OSS developers, and data analysis. The processes of instrument design, pilot test for the instrument, sampling, and data collection are all discussed. The data analysis refers to how to measure the reliability of constructs and validate the research models. Furthermore, hypothesis testing and the limitations of this research design are spelled out. II

PAGE 27

In Part IV, I summarize the research findings, provide several suggestions and conclude this study (Chapter 7) Chapter 7 presents the discussion regarding the outcomes generated from the empirical testing of research models and the implications for academia and practitioners. This chapter also discusses the contributions of this study-and several issues therein-to future research on OSS development. 12

PAGE 28

CHAPTER2 LITERATURE REVIEW OF OPEN SOURCE SOFTWARE The review of related literature on OSS is comprised oftwo phases. First, this treatise contains a definition of OSS, a brief history of open source development, software licenses, open source community structure, and several structural/procedural issues related to OSS development, such as development mode, acceptance process, communities of practice, and private-collective innovation. Second, prior studies of OSS developers are reviewed, including studies on developers' motivation for participation in OSS projects, their effort as well as performance in relation to OSS development, and certain research models developed to empirically test the relationship among motivation and developers' behavior. Definition of Open Source Software Significantly different from traditional closed source software whose source code is protected as a commercial secret by the software vendor, OSS means that the source code written in any programming language should be freely available to any person who requests it. According to "Open Source Definition 1 ,"the major implication of open source software primarily lies in the condition that anyone should be able to unrestrictedly access computer 1 http://www.opensource org, Open Source Definition 13

PAGE 29

program code, modify it and redistribute the modifications at zero or nominal cost. This allows any individual to make software improvements and fixes. OSS has also been called "free software", and both terms have been used interchangeably Actually, "free" in "free software" mainly emphasizes the "free(dom) to modify the program's source code", rather than "at no cost" (Appelbe, 2003). Just as used by the Free Software Foundation, "free" essentially refers to the freedom to access, modify, and redistribute the software's source code. Therefore, this potential ambiguity in the meaning of the term "free software" has led to "open source software" becoming the more common use term On the other hand, open source software is also distinguished from freeware and shareware since freeware is distributed for free but users do not get access to the source code and have no right to modify or extend the software and shareware is often offered for free for a trial period (Schmidt and Schnitzer, 2003). In summary, open source software has three main features (I) the source code must be available to anyone, (2) anyone may modify the source code or derive new works from it, and (3) there are no restrictions on redistributing the software; that is, anyone may give away or charge the software for a nominal price, and companies are free to package and sell products containing OSS, and consumers who buy these products are free to redistribute that OSS at no cost if they desire so. 14

PAGE 30

Evolution of Open Source Development The open source phenomena have lasted for a half century. Table 2.1 demonstrates the major events in the history of open source. Table 2.1 Important events in open source history Era Event 1950sto Software source code is distributed without restrictions Early 80s in some scientific computing communities (e.g ACM's algorithm section) and user groups among computer vendors such as IBM, DEC. The very first version of Unix and C Language were introduced, and their source code was freely available. BSD (Berkeley Software Distribution) Unix was freely distributed. Sendmail was originally developed by Eric Allman, a computer science student at UC-Berkeley, in the late 70s. Early 1980s Ricard Stallman, a programmer at the MIT artificial to Late 80s intelligence laboratory, established the Free Software Foundation in 1983, and published GNU2 (a recursive acronym for GNU's Not Unix) manifesto fonning GPL (General Public License) to seek for making software freely available Perl (Practical Extraction and Reporting Language) was created in 1987 by Larry Wall, a programmer at Burroughs (a computer mainframe manufacturer now part ofUnisys) The source code of Minix, a version ofUnix for PC and Mac, was released in 1987 by its developer, Andrew Tanebaum. 2 GNU is a computer operating system composed entirely of free software. Its name is a recursive acronym for GNU's Not Unix, which was chosen because its design is Unix-like, but differs from Unix by being free software and by not containing any Unix code. GNU was founded by Richard Stallman and was the original focus of the Free Software Foundation (FSF). 15

PAGE 31

Table 2.1 (Con't.) Era Event Early 1990s Linux, a Unix variant based on Minix, was created in to Late 90s 1991 by Linus Torvalds, a Finnish computer science student at University of Helsinki. FreeBSD 1.0 was released in 1993 Debian Linux, a new Linux distribution, was created by Ian Murdock in 1993. The development of Apache began in 1994 Marc Ewing founded Red Hat, the leading Linux distributor, in 1994. Star Office, the equivalent of Microsoft's Office suite, was originally developed in mid 90s by SUN Microsystems, and its introduction represented that OSS development started to cover desktop applications. Debian Free Software Guidelines developed in 1995, were adopted in early 1997 by a number of OSS developers and subsequently became known as 'Open Source Definition'. Netscape in 1998 gave away the source code of its Communicator 5.0, which eventually became an open source project. 2000 to December 11 2000, IBM Chairman and CEO Lou Today Gerstner announced that the company would invest $1 billion in Linux development for promoting the diffusion of its e-business platforms. Open Office, which evolved from Star Office is now 99.9% Word and PowerPoint compatible. Source: Barahona et al. 1999; Hars and Ou 2002; Lerner and Tirole 2002; Appe1be 2003 Open source software's origins can be dated from the earliest days of software development in the 1950s. Then, software source code developed by some scientific programming communities such as academic/research institutions was frequently and mutually given away to other similar organizations without any restrictions. As a result, this kind of benevolent behavior let those software beneficiaries also contribute their own 16

PAGE 32

improvements back to the whole community just as scientists publish or give away their research results so that other scientists can build on their results to further continual innovation (Barahona et al, 1999; Comerford, 1999; Dempsey et al, 2002). According to Appelbe (2003), there is an implicit "honor code" in science, which says that all research results and related developments should be shared with the scientific community scientists view colleagues who are not willing to share their results and work openly with deep suspicion. Thus as early as the late 1950s, scientific software was being freely distributed. During the 1960s and 1970s, when proprietary software flourished all over the world, two major software products emerged as the most important milestone of OSS development, i.e. Unix and C Language. Unix, an operating system capable of running on multiple platforms, and C, a programming language originally developed at AT &T's Bell Labs in the early 1970s were available for distribution in academia and research labs at zero or nominal cost with source code, i.e. open source. From then on, many of the sites where Unix and C Language were installed made further innovations, which were in turn shared with others (Lerner and Tirole, 2002). For instance, by using this open source, the University of California at Berkeley enhanced Unix in the late 1970s, and consequently produced the BSD (Berkeley Software Distribution) of Unix whose source code were freely distributed and widely used in academia. 17

PAGE 33

Since the early 1980s, more and more computer vendors had adapted BSD Unix to run on their machines, which led to many of the early commercial versions of Unix that were based on BSD Due to the growth of Unix's commercial success, AT&T then addressed a legal problem concerning its intellectual property rights related to Unix. This issue evolved as a controversial topic at that moment. Those maintaining BSD Unix had to redevelop the operating system from scratch to prevent from violating AT &T's intellectual property right. Nonetheless, a free version ofBSD Unix was eventually released in the late 1980s and it is still freely available in source code form"open source". Another important event in the history of open source occurred in the mid 1980s. Richard Stallman, a programmer from MIT's Artificial Intelligence Lab, founded the Free Software Foundation (FSF) in 1984 since he was extremely dissatisfied with the trend for more and more software to be proprietary and strongly believed in the importance of sharing source code without limitation. 3 Through the efforts of Stallman and a group of volunteers around the world, these open source software contributors not only initialized the GNU (a recursive acronym for Gnu's Not Unix) project to implement a totally free version of the Unix operating system but built a variety of free versions of software utilities and development tools as a preliminary to developing the operating system itself. According to Raymond (200 I), for more than a decade after its founding, FSF would largely 3 http : / / www.gnu org 18

PAGE 34

define the public ideology of the hacker culture, and Stallman himselfwould be the only credible claimant to leadership of the tribe. Apparently, Stallman's endeavor has been regarded as an open source software crusade because he has constantly advocated the importance of freely available source code. Throughout the 1990s, with the prevalent usage of the Internet and the World Wide Web, the development of open source software stepped into a new era. Especially, the presence of Linux, a new Unix variant with an amalgam of 'Linus' and 'Unix', brought open source activity unprecedented prosperity. In 1991, Linus Torvalds, a Finnish computer science student at Helsinki University, started creating an open source Unix-type kernel from scratch, and afterward encouraged others' contributions in a sequence ofpostings to online bulletin boards for free implementation ofthe software. Because of previous BSD Unix's legal problems, many people interested in running Unix on their personal computers turned to Linux. Consequently, the community dramatically spread, and Linux soon became one of the most popular and leading open source software projects. The key to Linux 's success has two important factors: (1) Torvalds "opened it up" for community improvement and development, and the software community jumped on the bandwagon; (2) the rise of the Internet provided software developers with a very expedient tool to communicate, collaborate, and distribute software (Appelbe, 2003). In addition to operating systems, open source development of application software was also launched in the mid 1990s. One of the most well-known 19

PAGE 35

products was "Star Office", the comparable of Microsoft's (MS) Office suite developed by SUN Microsystems in the mid 1990s. Although Star Office's functionalities could not compete with those of MS Office and live up to its original expectation oftaking up some of the office desktop market, its derivativeOpen Office about 99 9% Word and PowerPoint compatible had been able to catch up with the majority ofMS Office's features by the early 2000s.4 With respect to databases, the performance ofmySQL, an open source database management system, is close to that of commercial closed source products such as Oracle and DB2. In recent times, even on the enterprise resource planning side, an open source product-Compiere ERP/CRM has also emerged as a strong competitor to proprietary software such as Peoplesoft. Actually, more and more open source products have taken dominant positions in their relevant areas For instance, Apache takes the leading position in the web server market (Hann et al. 2002; Lerner and Tirole 2002; Appelbe 2003). Sendmail, an open source e-mail transfer agent, handles about three quarters of all Internet e-mail traffic around the world (Lerner and Tirole 2002) Due to the growth in popularity of open source, a number of websites such as SourceForge.net, Collab.net and bkbits.net are devoted to making it easier to find, download, install and maintain a wide variety of open source software (Appelbe 2003). 4 http: //www.openoffice.org 20

PAGE 36

To date, there has been almost no arena of software for which there is not an open source alternative to proprietary software (Appelbe 2003). Moreover, many governments worldwide have also begun legislative intervention to foster the open source movement and to spread the usage of open source software in public administration and educational institutions (Schmidt and Schnitzer 2002; Appelbe 2003). Types of Open Source Software Licenses Although the source code of OSS is freely available, open source programs are commonly distributed under very precise licensing agreements generally named "Copyleft" licensing scheme. Copyleft licensing was created for ideological purpose by Richard Stallman and the Free Software Foundation (GNU project, 2000b) The main feature ofcopyleft is that once a source code is licensed by the initial developer, the subsequent code based on the original must also be licensed in the same way. Copyleft is devised for linking the programmer and his/her contribution permanently together. According to Mustonen (2003), copyleft creates an environment where talented programmers have an incentive to signal their abilities via the open source community. Currently, there are two organizations Open Source Initiative (OSI) and Free Software Foundation (FSF), in charge of maintaining a series of copy left licenses. Among these licenses, Berkeley Software Distribution (BSD)-style license, FSF's GNU General Public License (GPL) and the Lesser GPL (LGPL) are the most important and popular. On the 21

PAGE 37

other hand, the emergence of the commercially oriented licenses such as Mozilla Public License (MPL) and IBM Public License (IPL) has also been quite influential. Table 2.2 shows the various characteristics of these open source licenses. The followings are their detailed descriptions. Table 2.2 Characteristics of open source licenses Licenses Does it impact derived Can it be works? closed? BSD (Berkeley Software No Yes Distribution) License GPL (General Public License) Yes No LGPL (GNU Lesser GPL) No No MPL (Mozilla Public License) No Yes IPL (IBM Public License) No Yes Source: Gacek and Arief2004 The BSD License The Berkeley Software Distribution (BSD) Licenses represent a family of permissive free software licenses. The original was used for the Berkeley Software Distribution, a Unix-like operating system for which the license is named. The original owners ofBSD were the Regents of the University of California because BSD was first written at the University of California, Berkeley. The first version of the license was revised, and the resulting licenses are more properly called modified BSD licenses. Permissive licenses, sometimes with important differences pertaining to license compatibility, are referred to as "BSD-style licenses". Several BSD-like licenses, including the 22

PAGE 38

New BSD license, have been vetted by the OSI as meeting their definition of open source. The BSD Licenses impose very few limitations on revising or redistributing source code compared to other free software licenses such as the GNU GPL or even the default restrictions provided by copyright, putting it relatively closer to the public domain. The licenses permit a software vendor to base its products on open source software, add its own custom enhancements, and then sell the final products as commercial software where the source code needs not to be open source. The MIT license and Apache Software License approved by the OSI are two well-known examples of such licenses. The BSD licenses have been referred to as copy-center, as a comparison to standard copyright and copyleft free software. Many commercial software companies find this type of license very attractive since they can incorporate the software that falls under this license into their products and then sell the packages without turning them into open source. General Public License The GNU General Public License (GPL) is one of the most common and stringent copyleft licenses. Its main idea is to keep software free in perpetuity. All source code that incorporates GPL source code legally becomes open source code itself. According to Stallman (1996), anyone who lawfully obtains a program covered by the GPL automatically inherits the full 23

PAGE 39

rights to use, copy, modify, or distribute the source code in any manner desired but subject only to the terms of GPL itself. Based upon the GPL, any modification to the software must be subsequently distributed under the terms of the license itself. In other words, a person who enhances software in terms of the GPL must contribute the enhancements back to the open source project/community. Thus, many people regard the GPL as being anti-commercial due to its rigid restriction. However, the GPL does not prohibit charging a positive price for the source code covered by the license as long as the code is not turned into closed source. Lesser General Public License Based upon the nature of the GNU GPL, all source code consisting of it must be released under the GPL. A modified license, the Lesser GPL (LGPL, formerly the GNU Library General Public License) emerged as GPL had been proved impractical. The LGPL, published by the FSF, was developed as a compromise between the GPL and the BSD. The LGPL places "copyleft" restrictions on the code itself but does not apply these limitations to other software that merely links with the code 5 Although the LGPL imposes fewer restrictions than the GPL, there are some other restraints. Essentially, it must be possible for the software to be linked with a newer version of the LGPL-covered code. The most common 5 http :// www gnu org/licenses/lgpl.html 24

PAGE 40

method for doing so is to use a suitable shared library mechanism for linking. Alternatively, a statically linked library is allowed if either source code or linkable object files are provided. Overall, the major difference between the LGPL and the GPL lies in the fact that LGPL is intended for use with software libraries and it may be linked with proprietary code (Fitzgerald 2006). Mozilla Public License The Mozilla Public License (MPL) was developed by Mitchell Baker when she worked as a lawyer at Netscape Communications Corporation and at the Mozilla Foundation 6 It is often regarded as being the middle-ground between the strictness of the GNU GPL and the tolerance of the BSD License. It is not used anywhere near as widely as either the GPL or the BSD license, but its flexibility and thoughtful drafting mean that it is becoming more popular. The MPL is commercially oriented. The license is regarded as a weak "copy left", although source code copied or changed under the MPL must stay under the MPL. Unlike strong copy left licenses, the code under the MPL may be combined in a program with proprietary files which would otherwise be derivative works ofthe MPL code.7 For example Netscape 6 and later releases were proprietary versions of the Mozilla Application Suite. Additionally, the MPL is the license for the Mozilla's families, such as Application Suite, Mozilla Firefox, Mozilla Thunderbird and other Mozilla 6 http: // www.opensource orgllicenses/mozilla l l.php 7 http :// www.mozilla.org/MPL/MPL-I.I.html 25

PAGE 41

software. The MPL has been adapted by others as a license for their software, most notably Sun Microsystems, as the Common Development and Distribution License for OpenSolaris, the open source version of the Solaris 10 operating system. IBM Public License The IBM Public License (IPL) is a free and open source software license used by IBM It is approved by the OSI and the FSF. Unlike GNU GPL, IPL places the liability on the publisher or distributor of the licensed program. This is to facilitate commercial use of programs, without placing the contributor in risk of liability. Its proponents say it has a clearer definition of who's responsible for the program than the GPL has. The IPL is incompatible with the GPL because it contains restrictions which are not in the GPL. According to the FSF, it requires certain patent licenses be given that the GPL does not require. 8 It differs from the GPL in the handling of patents, as IPL terminates the license upon patent disputes This license has also been criticized because of provisions in section 4 which require commercial distributors of code covered by this license to indemnify all upstream originators for legal costs relating to lawsuits brought about by users of the software. It has been argued that this exposes small distributors (e.g., Linux distributions that happen to sell CDs) to unbounded legal costs, possibly arising from vexatious claims 8 http: //www gnu.org/philosophy/ license-list.html 26

PAGE 42

Open Source Communities Since OSS grants contributors and users the right to freely read and modify the source code, their activities associated with OSS development become communities of practice. Generally, the community members interact with each other for knowledge sharing and collaboration in pursuit of solutions to certain software problems. Thus, OSS development is impossible to be successful if there is no accompanied community that provides the stage for contributors and users to collaborate with each other. The open source community of practice is described as follows. Open Source Participants Based on project type, an open source participant can take on different roles in different projects. According to Nakakoji et al. (2002), members of an open source community assume certain roles by themselves according to their personal interest in the project rather than being assigned by someone else Table 2.3 shows the roles on which OSS participants might take in their project/community. 27

PAGE 43

Table 2.3 Major participants in open source communities Role Duty Project Leader The person who initiates the project and is responsible for the vision and overall direction of the project. Core Member The persons who are responsible for guiding and coordinating the development of an open source project. Usually, core members are those who have been involved with the project for a relative long time and have made significant contributions to the development and evolution of the system. In particular, if a single project manager quits, the core members would form a council to take the responsibility of directing the project development. Active The persons who regularly contribute new features and Developer fix bugs. Typically, they are one of the major development forces of open source projects. Peripheral The persons who occasionally contribute new Developer functionality or features to the existing system. Their contribution is irregular, and the period of involvement is short and sporadic. Bug Fixer The persons who fix bugs either discovered by these participants or reported by other members in the community Bug Reporter The persons who discover and report bugs. They do not fix the bugs themselves, and may not read source code either. They assume the same role as testers of the traditional software development model. Reader The persons who are active users of the system. They not only use the system, but also try to understand how the system works via reading the source code. Passive User The persons who only use the system in the same way as most users of commercially closed source software Source : Nakakoji et al. 2002; Ye and Kishida 2003 Structure of Open Source Communities The structure of most OSS development teams is demonstrated as having a hierarchical or onion-like form (Moon and Sproull2000; Nakakoji et al. 28

PAGE 44

2002) According to Nakakoji et al (2002), at the core is a project manager surrounded by a few core members and then more active developers and peripheral developers respectively. Surrounding the developers are bug fixers followed by bug reporters. Readers are located between bug reporters and passive users. At the outermost layer of the onion are passive users. In essence, the core of an open source community is kept as small as possible since it would be difficult to control if the core group were large. In addition the roles closer to the core have a greater scale of influence. A project manager's action would have more impact than that of a core member which in turn has a more significant influence than an active developer and the like. Users have the least influence, but they still can make certain contributions to the community since they use the latest product releases and usually contribute bug reports or feature requests (but not code). Actually, non-developers may also contribute by writing documentation or translating the system. Generally, every open source community has its own features based upon the project type and the number of its members. The structure differs in the percentage that each role in the community changes (Ye and Kishida 2003). Mode of Software Development OSS development has several particular features driven by its loosely organized community and numerous highly distributed developers. The most significant characteristic is modularization. The success of an open source 29

PAGE 45

project heavily depends upon the ability to break the project into distinct components (Lerner and Tirole 2002). Modular design's benefits are a direct result that it supports increased understanding during design and concurrent allocation of work during implementation (Gacek and Arief2004). A well-defined interface and modularized source code are a prerequisite for effective remote collaboration, since OSS development is globally distributed (Bollinger et al. 1999). The second feature of OSS development lies in its architecture. A computing system's software architecture represents its structure and comprises its components, the components' externally visible properties and their relationships (Bass et al. 1998). OSS system's architecture might be available or not. An unintentionally unavailable software architecture suggests that the structure exists in some people's minds only (Gacek and Arief2004). Documentation and testing are two important tasks for software development. Good documentation enables users to accurately use the software and understand how to modify the software. Testing can give users and/or developers confidence that the software would work as expected. However, these two aspects are often ignored or changed broadly during OSS development. OSS developers tend to be more interested in coding than documenting or testing (Raymond 2001; Gacek and Arief2004). This is probably because OSS tries to replace the formal testing process with "many 30

PAGE 46

eyeballs" approach to fixing bugs and also developers usually consider that the addition of comments to the source code is sufficient documentation. Numerous open source projects implement some types of configuration management in order to facilitate concurrent software development and controlled evolution They do this by utilizing Concurrent Versions System (CVS) or other ad hoc web-based supports. Therefore, this has also become one ofOSS important development's features. Process of Submissions Acceptance In the settings of OSS development, source code or patches are usually submitted from a variety of sources. According to Gacek and Arief (2004), open source projects often post the areas for which they want to receive submissions, and often they might receive multiple concurrent submissions addressing the exact same problem. Therefore, projects usually have specified processes for accepting submissions and handling multiple concurrent submissions. Generally, an open source project accepts participants' contributes through three steps: (I) choosing the work area; (2) making decision for approving the submissions; (3) disseminating the approved submission (Gacek and Arief 2004 ). First, code or patches are written by the developer and are submitted to the specific work area which addresses the project's corresponding problems. Second, the decision to accept the submissions is based on four aspects : (1) quality goals; (2) acceptance criteria; (3) decision 31

PAGE 47

group's cognitive abilities; and ( 4) the project's social structure Third, a project might passively disseminate the accepted contributions through newsgroups or comments in the code itself, or actively distribute the information via email or mailing lists. With respect to the decision making process, quality goals and acceptance criteria extensively differ from project to project and from one community to another. Likewise, the decision making for approving the submissions varies among projects and potentially within projects since it highly relies on the decision group's cognitive abilities to identify appropriate solutions. In addition, the social structure of an open source project/community also plays an important role in decision making. In a hierarchical structure, different groups of members evaluate different submissions, or some members might have greater power to make the final decision. If the structure is monolithic, the decision group might use consensus or majority vote to decide whether the submissions can be accepted Communities of Practice OSS communities possess the essential qualities that Wenger and Snyder (2000) identified as "communities of practice," which are formed by people, mostly practitioners, who are "informally bound together by shared expertise and passion for a joint enterprise." The term "communities of practice", first coined by Lave and Wenger ( 1991 ), refers to a theory that builds on learning as social participation (Wenger 1998). According to Wenger ( 1998), 32

PAGE 48

communities of practice are formed by people who engage in a process of collective learning in a shared domain of human endeavor: a tribe learning to survive, a band of artists seeking new forms of expression, a group of engineers working on similar problems, a clique of pupils defining their identity in the school, a network of surgeons exploring novel techniques, or a gathering of first-time managers helping each other cope. Wenger and Synder (2000) suggests that communities of practice can drive strategy, generate new lines of business, solve problems, promote the spread of best practices, develop people's professional skills, and help companies recruit and retain talent, since people in communities of practice share their experiences and knowledge in free-flowing, creative ways that foster new approaches to problems. Moreover, communities of practice can function in virtual and distributed environments, since communities generally are concerned with motivation, are self-generating, are self-selecting, are not necessarily co-located, and have a common set of interests motivated to a pattern of work not directed to them (Hildreth et a!. 2000). Therefore, a number of OSS researchers have introduced the idea of communities of practice to study OSS communities Communities of practice emerge in companies that thrive on knowledge (Wenger and Snyder 2000). The concept of communities of practice has been regarded as setting the stage for effective knowledge sharing. Thus, it is appealing to use the concept to describe and to analyze knowledge communities whose purpose is knowledge creation and communication. As 33

PAGE 49

knowledge communities are located in the virtual such as open source communities, the investigation of communities of practice may help further understand OSS communities Private-Collective Innovation The practices of OSS developers and communities present a novel and successful alternative to conventional innovation models. According to von Hippel and von Krogh (2003), OSS development is an exemplar of a compound "private-collective" model of innovation that contains elements of both the private investment and the collective action models. In the private-collective model, participants in OSS projects use their own resources to privately invest in creating novel software code. These innovators could then claim proprietary rights over their code, but instead they choose to freely reveal it as a public good. Clearly, the net result of this behavior appears to offer society the best of both worlds-new knowledge is created by private funding and then offered freely to all (von Hippel and von Krogh 2003) In the "private investment" model, innovation is supported by private investment and that private returns can be appropriated from such investments (Demsetz 1967). Innovating contributors should only freely reveal their innovations when the costs of free revealing are less than the benefits. It has been argued that such conditions can hold in many fields, including OSS (Harhoff et al. 2003). In contrast, the collective action model applies to the provision of public goods ranging from provision of a public bridge to 34

PAGE 50

provision ofOSS (von Hippe! and von Krogh 2003) It requires that contributors of innovation relinquish control ofknowledge they have developed for a project and make it a public good by unconditionally supplying it to a "common pool." This requirement enables collective action projects to avoid the social loss problem associated with the restricted access to knowledge of the private investment model. OSS projects display with respect to the assumptions about incentives embedded in the private investment and the collective action models of innovation is that contributions to OSS development are not pure public goods since they have significant private elements even after the contribution has been freely revealed (von Hippe! and von Krogh 2003). More specifically, the private-collective model of innovation occupies the middle ground between private investment and collective action models. In the setting of OSS development, contributors must engage in problem solving to create novel code When they freely reveal this code to the project, it becomes a public good However, the problem-solving process and effort used to produce the code have other important outputs as well, such as learning and enjoyment, and a sense of ownership and control over their work product. In other words, programmers contribute freely to the provision of a public good because they garner private benefits from doing so. Therefore, the emerging phenomenon of OSS development obviously does not undermine either the private investment or the collective action models of innovation. However, it does make clear the utility of combining 35

PAGE 51

both into a "private-collective" incentive model that can more effectively address the interlinked private and collective incentive structures observable in that field, and perhaps elsewhere as well (von Hippel and von Krogh 2003) Prior Research on OSS Developers Open source development involves lots of developers spread all over the world. Generally, developers from different geographic areas work asynchronously on a large variety of open source projects. Moreover, the majority of developers are volunteers who donate their time, energy, and knowledge to the projects without any financial compensation. According to Mockus et al. (2000), there are potentially very large numbers of people engaged in open source software development, and often hundreds or thousands of volunteers contributing to the exact same project. Therefore, this raises an intriguing and widely debated issue: what motivates open source developers to contribute their effort? This topic has been explored from a variety of theoretical viewpoints such as social psychology, organizational behavior, economics, and so on. Related studies and empirical findings on OSS developers' motivation and effort are described as follows. Studies on Motivation The OSS literature suggests that OSS developers show both intrinsic and extrinsic motivations (Bitzer et al. 2004). OSS developers' motivations are considered to be related in complex ways rather than independent, and 36

PAGE 52

different motivations have an impact on developers' participation in different ways (Roberts et al. 2006). The list of all possible motivators is too extensive to include all motivators in one study Therefore this research focuses on those motivators that appear to have the strongest theoretical arguments. The various theoretical viewpoints reviewed here are summarized in Table 2.4 Table 2.4 Major motivations of OSS developers Motivation Descriptions Type Helping Aiding others to increase the welfare of Intrinsic others Enjoyment Enjoying writing programs Intrinsic Peer Recognition Receiving recognition from others Intrinsic within the community Enhancing Human Accumulating skills and knowledge via Extrinsic Capital adult learning_ Career Advancement Demonstrating capacities and Extrinsic skillfulness to signal potential employers Satisfying Personal Acquiring software patches or Extrinsic Needs components for personal use or solving iob related technical problems Intrinsic Motivation Intrinsic motivation is defined by Deci (1975) as the doing of an activity for its inherent satisfactions rather than for some separable consequence Based upon Ryan and Deci (2000), an individual when intrinsically motivated is moved to act for the fun or challenge entailed rather than because of external prods, pressures, or rewards. Deci's and Ryan's theory has been 37

PAGE 53

regularly applied to explain why open source developers voluntarily contribute their efforts. Empirical analyses show that intrinsic motivations are essential in determining the participation of individual programmers in open source projects (David eta!. 2003) Generally, from the perspective of intrinsic incentives, open source developers rank helping others, enjoyment to program, and peer recognition among the most important reasons to take part in open source movement (Bates eta!. 2002; Ghosh eta!. 2002; Hars and Ou 2002; Lakhani and von Hippe! 2003). Helping behavior. Open source community is a "gift culture" that is motivated by altruism and reciprocity (Raymond 2001; Hars and Ou 2002; Bonaccorsi and Rossi 2003). "Altruism, understood as doing something for someone else at some cost to oneself, is contrasted with selfishness ... altruism is a natural part of human nature that it is not just found in a few rare people that it has evolutionary value and is exhibited in some manner by everyone" (Ozinga 1999). People often behave altruistically and pro-socially, contributing to the welfare of others without apparent compensation (Schwartz 1970). Altruism exists when people derive intrinsic enjoyment from helping others without expecting anything in return (Krebs 1975; Smith 1981 ). OSS developers' voluntary behavior has been regarded as an altruism-driven incentive since they are willing to proactively contribute their work with no reimbursement. Giving away goods and services for free allows individuals to make and maintain social links and entails the duty to reciprocate (Mauss 1959). 38

PAGE 54

Reciprocity has been highlighted as a benefit for individuals to engage in social exchange (Blau 1964). Wasko and Faraj (2000) indicate that people who share knowledge in online communities believe in reciprocity since they feel that sharing knowledge to help others not only feels good but that everyone is better off when knowledge is shared. Moreover, people who regularly help others in virtual communities seem to receive help more quickly when they ask for it (Rheingold 2000). This suggests that such a personal benefit is more likely to accrue to individuals who actively participate and help others in open source communities (von Hippel and von Krogh 2003). Rheingold ( 1994) points out that information-sharing based on a hunger for intellectual companionship is initially found most commonly among professionals who work more or less on their own, e.g journalists, freelance artists and designers, programmers, etc. Open source community is organized by numerous devoted programmers who are both producers and users. The creation and maintenance of social relations in such a community are not regulated by the possession or exchange of money or commodities but the economy of gift exchange (Bergquist and Ljungberg 2001 ). Literature suggests that OSS developers voluntarily make contributions because they would like to lend a hand to other members and simultaneously give something back to those who have ever assisted them In a gift culture's setting, given the abundance of resources, social status is determined not by what you have but what you give away (Raymond 2001 ). This phenomenon is most likely to happen when the exchange is not in favor of well-known 39

PAGE 55

individuals but of a community ofunknown subjects (Bonaccorsi and Rossi 2003), such is the case of the OSS community. Enjoyment. According to Csikszentmihalyi (1975), enjoyment-based motivation refers to a satisfying flow of activity from which the enjoyment can be derived, such as playing sports or collecting stamps for pleasure. Csikszentmihalyi provided the concept of a state of flow to explicate the phenomenon. Based upon the theory, enjoyment is maximized in a state of flow. Flow states arise when a person's skill matches the challenge of a task. Moreover, there is an optimal zone of activity in which flow is maximized. Enjoyable activities are found to provide feelings of creative discovery, a challenge overcome and a difficulty resolved (Csikszentmihalyi, 1975). Following the theory of flow states, Lakhani and Wolf(2003) suggest that OSS developers may be seeking flow states by selecting projects that match their skill levels with task difficulty, a choice that may not be available in their regular jobs, when they participate in an open source project. Additionally, the literature in electronic networks of practice has shown that individuals are motivated intrinsically to contribute knowledge to others because engaging in intellectual pursuits and solving problems is challenging or fun (Wasko and Faraj 2000). Lerner and Tirole (2002) point out "the programmer compares how enjoyable the mission set by his/her employer and the open source alternative are. A "cool" open source project may be more fun than a routine task." Thus, enjoyment has been regarded as one of the most important incentives that motivate OSS developers to participate in open source projects. 40

PAGE 56

Peer Recognition. Peer recognition has been viewed as one type of signaling incentive. Lerner and Tirole (2002) suggest that programmers participating in open source projects based on this motivation may also want to signal their abilities to different subjects although they may shun future monetary rewards. According to Hars and Ou (2002), peer recognition stems from the desire for fame and esteem which is associated with future returns Following Maslow (1970), "All people in our society (with a few pathological exceptions) have a need or desire for ... reputation or prestige, status, fame and glory, dominance, recognition, attention, importance, dignity, or appreciation." Hence, it is reasonable that programmers get involved in OSS development simply because they expect the acquisition of recognition from their counterparts. Essentially, OSS development is quite similar to the publication of scientific research. In scientific societies, sharing results enables researchers both to improve their results through feedback from other members of the scientific community and to gain recognition and hence prestige for their work (Banaccorsi and Rossi, 2003) In the settings of OSS development, while releasing the source code to the communities, OSS developers can not only obtain other members' fast feedback helping them refine their work but also gain recognition from those who are using their contributions and probably bring to themselves future employers' notice. As a result, successful contributors of open source projects will have potential opportunities to advance their career because ofthe effect of recognition. 41

PAGE 57

Extrinsic Motivation Ryan and Deci (2000) define extrinsic motivation as a construct that pertains whenever an activity is done in order to attain some separable outcome. They describe that extrinsic motivation refers to behavior where the reason for doing it is something other than interest in the activity itself (Deci and Ryan, 1985). Intriguingly, economics appears to have become the most dominant theory used to explain open source developers' extrinsic motivation. According to economic literature, a programmer is expected to participate in an open source project only if a benefit can be derived from engaging in the activity (Lerner and Tirole 2002). The major benefits from participation mainly consist of enhancing human capital advancing career opportunities, and satisfying personal needs for software. Enhancing human capital. A number of economists and OSS researchers regard enhancing human capital as one of the important incentives that motivates open source participants to engage in open source projects. From the standpoint of labor economics, human capital as a determinant of productivity refers to personal skills, capabilities, and knowledge (Mankiw 2004). Human capital is the economist's term for the accumulation of investments in people such as education and on-the-job training "The most important type of human capital is education .. education represents an expenditure of resources at one point in time to raise productivity in the future .. Human capital includes the skills accumulated in ... on-the-job training for adults in the labor force" (Mankiw 2004, p. 412, p. 542) 42

PAGE 58

OSS development provides an excellent learning environment for participants who seek opportunities to enhance their human capital. Developers who start an open source project are generally master programmers and their works are the products of fine craftsmanship, providing examples of excellent programming for less skilled developers (Ye and Kishida 2003). In addition the freedom to choose tasks not only enable OSS developers to select the learning experiences that meet their needs and interests, but also enable those entry-level programmers like college students to participate in realistic projects at a very early stage (Hars and Ou 2002) Thus, those open source participants can develop skills, such as programming that are going to be useful when they enter the labor market (Bonaccorsi and Rossi 2003). Career advancement Participation in OSS development has the potential to advance one's career in two ways. The first is the opportunity for OSS participants to demonstrate (advertise) their capabilities and skillfulness The larger the contribution of an individual to open source projects, the more likely it is that the commercial software vendors will recognize the value of the individual (Hars and Ou, 2002) OSS development is an opportunity for participants to signal potential employers. According to economic signaling theory, signaling refers to actions taken by an informed party to reveal private information to an uninformed party (Mankiw 2004). Signaling within an open source context would be particularly valuable if employers have difficulty assessing potential 43

PAGE 59

employees. Instead, employers could use success at developing OSS as an indicator of ability. Second, participants may also use OSS involvement to acquire access to venture capital or to launch a new business endeavor. The most well known examples are the founders of Linux, Sun, and Red Hat (Lerner and Tirole 2002). However, the majority of OSS developers are more likely to use open source communities for job advancement or acquiring shares in commercial open source-based companies than accessing venture capital. Satisfying personal needs. Numerous open source projects were initiated by the need that the founders had for specialized software or software patches since they had met certain technical problems in their day-to-day work (Hars and Ou 2002; Lerner and Tirole 2002). For example, one of the origins of the free software movement was Stallman's inability to improve a printer program because its owner, Xerox, refused to release the source code (Lerner and Tirole 2002). 9 During the development of the Apache web server a similar problem was encountered when the original developers struggled to find patches for revising the server software. Definitely, many OSS projects take shape because the people promoting them have looked in vain for a program to perform a particular function. They arise to satisfy a work-related demand for which there is no corresponding supply, in short to "fill an unfilled market" (Bonaccorsi and Rossi 2003). Thus, participants in OSS efforts may directly benefit from the 9 Richard Stallman founded the Free Software Foundation in 1984. Then, he was a programmer from MIT's Artificial Intelligence Lab 44

PAGE 60

software and software improvements they develop because they have a personal or job related use for them (von Hippel 1988; Raymond 200 I). In addition, a major advantage of OSS is the freedom to incorporate the software into other products or to produce add-on products that enhance the features ofOSS. In many cases, the ability to modify OSS to satisfy exact personal or business requirements or knowing how to incorporate OSS into other products is valuable. Developing OSS is one way of acquiring the knowledge needed to produce new products that utilize OSS. Effort and Performance For the sake of the sustainability of an OSS project, an understanding of OSS developers' effort and performance is an important issue on OSS development. According to Christen et al. (2006), effort is an input to work while performance is an output from this effort. In the setting of OSS development, there has been several research concerning developers' effort and performance Effort Effort defined as the number of hours per week spent on an open source project has been used in a number of previous OSS studies (Hars and Ou 2002; Hertel et al. 2003; Lakhani and Wolf2003) and provides an appropriate proxy for developers' contribution and interest in OSS projects (Lakhani and Wolf 2003) With respect to the success of an OSS project, it is imperative to 45

PAGE 61

understand whether all types of motivations affect OSS developers' effort equally or in the same way. Some motivations may strongly affect the effort whereas others may not be as salient. Academic theorizing on individual motivations for participating in OSS projects has posited that external motivational factors in the form of extrinsic benefits (e.g., career advancement) are the main drivers of effort (Lakhani and Wolf 2003). In contrast, Hars and Ou (2002) argued that open source programmers with intrinsic motivations will spend more time and effort in open source projects since intrinsically motivations have been suggested to be associated with most effortful behaviors (Sheldon and Elliot 1997). Therefore, there is no consensus in the OSS literature as to which motivation has the most dominant impact on individual OSS contributors' effort. Hars and Ou (2002) performed a simple correlation analysis, finding a weak correlation between effort and altruism and strong correlations between effort and both building human capital and signaling employers However, Lakhani and Wolf (2003) argued that the most significant and pervasive determinant of effort dedicated to OSS projects was enjoyment-related intrinsic motivation in the form of a sense of creativity, followed by extrinsic motivation in form of payment. Moreover, they revealed that paid contributors would spend more hours per week than volunteer contributors on OSS projects, which implied financial subsidy is substantial to OSS development. Contrary to experimental findings on the negative impact of extrinsic rewards on intrinsic motivations (Deci et al. 1999), Lankhani 's and 46

PAGE 62

Wolf's results suggested that both being paid and feeling creative on OSS projects do not have a significant negative impact on developers' effort. Performance One of the tenets of OS S projects is the frequent provision of feedback to contributors (Moon and Sproull 2000) In a number of well known open source communities like Apache Software Foundation, continued contribution is rewarded with a change in performance ranking (Roberts et al. 2006). They periodically evaluate the actual contributions of their members and assign each member a certain performance ranking. These rankings are based on merit and reflect the contributors' level of participation in the OSS community. Advancement within the meritocracy recognizes individuals' commitment and contributions to the OSS projects (Fielding 1999) In essence, performance is broadly defined as an aggregate construct of effort, skill, and outcomes (Walker et al. 1977; Behrman and Perreault 1984; Lusch and Serpkenci 1990 ). It refers to an evaluation of the results of an individual's behavior usually by someone other than the individual, and involves determining how well or poorly an individual has accomplished a task (Kanfer 1990). Experimental research in psychology also suggests that performance is the outcome of an evaluation by others of an individual's behavior and this behavior is often manifested by individuals' task output (Mitchell and Daniels 2003). 47

PAGE 63

Theoretically, motivation has an important influence on performance, and it has been deemed as an antecedent of performance. According to Roberts et al. (2006), motivations vary across individuals and combine with individuals' knowledge, skills, and abilities to produce task relevant behaviors. These behaviors contribute to individual performance. Studies have shown that motivation focuses attention on particular task elements and produces effort as people work harder when they are motivated (Roberts et al. 2006). In the OSS context, Roberts et al. (2006) argued that motivations influence OSS developers' participation in OSS projects as exemplified by the level of their contributions to the source code. Over time, contributors' participation is evaluated by the OSS community. This performance evaluation may lead to a rise in a contributor's rank within the community, which can, in turn, act as feedback to influence the future motivations of contributors. Example Models A number of researchers have explored motivational processes and performance of persons who voluntarily engage in OSS development. A rather intuitive approach to understand the motives of OSS developers is built on expectancy-valence theory that explains motivational processes in individual developers' mindset. Another method relevant for the understanding of developers' motivation and performance stems from 48

PAGE 64

systematic approaches provided by a theoretical model from intrinsic and extrinsic motivation The two relevant studies are described as follows. Wu, Gerlach, and Young model (2007) + Motivation on ns Helping ns Motivation on + .-----------Enhancing I Human Caoital + Motivation on ns Career ---------I Advancement ns Motivation on + I ------Satisfying I Personal Needs I I I I I + + oss Satisfaction Continuance Intent io n Figure 2.1 Open Source Participants' Continuation Model Wu, Gerlach, and Young (2007) conducted a field survey of 148 OSS participants to identify salient determinants of open source participants' intention to continue making OSS contributions. Toward this goal, the concepts from expectancy-valence theory were adapted to build a theoretical model of OSS participants' continuance intention shown in Figure 2.1. 49

PAGE 65

According to the findings satisfaction with participating in open source projects was the strongest influence on OSS participants' intentions to participate in future OSS projects, followed by their motivation on enhancing human capital and job related problem-solving. These associations demonstrated that satisfactory experience and recognition of benefits from solving job related technical problems and increasing human capital influence OSS participants' continuance behavior. The results asserted that satisfaction is the key to continuance, participants may subconsciously pursue instrumental behavior because of the requirement for specific software functionalities, and the theory that learning by doing is of essential importance to the success of OSS projects and the sustainability of OSS communities. Roberts Hann and Slaughter model (2006) To understand what motivates OSS developers to participate in OSS development, Roberts et al. (2006) conducted a study on the developers of the Apache projects by revealing how different motivations were interrelated, how these motivations influenced participation leading to performance, and how past performance influenced subsequent motivations. Drawing on theories of intrinsic and extrinsic motivation, Roberts et al. developed a theoretical model relating the motivations, participation, and performance of OSS developers. The model is shown in Figure 2.2. They evaluated the model using survey and archival data collected from a longitudinal field study of software developers in the Apache projects. 50

PAGE 66

,---I I I I I ns 1 I I Past Performance n s Intrinsic + Performance Education Experience Figure 2.2 OSS Developers' Motivation and Performance Model The theoretical framework for Roberts et al. 's study leveraged the general model of motivation and performance in organizational and social psychology. In their model, extrinsic motivations include being paid to contribute, enhancing status/career opportunities, and use value (i.e., the desire to fix a bug or solve a problem of immediate relevance to the contributor), while intrinsic motivations refer to needs for competence, control, and autonomy. Performance is related to promotion to a higher rank within the Apache hierarchy, which is awarded after one or more cycles of contribution followed by a positive peer review and is consequently an acknowledgement of an individual's substantive contributions to the project. 51

PAGE 67

This study argued that motivations varied across individuals and combined with individuals' knowledge, skills and abilities to produce task-relevant behaviors and influence the level of OSS developers' contribution to the source code. In addition, motivation has an important influence on performance because it focuses attention on particular task elements and produces effort as people work harder when they are motivated. This performance evaluation may lead to an increase in a contributor's rank within the community. Roberts et al. 's results revealed several important findings. First, OSS developers' motivations were not independent but rather were related in complex ways. Being paid to contribute to Apache projects was positively related to developers' status motivations but negatively related to their use value motivations. Moreover, Roberts et al. found no evidence of diminished intrinsic motivation in the presence of extrinsic motivations; rather, status motivations enhance intrinsic motivations. Second, different motivations differentially influenced participation. Developers' paid participation and status motivations led to above average contribution levels, but use value motivations led to below average contribution levels, and intrinsic motivations did not significantly impact average contribution levels. Third, developers' contribution levels positively influenced their performance rankings. Finally, past performance rankings enhanced developers' subsequent status motivations. 52

PAGE 68

Summary OSS development affords a particularly rich context in which to examine individual motivations. Exploratory empirical research has been done pertaining to the motivations of OSS developers. For years, OSS researchers have identified a variety of motivators and suggest that developers' motivation may contain various sources and, simultaneously, different motivations could differentially relate to participation in open source projects OSS developers' motivations have been classified as intrinsic or extrinsic. Intrinsic motivators include helping, enjoyment, and peer recognition, while extrinsic motivators include enhancing human capital, career advancement, and personal requirements for the software. The empirical findings from extant research on developers help us understand the relationships between motivations and developers' behavior as well as related issues associated with OSS development, such as effort and performance. For example, participation in open source projects may be based on multiple incentives; developers may vary their effort according to different motivators; motivations influence developers' performance; and developers' satisfaction level influences their continuance intention. Although extant research has examined OSS developers' motivations, behavior, and the relationship of motivation to effort and performance, no research heretofore has investigated OSS contributors' motivations as a whole. In other words, these studies do not take into consideration the influence of each individual motivator. Moreover, many arguments have not been 53

PAGE 69

empirically and rigorously measured with ground theories such as expectancy-valence theory, human resource theory, and volunteerism. Based on different motivators, what are developers' expectations and attitudes in association with OSS development after participating in open source projects? Further, what impact will an individual motivator have on developers' effort? Clearly, additional research is needed to address these issues. 54

PAGE 70

CHAPTER 3 REVIEW LITERATURE OF VOLUNTEERISM Research on OSS developers suggests that open source communities cannot exist or prosper without the contributions of highly motivated developers who voluntarily donate their time and effort to the community. Since the majority of OSS developers are strictly volunteers rather than traditional paid employees, it is not possible to solely rely on employment relationships or employment contracts to manage these persons (Roberts eta!. 2006). Therefore, an understanding of volunteers' motivations and their behavior is considered an important first step in exploring how to motivate OSS developers and how to direct, sustain, and influence their volunteer behavior. Volunteering is routinely identified as work that is unpaid and taken on freely (Mutchler eta!. 2003). For many individuals, volunteering may provide rewards that paid employment does not, e.g., self-fulfillment through challenging and interesting activities (Miller 1985) However, seeking more time and more work from a volunteer is usually not a welcome request (Galindo-Kuhn and Guzley 2001) Because conventional rewards are non-existent in volunteer work, one must tum to intrinsic rewards instead (Gidron 1984). Therefore, paid work is infused into volunteer activities when there is an economic and social necessity for it (Gidron 1983). 55

PAGE 71

In the context of OSS development, a number of open source communities allow companies benefiting from OSS to involve their paid employees in open source projects. Alternately, communities may financially support their own volunteer developers to devote their full attention to the tasks assigned by the projects such as Debian and GNOME (GNU Network Object Model Environment). The combination of volunteer workers with paid employees may have a certain impact on volunteers' mindset and behavior which, in turn, may influence the progression of their activities. For example, the influence of paid employees whose job is GNOME development has been instrumental in the success of the project, while Debian has been suffering serious delay. In this chapter, volunteer motivation, work attitudes, and other related issues such as turnover and effort are elaborated. Volunteerism One important manifestation of human helpfulness is volunteerism (Clary et al. 1998). Volunteerism is conceptually defined as an ongoing activity aimed at improving the well-being of others (Omoto and Snyder 1995). It is an activity that can be found all over the world (Curtis et al. 1992). Every year, millions of people share significant amount of their time and talents as volunteers to helping others, and the number is on the rise (Clary et al. 1998; Skoglund 2006). 56

PAGE 72

The study of volunteerism is one element in the large effort to study helping behavior (Mowen and Sujan 2005). Specially, it has long captivated researchers of social behavior that an individual would make significant personal sacrifices for another person particularly when that person is a stranger (e.g Latane and Darley 1970; Staub 1978; Piliavin et al. 1981; Eisenberg 1986; Batson 1991; Schroeder et al. 1995). Existing volunteerism literature speaks largely to varieties of helping somewhat different from volunteerism, focusing on helping in contexts where a potential helper is faced with an unexpected need for help, calling for an immediate decision to act and an opportunity to provide one and only one relatively brief act of help (Benson et al. 1980; Bar-Tal 1984; Piliavin and Charng 1990). Clary et al. ( 1998) suggested that factors revealed by research on the helping that occurs in these kinds of contexts, sometimes referred to as spontaneous helping, may have the important impact on volunteerism. Volunteerism seems to be a rather different kind of helping, a kind that is prototypic of planned helping, which often calls for considerably more planning, sorting out of priorities, and matching of personal capabilities and interests with type of intervention (Benson et al. 1980; Clary et al. 1998). Therefore, based upon Clary et al. ( 1998), volunteers should have three characteristics(1) they often actively seek out opportunities to help others, (2) they may deliberate for considerable amounts of time about whether to volunteer, the extent of their involvement, and the degree to which particular activities fit with their own personal needs, (3) they may make a commitment 57

PAGE 73

to an ongoing helping relationship that may extend over a considerable period of time and that may entail considerable personal costs of time, energy, and opportunity. Motivation to Volunteer It is essential to understand volunteers' initial motivation to volunteer and their expectations of the volunteering experience, since recruitment and selection of volunteers is a costly process and it can be critical to the future use of volunteers in an agency (Cnaan and Goldberg-Glen 1991; Ralston and Rhoden 2005) Although Gidron (1984) argued that the motives that initially influence people to volunteer may differ from those that influence their decision to continue to volunteer, it is extremely important to understand the initial motivation of those who remain as volunteers for a long run (Cnaan and Goldberg-Glen 1991 ). Moreover, based upon the definition and characteristics of volunteerism, Clary et al. ( 1998) indicated that it may be productive to adopt a motivational perspective and to inquire about the motivations that may dispose individuals to seek out volunteer opportunities, to commit themselves to voluntary helping, and to sustain their involvement in volunteerism over extended periods of time. Reasons for Volunteering Motivation is a difficult concept in general, because it is subconsciously constructed and it is neither systematic nor consistent (Cnaan and 58

PAGE 74

Goldberg-Glen 1991 ). Gillespie and King ( 1985) suggested that we will never know the answer if we do not ask people what motivates them to volunteer. According to Tschirhart et al. (200 1 ) volunteering can reflect multiple motivations (including altruistic instrumental, social, self-esteem, and other goals) that can be collapsed into one or more factors or dimensions. Cnaan and Goldberg-Glen ( 1991) summarized twenty eight significant motives for volunteering in human services on the basis of motivation to volunteer literature. Those motivations are listed in Table 3.1 Table 3.1 Motives for volunteering No. Reason I It is God's expectation that people will help each other. 2 I adhere to the agency's specific goals 3 If I did not volunteer there would be no one to carry out this volunteer work. 4 I did not have anything else to do with my time. 5 I was lonely. 6 I have more free time (i.e kids have left home, retired, widowed, divorced). 7 I wanted to gain some practical experience toward paid employment (or new career). 8 I wanted to broaden my horizons 9 Being involved with this agency is considered prestigious. 10 Volunteering for others makes me feel better about myself. 11 Volunteering in this agency provides challenging activities. 12 Most people in my community volunteer 13 Helping people in need improves my attitude regarding my own life situation. 14 Volunteering creates a better society. 15 My employer-school expect their employees-students to provide volunteer community service. 16 Volunteering is an opportunity to change social injustices. 17 Volunteering is an opportunity to develop relationships with others. 18 Volunteering is an opportunity to work with different age groups. 59

PAGE 75

Table 3.1 (Con't.) No. Reason 19 Volunteering is an opportunity to do something worthwhile. 20 Volunteering is an opportunity to return good fortune. 21 A relative of friend is/was a client of this agency. 22 I have past experience providing similar service. 23 I am able to relate better to the patients/residents situation because of my own similar experience. 24 This volunteering gives me an opportunity to vary my weekly activities. 25 Previous contact with professionals in this agency. 26 Volunteering for this agency enables it to provide more care for less money. 27 It's a way to continue a family tradition of helping in need. 28 This is an excellent educational experience Source: Cnaan and Goldberg-Glen 1991 Functional Approach Based upon Clary et al. ( 1998), the fundamental concerns of motivational inquiry with understanding the processes that move people to voluntary action are precisely the concerns engaged by the questions: "Why do people volunteer?" and "What sustains voluntary helping?" A useful approach to answering these questions begins with the premise that volunteering serves different functions for different people, i.e functional analysis (Houle et al. 2005). The functional approach is explicitly concerned with the reasons and the purposes, the plans and the goals, that underlie and generate psychological phenomena, that is, the personal and social functions being served by an individual's thoughts, feelings, and actions (Snyder 1993). 60

PAGE 76

Houle et al. (2005) suggested that the functional approach may help to discover underlying motivations of volunteering since it implicates the importance of matching volunteer motivations to the benefits that volunteerism provides. According to a functional analysis of volunteerism, people engaging in similar acts may have different underlying motivations for doing so (Houle et al. 2005) Clary et al. (1998 ) adopted the strategy of functional analysis to analyze volunteer motivations. They catalogued six functions of volunteerism: values, understanding, social, career, protective, and enhancement. The six functions are described as follows. The values function refers to concerns for the welfare of others, and contributions to society (Clary et al. 1998) The function has been likened to altruism (Clary and Miller 1986), the value-expressive attitude function (Katz 1960), and the quality of expressiveness (Smith et al. 1956). An empirical evidence for the values function indicated that over 70% of the respondents endorsed to help others as a reason for volunteering (Anderson and Moore 1978) The understanding function served by volunteering involves the opportunity for volunteerism to permit new learning experiences and the chance to exercise knowledge, skills, and abilities that might otherwise go unpracticed (Clary et al. 1998). This function is related to Katz's (1960) knowledge function and Smith et al. 's ( 1956) object appraisal function. In support of this function, Gidron ( 1978) found that young volunteers (high 61

PAGE 77

school and college students) tended to view their volunteer work as a learning and self-development experience. A third function served by volunteering is the social function in which an individual volunteers due to strong normative or social pressure, or to get along with others in his or her reference group (Clary et al. 1998; Haule et al. 2005). Conceptually, this function is similar to Smith et al. 's ( 1956) social adjustive function and Francies ( 1983) need to respond to the expectations of others Piliavin et al. (1984) found evidence for the social function in their investigation of motives for donating blood. They found that some individuals donate blood due to external, social motives. A fourth function that may be served by volunteering is concerned with increasing one's job prospects and enhancing one's career that may be obtained from participation in volunteer work (Clary et al. 1998). For example, Beale (1984) suggested encouraging students to volunteer as the experiences may serve as steppingstones to employment. Jenner (1982) indicated that some Junior League volunteers perceived volunteering to be a means of preparing for a new career or of maintaining career-relevant skills. A fifth function that may be served by volunteering is the protective function in which an individual volunteer tries to reduce feelings of guilt about being more fortunate than others, or to escape from one's own problems (Clary et al. 1998). Thus function could be likened to Katz's (1960) ego-defensive function, Smith et al. s ( 1956) externalization function, and Francies' (1983) need to express feelings of social responsibility. Schwartz 62

PAGE 78

(1970) found support for the protective function in his study of volunteering to be a bone marrow donor. His results showed that individuals had a greater level of commitment to volunteer when the salience of personal responsibility for others was high. A sixth function is the enhancement function in which volunteerism serves to enhance an individual volunteer's self-esteem, self-confidence, and self-improvement (Clary et al. 1998; Houle et al. 2005). Results of related studies have found support for the esteem function. For example, volunteers working in mental hospitals showed an increase in self-acceptance as a consequence of their volunteer participation (Holzberg et Ia. 1964; King et al. 1970). Based upon the approach of functional analysis, volunteerism may serve more than one motive for an individual and different motivations may be served within a group of volunteers performing the same activity (Houle et al. 2005). Moreover, in the same individual, different motives may be primarily engaged by different volunteer activities. Paid v.s. non-Paid Volunteer work in the human services has many similarities to paid work (Gidron 1983). According to Galindo-Kuhn and Guzley (2001), both paid and unpaid workers interact with the organization and other people in the organization, and they also have certain expectations about what their participation will provide for them. Whether paid or unpaid, workers are 63

PAGE 79

provided with a job that they are expected to do to the best of their abilities. It involves a situation where there is ajob to be done, the job can utilize one's skills and creativity, one's efforts can bear fruit in the form of results or achievements, and one can be recognized for it (Gidron 1983). Nonetheless, volunteer work is somewhat different from paid work. The volunteerism literature (e.g., Gidron 1983, 1984; Cnaan and Goldberg-Glen 1991; Galindo-Kuhn and Guzley 2001) makes a distinction in motivations between volunteers and paid labor. The dissimilarities are described as follows: First, volunteer work is by definition an act of free will; individuals engage in it and discontinue it at will. In contrast, in paid work, the pay element represents for most people the necessity to work, the fact that one has to work in order to live. It is related to one's survival. This notion is well-grounded in most cultures. It means a different form of relationship to the workplace, a different form of compliance (Gidron 1983). Second, volunteer work per se is usually not related to an occupational career in the field in which one works (Gidron 1983). If a volunteer decides to develop a career in the field where he or she volunteers, the volunteer usually crosses the lines andjoins the paid work force (Gidron 1983). Third, volunteerism involves volition. Implicit in the very word volunteer is the root concept of volition (Ellis and Noyes 1990). There is volitional nature to unpaid work that goes above and beyond any economic or social necessity; volunteers choose to engage in unpaid work simply because 64

PAGE 80

that is how they choose to spend their leisure time. In contrast, paid work is subtly coercive in its origin. Fourth, volunteering and going to work represent vastly different psychological approaches to organizational participation (Pearce 1983). The primarily expressive orientation of volunteer work can best be described in terms of whether there is a goal of social responsibility (Ellis and Noyes 1990). People engage in volunteer work because they want to help others (Schram 1985; Cnaan and Goldberg-Glen 1991; McSweeney and Alexander 1996). This orientation can also be found among some paid workers, but paid work is predominantly instrumental in orientation (Galindo-Kuhn and Guzley 2001). The primary emphasis for most paid workers, even paid work in the public and nonprofit agencies, is towards responsibility for self and self-benefit (Mirvis and Hackett 1983; Pearce 1983; Vinokur-Kaplan et al. 1994). Fifth, the perceived value of reward that is obtained by volunteers is quite different from by paid workers. Volunteers attach a stronger sense of reward value to the incidental outcomes of the work experience such as friendships (Mirvis and Hackett 1983; Pearce 1983; Vinokur-Kaplan et al. 1994). The traditional work reward of material compensation is of greater value to the paid worker Satisfaction According to Gidron ( 1984 ), what originally motivated the volunteer to get involved may not be sufficient to sustain their involvement in the long 65

PAGE 81

term, unless they feel satisfied with their volunteering experience. Satisfaction and psychic benefits of various kinds are not just a by-product for those engaged in volunteer work, but are expected by volunteers (Smith 1981 ). Volunteer work is perceived as an exchange between the volunteer and his/her work situation (Sharp 1978; Kemper 1980), whereby time and effort are exchanged for satisfactions and psychic rewards to the individual (Qureshi et al. 1979). The descriptive and prescriptive literature on volunteering suggests that volunteers should be satisfied on their job in order to persevere with it (Naylor 1967). Satisfaction increases the likelihood of predicting retention-related outcomes, namely turnover potential (Galindo-Kuhn and Guzley 2001). Individuals who believed that they had satisfied their initial motivations for volunteering were more likely to intend to continue to volunteer than those who did not believe that they had received benefits that matched their initial motivations (Clary et al., 1998). People continue to volunteer because they enjoy what they are getting from the experience; they value the rewards they are receiving and want to maintain and expand them (Gidron 1983, 1984). What constitutes satisfaction from volunteer work? What is it in the job that volunteers find satisfying? Volunteer satisfaction is founded in a link between motivations, expectations and actual experience (Ralston and Rhoden 2005). However, little research exists on volunteer satisfaction, since for many years volunteer work was perceived as a purely altruistic act in which the volunteer is treated as a benevolent person who sacrifices something of 66

PAGE 82

oneself in order to give to others (Gidron 1983). This perception of volunteer work creates the bias against viewing volunteer work as a satisfying endeavor can cause volunteers to hesitate to discuss their real feeling about their work freely and openly (Gidron 1983). By conducting a multiple regression analysis, Gidron ( 1983) identified twelve factors that have the strongest effect on overall volunteer satisfaction. The twelve variables are work itself, task achievement, task convenience, stress factors, family (outside support), supervisor-instrumental (information), supervisor-expressive (emotional support), professionals (staff relationships), perceived social acceptance of volunteer work, client, recognition, and other volunteers. Gidron's (1983) work is reinforced by Cnnan and Goldberg-Glen (1991) and Galindo-Kuhn and Guzley (2001) who, in an attempt to predict volunteers' intention to remain, correlate wider motivational factors with more prescriptive volunteer expectation and satisfaction factors: quality, clarity and adequacy of communication and job-related information; feedback and recognition; suitability, convenience and autonomy of work assignment; importance of their role, benefits to others, fulfillment of intended contribution; quality of training, emotional and organizational support; social aspects and group integration (Galindo-Kuhn and Guzley 2001). Volunteers' satisfaction increased the more they became involved with the organization and took on additional responsibilities (Miles et al. 1998). Satisfying the volunteer and meeting their task-related expectations, particularly at the beginning of the volunteer-organization relationship, is 67

PAGE 83

more important than satisfying the needs of organization if volunteer attrition is to be avoided. Volunteers who enjoy their work, who think their work makes a difference and who believe their efforts are of value are more likely to persevere (Ralston and Rhoden 2005). Organizational Commitment Organizational commitment is defined as the identification with a particular organization, willingness to exert considerable effort on behalf of the organization, and the desire to maintain membership in the organization (Porter et al. 1974; Mowday et al. 1982; lgbaria et al. 1991; Igbaria and Greenhaus 1992; lgbaria and Guimaraes 1993; Thatcher et al. 2002-3). This affectively-oriented definition for the basis of commitment recognizes the importance of emotional attachment to the organization (Dailey 1986). Organizational research on volunteers has shown that volunteers are important human resources relative to the productivity of human services organizations (Johnson 1981; Gamm and Kassab 1983). If volunteers are well managed then they are likely to continue their contributions and influence others to become volunteers (Dailey 1986). In reality, research on volunteers' organizational commitment is sketchy. Related studies have most often been concerned with profiling the personality characteristics of the volunteer for various volunteer activities (Dailey, 1986). Pearce (1983) reported that volunteers tended to have higher socioeconomic status than non-volunteers and they were less likely to leave their 68

PAGE 84

organizations than paid employees working in comparable organizations. Werner ( 197 6) found individuals with moderate levels of independence and nonconformity, plus a willingness to reject authoritarian attitudes to be more likely to volunteer. These studies confirm findings in the organizational behavior literature since researchers routinely describe willingness to stay with the organization as a facet of the organizational commitment construct (Porter et al. 1974). Organizational commitment has been regarded as a useful index to predict future levels of volunteer involvement (Cuskelly 1995). It is formed by subsequent attitudes of job satisfaction (Bateman and Strasser 1984 ) According to Gamm and Kassab ( 1983 ), substantial proportions of decisions to volunteer and exert sustained effort on behalf of the organization are functions of higher order need strength. Thus, opportunities to satisfy these needs are prominent precursors for sustained job satisfaction among volunteer workers (Dailey 1986). Additionally, the more focused attitudes of job satisfaction and job involvement can be partially sustained through the positive impact of organizational commitment (Dailey 1986). Job satisfaction has a well documented history of being related to organizational commitment (Hall and Schneider 1972; Porter et al. 1974; Stevens et al. 1978). Most researchers cast job satisfaction as a predictor of organizational commitment (Hall and Schneider 1972; Stevens et al. 1978; Bateman and Strasser 1984). However, Steers and Mowday (1981) argued that commitment involves a wider individual perspective regarding the entire 69

PAGE 85

organization while job satisfaction centers on the person's reactions to specific facets of the job (e.g., co-workers, pay, promotions, the work itself and supervision). He indicated that job satisfaction is very volatile depending on rapid changes in the environment of work while organizational commitment develops more slowly but consistently over time Research has shown that both job satisfaction and organizational commitment have central roles in the job scope and turnover literature and they have been shown to be related to intention to stay and actual turnover behavior (Mowday et al. 1982 ; Bateman and Strasser 1984). Job involvement is an important predictor of organizational commitment (Dailey 1986). It is a work attitude which develops at the level of person-job interaction while organizational commitment forms as an affective attachment to the organization. Job involvement is conceptually defined by Rabinowitz et al. ( 1977) as the degree to which employees identify with their jobs, the degree to which employees actively participate in their jobs, and the degree to which they believe that their jobs are important determinants of their self-worth. Job involvement focuses on the affective relationship between the person and work, and thus it is understandable that job involvement is viewed as an individual-situation construct (Dailey 1986). Empirical findings in the literature of individual differences and job scope indicate that job involvement is a logical choice for predicting organizational commitment among volunteer workers (Dailey 1986). 70

PAGE 86

Retention and Turnover Volunteer retention, in its simplest form, is making volunteers feel good about their assignment and themselves (Lynch 2000) If the volunteer experience makes the volunteers feel good, then they will continue to want to volunteer (McCurley and Lynch 1996). Retention has an evil twin: turnover or the number of volunteers who leave the organization that have to be replaced (Skoglund 2006) Retention and turnover are important variables to volunteer program managers because they present serious problems for organizations that depend on volunteers to execute their mission statement (Skoglund 2006). Flowers and Hughes (1973) argued that the concepts of retention and turnover are inseparable since any effort to reduce turnover must take into account the reasons people stay and the reasons people leave. Retention Retention is the ability to keep volunteers involved (Jamison 2003). It is operationalized as active volunteers who have maintained the same level of service they started volunteering at the agency (Jamison 2003). This term refers to the number of volunteers who successfully complete their initial commitment to agencies, including those who renew and continue serving at the agency (Connors 1995). McCurley and Lynch (1996) suggested that studies of volunteer retention have determined that the first 6 months of volunteers' experience is critical toward their retention, as the greatest losses of volunteers occurs during this period 71

PAGE 87

According to Skoglund (2006), volunteers start their service in a honeymoon stage, which is composed of euphoria, self-congratulation, and eagerness to give of themselves. Upon gaining some experience, volunteers regress to a post-honeymoon blues phase. The idealism motivating their initial endeavor has now dissipated. This regression may occur when volunteers realize they are not able to accomplish what they had initially anticipated or when they realize that the organization does not represent the values or issues they originally thought. Such realization put a damper on an individual's initial motivating forces, and it merely becomes a matter of time before the volunteer steps out of the role of servitude. Volunteers also require more attention at anniversaries at the end of large projects, or at the completion of an agreed term of participation (McCurley and Lynch 1996). Based upon Skoglund (2006), several things contribute to a positive volunteer experience, which in turn increases retention and reduces the risk of turnover. First, retention of volunteers is accomplished through the development of feelings of importance and belonging to a particular agency (Murk and Stephan 1991). Starns and Wymer (2001) argued that volunteers will be satisfied with volunteering if they have the chance to develop friendship, share experiences, communicate with others, and develop support groups Volunteers will feel positive about their experience if they have an opportunity to cultivate their role identity (Grube and Piliavin 2000). Role identity is defined as one's concept ofthe self that corresponds to the social 72

PAGE 88

roles held by the individual (Grube and Piliavin 2000). A volunteer should perceive his or her role as important to the overall success of the organization When this occurs, self-esteem should be increased, thereby fostering commitment to the indi v idual's role identity as volunteer. In essence, general role identity as a volunteer will predict volunteer role performance (Skoglund 2006). Finally a positive volunteer experience can also be achieved when the volunteer experience new learning opportunities with the potential for personal or professional growth (Starns and Wymer 200 I). Starns and Wymer (200 1) indicated that one of the most frequent motivations for discontinuing volunteer service is inadequate training. Training for volunteers is overlooked when agencies view their volunteers as employed professionals who are just giving of their spare time In contrast that someone is a professional does not mean that person does not need training in volunteer work (Logue 2001 ). Training not only helps volunteers work better, but also helps to motivate them to donate time (Skoglund 2006) There is no money to pay volunteers "so we always have to work on motivation ... experience is the best way for volunteers to learn and stay motivated" (Logue 2001). Turnover The term turnover is operationalized as inactive volunteers who had a prior affiliation with their agencies but have decreased the number of hours 73

PAGE 89

per week (or month) or who have completely stopped volunteering at the agency (Jamison 2003). In deciding whether to quit a job, paid and volunteer employees undoubtedly differ in the extent to which they weigh various concerns (Miller et al. 1990). Volunteer turnover is to be expected in volunteer organizations and creates opportunities for organizational change, but high rates of turnover can hinder the capacity of organizations to deliver the quality or range of services and programs clients come to expect (Razzak 2001 ). Volunteers may leave agencies because they relocate to another area; because they become full-time employees, students, or parents; or because they experience health or transportation difficulties (Jamison 2003). Miller et al. ( 1990) suggested that turnover among volunteers, as with paid employees, can have some benefits. The departure of some volunteers may create opportunities for others to move into doing work that they prefer, and new volunteers who are brought in to fill vacancies may be better performers or have new ideas that result in positive changes for the organization (Dalton and Todor 1979; Staw 1980; McEvoy and Cascio 1987; Watts and White 1988). However, high turnover rates are critical when there is a need for volunteers with special skills or intensive training; volunteer responsibilities that require long-term commitments; the clients served by volunteers are disrupted by the absence of volunteers; and there is a shortage of qualified volunteers (Starns and Wymer 2001 ). Due to the nature of the organizations that tend to rely on volunteer work, turnover can have very adverse effects on service delivery and financial 74

PAGE 90

resources (Miller et al. 1990). For examples, the resignation of a volunteer tutor can seriously disrupt learning in a literacy program, and the resignation of a "Big Brother" or "Big Sister" can, contrary to program goals, contribute to a child feeling abandoned and unwanted. Moreover, many organizations rely on the services of volunteers in part because they are understaffed and strapped for funds; constantly having to recruit and train new volunteers can be a severe financial drain. Unlike most paid employees, volunteers are free to resign without first locating alternative employment (Miller et al. 1990). Therefore, those individuals who manage and oversee volunteer programs face a twofold challenge: orienting, training, and monitoring volunteers as well as retaining these volunteers (Forsyth 1999). An effective orientation and training of a program's volunteers will engage participants in a way that results in volunteers' willingness to participate in the agency's volunteer program for a significant period of time. However, if there is a breakdown in one of these functions, it is not long before an organization's volunteer program starts to flounder (Skoglund 2006). Determinants of Volunteers' Effort Volunteers provide organizations with many vital resources in the form of expertise, skills, knowledge, and labor. Volunteer work is regarded as an expenditure of time and effort on behalf of an organization or another individual (Mutchler et al. 2003). According to Dorsch et al. (2002), one 75

PAGE 91

factor that determines the value of these volunteer resources is the amount of effort volunteers expend while engaged in their voluntary activity. Dorsch et al. (2002) conducted a survey on several volunteer organizations (e.g., sports, culture, recreation), and found that four important factors were the most useful for predicting volunteer effort. These factors described as follows are role acceptance, role clarity, specific aspects of satisfaction, and role efficacy. Dorsch et al. (2002) suggested that the most important factor influencing volunteer effort was role acceptance. When people volunteer with an organization, they are either recruited to fill a particular role or are assigned to a role the organization needs to fill. In both instances, volunteers must accept the responsibilities of the role they occupy. Survey suggested that volunteers who accept the responsibilities associated with their role are more likely to expend greater effort to carry out their tasks than those who do not accept their responsibilities. Based upon Dorsch et al. (2002), role clarity was the single strongest contributor to role acceptance, and also was important in determining volunteer effort. Volunteers need to have a clear picture of what they will be doing, where they are doing it, and how it fits into the bigger picture. If volunteers are not certain about the specific nature of their roles, it is very difficult for them to accept those roles and work hard on the tasks associated with them. Role clarity was influenced by two most important factors: satisfaction with the organization's performance and role efficacy. A volunteer's level of satisfaction with how the organization is performing (e.g., 76

PAGE 92

meeting goals) has a direct impact on his or her perception of how clearly defined the volunteer role is in the organization Organizations with clearly defined goals tend to have clearly defined roles for their volunteers. Moreover, it tends to be easier to determine whether the organization is performing well if the organization has clearly defined goals, which results in greater satisfaction among volunteers Satisfaction with social service was a moderate predictor of role acceptance, and thus contributes indirectly to volunteer effort. Volunteers who are satisfied that they are being of service to others tend to put more effort into their activity. This factor must be seen in light of volunteer motives. Dorsch et al. (2002) suggested that two most important motives for volunteering are to help the community and other people. If the most important motivation of volunteers is to help and their perception is that they are doing so, then they will be much more accepting of the roles they have been assigned. This type of satisfaction also contributes to role clarity, when volunteers are satisfied that they are helping their community and other people they believe that their role has some meaning in the larger societal context. Volunteers' roles not only must be clearly stated and individuals perceive that they are contributing to the welfare of others but volunteers must feel confident that they have the ability to carry out their assigned tasks Role efficacy was not as strong a predictor of role acceptance as role clarity, but it was still a factor influencing volunteer effort. Moreover, role efficacy also 77

PAGE 93

contributes to role clarity, so it cannot be overlooked when considering how much effort volunteers will expend. Generally, organizations that want to achieve high levels of effort from their volunteers must ensure that volunteers accept their roles. This means ensuring that volunteers should have clearly defined roles, understand these roles, feel a sense of confidence in their ability to fulfill their roles, are satisfied with the extent to which they perceive themselves as helping their community and others, and are satisfied with the organization's overall perfonnance. Example Models Empirical studies on volunteers have been presented in the literature concerning the psychological mechanisms underlying individual differences in volunteers' behaviors and motives and the nature and extent of their participation in volunteer activities Several models have been empirically tested to validate volunteerism theories The followings are example models. Mowen and Sujan Model (2005) Mowen and Sujan (2005) conducted a three-phase study to explore the antecedents of volunteer behavior within a hierarchical model of motivation and personality shown in Figure 3 .I. The goal of their study was to identify individual difference variables predictive of a set of volunteer behaviors and the relation between a functional motive approach and a trait approach for 78

PAGE 94

predicting volunteer behavior. The functional motive approach seeks to identify the reasons and purpose that motivate a person to engage in behavior. In contrast, the trait approach was used to find enduring dispositions that influence behavior. The model was measured with a confirmatory factor analysis and the data were collected from the surveys of 600 members of a consumer panel run by a marketing company, 138 students at a Midwestern university and 630 members of an organization promoting volunteerism at the community and state levels. Compound Situational Functional Outcomes Traits Trait Motives Learning Altruism Help Others Ne ed for Activity Make Friends Volunteer Behaviors Self-enhance Need for Learning Career Present Time Focus Self-protection Figure 3 1 Trait plus Functional Motive Model 79

PAGE 95

In the first phase study, Mowen and Sujan developed measures of altruism and volunteer orientation and combined them with four compound traits (need for activity, need for learning, altruism, and present time focus) to predict a set of volunteer behaviors. In study 2, they examined the use of functional motives to predict volunteer behaviors. The functional motives developed and tested by Clary et al. (1998) included learning, help others, make friends, self-enhance, career, self-protection In study 3, Mowen and Sujan used an overarching model of individual difference regarding motivation and personality, developed by Mowen (2000), to examine the relationships of functional motives with compound and situation traits. Mowen and Sujan's study had several intriguing results. First, volunteer orientation and the functional motive of helping others were positively related to volunteer behaviors. Second, the situational trait of volunteer orientation was a significant predictor of volunteer behaviors and all functional motives. Third, altruism was a significant predictor of volunteer orientation. Fourth, the self-enhancement functional motive was negatively related to volunteer behaviors. Miller, Powell and Seltzer model (1990) By conducting a survey of 158 active volunteer workers at two hospitals, Miller et al. ( 1990) examined the causal sequencing of attitude, personal situations, and behavioral intentions as determinants of turnover among hospital volunteers. They intended to investigate whether turnover can be 80

PAGE 96

explained by factors analogous to those included in models of turnover among paid employees Therefore, Miller et al. developed a research model to study the influences of three general types of presumed determinants of turnover: attitudes of volunteers toward their volunteer assignments and organizations (i.e. job satisfaction organizational commitment satisfaction with the work itself), personal situation factors (convenience of schedule, experience) believed to be particularly relevant for volunteers, and intentions to quit the volunteer assignment. Additionally age was added to the model to be tested in order to examine its role as a determinant of turnover, since studies with paid employees have frequently found that age is inversely related to turnover. Miller et al. 's model depicted in Figure 3.2 was assessed by structural equation modeling. Convenience Experience Attitudinal Factors Age Intention to Figure 3 2 Hospital Volunteers Turnover Model 81

PAGE 97

Miller et al. 's analysis focused on understanding volunteer turnover at the individual level and examined psychological predictors of turnover. They argued that volunteers are a critical human resource for not-for-profit organization (NFPs) because they are frequently used in lieu of paid employees. As with paid employees, turnover among volunteers can be an important problem and major cost factor (Miller et al. 1990). Prevailing models of turnover among paid employees have proposed that attitudes and personal situations affect behavioral intentions to quit and that, in turn, behavioral intentions affect turnover (Miller et al. 1979; Mobley et al. 1978, 1979; Steers and Mowday 1981 ). Miller et al. ( 1990) argued that turnover among volunteers may result from the same general influences and decision-making processes as turnover among paid employees. According to Miller et al's findings, attitudinal factors and the volunteers' personal experience influenced turnover indirectly with intention to leave acting as a mediating factor. Of the variables originally designed as potential predictors, only convenience of schedule had a direct effect on turnover. In addition, the results revealed that age played an important role in determining turnover for volunteers, much as it did for paid employees. Miller et al. concluded that these findings were consistent with several studies of turnover among paid employees and of behavioral domains other than turnover. 82

PAGE 98

Dailey model (1986) I Personal Job Characteristics Involvement Organizational Commitment Job Job Characteristics Satisfaction I Figure 3.3 Volunteer Workers Organizational Commitment Model To measure a proposed model of organizational commitment for volunteer workers, Dailey (1986) administered a survey of 138 campaign workers who volunteered to work as solicitors of contributions and coordinators of fundraising activities for a major national charitable organization Dailey intended to examine the personality, job characteristics, and attitudinal antecedents of organizational commitment for volunteers since prior research did not shed light on the psychological processes which underlie organizational commitment among volunteer workers. The proposed model shown in Figure 3.3 was empirically tested with multiple regression analysis Dailey's studies including job scope constructs, individual differences, job satisfaction, and organizational commitment had been cross-sectional in design. Dailey did not conduct causal tests of hypothesis directionality, since 83

PAGE 99

the importance of the constructs in the literature of individual attitudes and behavior in the work place was doubtless. However, this assertion cannot be applied with equal confidence to the scant literature of commitment for volunteer workers. Therefore, given the central theme of the organizational commitment construct in the organizational behavior literature, it was reasonable to enhance understanding of the construct and its theoretical antecedents for volunteer (Dailey 1986). Dailey's research model shown in Figure 3.3 presented a synthesis of the literature for prior study's constructs and it was exploratory and descriptive in nature. According to Dailey, the importance of job characteristics and perceptual judgments about job design variables cannot be underestimated for volunteer workers. The findings indicated that the variables used to study work attitudes and organizational commitment for paid employees were associated with the outcomes for volunteers. Additionally, Dailey suggested that researchers studying volunteerism should incorporate task characteristics into their research and they need to recognize there is a wide range of behaviors and attitudes that materialize and drive volunteer activity well after the decisions to join and donate energy and time have been made. Empirical Findings After an extensive review of volunteerism, we can find researchers in a variety of disciplines have investigated volunteerism or related constructs. Most of these studies were based either on a conceptual analysis including 84

PAGE 100

summary reports of a specific project or on an empirical analysis of certain theories. The empirical findings are summarized in Table 3.2, and the highlighted arguments represent they will be brought forward to examining the motivation of OSS developers and their behavior in this study. Table 3.2 Research findings on volunteerism Topic Findin2s Volunteering can reflect multiple motivations (including altruistic, instrumental, social, self-esteem, and other goals) that can be collapsed into one or more factors or dimensions. Some theorists have considered two basic motivations for volunteering: to satisfy self-regarding or instrumental interests and to satisfy other-regarding or altruistic interests. Volunteers act not from a single motive or a category of motives but from a combination of motives that can be described overall as a rewarding experience. Motivations to volunteer include anticipated benefits of the activity for other individuals and groups as well as perceived benefits for the Motivation individual engaged in the activity. Volunteerism can be the key to gaining higher levels of specific knowledge, skills and experience, which may enhance existing ones and/or prove useful in the future. The pleasure of new-found knowledge, seeking out new challenges and the opportunity to develop new skills and explore career options were important motivators and satisfiers. The opportunity to mix with people with a similar interest and to share their views was also seen as an important motivator and satisfier. A positive volunteer experience can also be achieved when volunteer experiences new learning opportunities with the potential for personal or professional 2rowth. 85

PAGE 101

Table 3.2 (Con't.) Topic Findings An individual's desire to "give back to the community," for instance, may be fulfilled by a period of service after which the individual pursues other goals that have become higher in priority. Although a volunteer's decision to start volunteering may be driven by strong altruistic goals, his or her decision to continue volunteering is more likely to be based on an evaluation of costs and rewards than Motivation altruistic motives. Over time, external forces can undermine intrinsic motivation and consequently altruistic behaviors. People do differentiate tasks based on the volunteer motives they satisfy. Not all tasks are equal and that a task can be classified in terms of the motive(s) it does or does not satisfy. The motivations influencing the decision to become a volunteer are different from those that influence the decision to continue. Volunteer satisfaction is founded in a link between motivation, expectations and actually experiences. Volunteers' satisfaction increased the more they become involved with the organization and took on additional responsibilities. Volunteers who received benefits that matched their motivations were more satisfied with their volunteer experience. Satisfaction General job satisfaction was predicted with moderate strength by 3 job dimensions (task significance, skill variety, and task identity). Volunteers do not randomly select tasks but base a substantial portion of their task selection on motive satisfaction. Volunteers who felt that they were unimportant to the population they served were not satisfied Job-fit for both skills and convenience has been associated with satisfaction. 86

PAGE 102

Table 3.2 (Con't.) Topic Findin2s Organizational commitment was strongly predicted by Organizational general job satisfaction, autonomy, feedback. Commitment Organizational commitment predicted future levels of volunteer involvement. Discussions of how to retain volunteers generally have focused on ways to satisfy the volunteers' needs or enhance their commitment. Meeting volunteer expectation is the key to ensuring satisfaction and thus improving retention rates. Volunteers with matching benefits had greater intentions to continue volunteering either at the same of different location in both the immediate and long-term future. Volunteer who performed tasks that met their Retention and motives did report more positive volunteer Turnover experiences and intended to continue volunteering in the future. Intent to remain with the organization has been found to be an effective predictor of turnover. People continue to volunteer because they enjoy what they are getting from the experience; they value the rewards they are receiving and want to maintain and expand them. If volunteers' expectations are met initially and they stay with the organization, then they may wish to expand and diversify their duties and contribution. Some volunteer roles require little investment of time and effort, and volunteers in these roles may give little thought to why they continue to volunteer. Scheduling convenience had a direct effect on intention to leave and turnover. 87

PAGE 103

Table 3.2 (Con't.) Topic Findings Major constructs used to study employee behavior and work attitudes also function in a similar fashion relative to very important work attitudes for volunteers. Both satisfaction and organizational commitment have been shown to be related to intention to stay and actual turnover behavior. Because volunteers are free to leave their work without first finding alternative employment, their Overall satisfaction with volunteering and commitment to the organization frequently have been stressed as important determinants of whether they stay. Because of the leisure connotation, volunteers expected their experience to provide an opportunity for flexibility and to work at their own pace, with clear instructions and less stress than their normal work situation, and a change from their normal routine. In terms of the number of hours contributed to volunteering, the gap between full-time workers and nonworkers or part-time workers is limited, amounting to a few hours per week. Summary Volunteer work is not as altruistic an activity as we would expect. The volunteerism literature suggests that most volunteer activities are the result of multiple causations since volunteering can reflect multiple motivations including altruistic, instrumental, social, and other goals. Apparently, OSS developers share several characteristics with traditional volunteer workers because OSS research shows that OSS contributors' motivation simultaneously contain various sources (e.g., altruism, learning, enjoyment); 88

PAGE 104

as such, different motivations could differentially relate to participation. Therefore, an understanding of the motivation of volunteers can help investigate what motivates OSS contributors to voluntarily spend their time and effort in open source projects Job attitudes (satisfaction and organizational commitment) and turnover in the paid workplace have been popular and extensively studied topics for decades. As with paid employees, studies on volunteers' job attitudes are very important as well, especially since volunteers are frequently used in lieu of paid employees to fill auxiliary roles of the complementary or supplementary nature of professionals. Empirical tests on volunteerism have revealed that satisfaction is a link between motivation and both organizational commitment and turnover rate. Moreover, both job satisfaction and organizational commitment are related to volunteers' intention to stay or turnover. Since OSS contributors bear a number of features analogous to that of volunteers, the findings from volunteerism literature can help analyze OSS developers' motivations and other related issues such as work attitude, continuance intention, and effort. 89

PAGE 105

CHAPTER4 THEORETICAL PERSPECTIVES The theoretical arguments underpinning this study are based on the expectancy-valence theory (EVT), human resource theory, and volunteerism. EVT is applied to analyzing developers' motivations The human resource theory and volunteerism are used to explore developers' work attitudes (i.e., satisfaction and commitment), continuance intention, and related issues such as expertise and the availability of schedule. Moreover, the influential factors of motivation: project characteristics, years of OSS experience, and participation status, are analyzed by volunteerism. The following sections elaborate on these topics. Motivation Motivation is a psychological process resulting from the interaction between the individual and the environment (Latham and Pinder 2005). It has been defined as a psychological force inciting an individual to exert effort toward particular individual or organizational goals; motivation serves as a mechanism for satisfying an individual need (Robbins 1998; George and Jones 1999). According to Latham and Pinder (2005), motivation is a resource allocation process that determines how energy is used to satisfy needs and how time and energy are allocated to an array of tasks Motivation has been 90

PAGE 106

conceptualized using a number of theoretical frameworks intended to explain outcomes as diverse as work performance (Miller and Monge 1986), goal achievement (Locke 1968), turnover (Vroom 1964), and job satisfaction (Iaffaldano and Muchinsky 1985). In reviewing the literature several theories of motivation have been proposed including needs theories goal setting, equity theory expectancy-valence theory, and the jobs characteristics model. Within the broad area of motivation research, the notion of expectancy valence has played a central role in explaining human motivation in the workplace (Vroom 1964 ; Katzell and Thompson 1990 ; Locke and Latham 1990; Ambrose and Kulik 1999), which is used as the fundamental theory in this study It has been used to account for everything from occupational preferences and job satisfaction to volunteer attendance decisions (Harrison 1995). EVT, a behavioral version of the rational choice model in economic decision making, states that behavior is determined by conscious choices (Singer and Coffin 1996). Its underlying premise is that behavior is a function of an individual's expectation that a response will bring reinforcement together with the perceived value of the reinforcement (Rotter 1954). According to Lynd-Stevenson (1999), EVT is a cognitive-motivational theory that relates to an individual s level of motivation to the expectations and value/valence-be it positive or negative-held by the individual on reaching a particular goal. EVT is useful for evaluating the relative strength of intrinsic versus extrinsic valences in the motivational 91

PAGE 107

process (Mitchell 1980), a key requirement for investigating the motivations of OSS developers. EVT research can be largely found in the fields of social and applied psychology According to Vroom ( 1964 ), work motivation is determined by the expectancies and valences associated with items currently of importance in the individual's decision space. The level of motivation for specific behavior can be predicted by the combined strength of the individual's expectancies toward particular outcomes and the valences (i.e., subjective desirability) the individual places on those outcomes. The mathematical product of expectancy and valence can be used to predict the force on a person to perform a particular act as follows: F; =J;[I(EifvJl; i=n+J .. m, fi>O, in}=; j=l J where F; = the force to perform act i, Eu = the strength of the expectancy that outcome j will follow from act i, and Y.i = the valence of outcome j. The force to perform act i is a monotonically increasing functionj; of the sum of the products of the valences of all the outcomes associated with it and the related expectancies. The individual, presumably cognitively or acognitively, compares the courses of action open to him and performs the one with the greatest positive or smallest negative force (Behling and Starke 1973). Vroom's theory has been used for the prediction of occupational choice, valence of good performance, effort on the job, and job satisfaction (Behling 92

PAGE 108

and Starke 1973) Moreover, EVT is considered to be particularly useful in exploring the behaviors of individuals alone or in a group setting (Feather 1982). Mitchell ( 1980) suggests that the theory might work best in organizational contexts where the behavior is in the repertoire of the subject; where the behavior is under the control of the subject; and where the time lags between behaviors and outcomes, as well as between the assessment of predictors and the criterion, are short. The notions of expectancy and valence are each discussed below Expectancy and Valence Expectancy is conceptually defined by Vroom (1964) as a momentary belief on the part of an individual that acting in a particular way will actually lead to a given outcome. Vroom suggested that a person's behavior is influenced by both his preferences among uncertain outcomes and the degree to which he believes these outcomes are probable whenever he chooses between alternatives that involve these outcomes In addition, he indicated that expectancy is an action-outcome association. Likewise, Bandura (1977) argued that outcome expectancy is a person's estimate that a given behavior will lead to certain outcomes. While these researchers regard expectancy as the estimation of action-outcome behavior, a variant definition of expectancy is proposed in this study. Essentially, Vroom's theory focuses primarily on pre-action expectancy, but not post-action expectancy However, in the setting of open source, this 93

PAGE 109

study intends to examine developers' confirmation of the relationship between action and outcome in order to better understand the perceived value of the open source experience Post-action expectancy is especially important since future expectations are generally heavily influenced by past experience, and over time, they approach steady-state equilibrium as they become more realistic and entrenched in observed behaviors (Festinger 1957). Therefore, in this study, expectancy is interpreted as an individual's beliefthat a given action should lead to some certain outcome This interpretation is reasonable since research subjects in this study are not novices and their expectations are supposedly aligned with their past experiences. Vroom defines valence as the strength of a person's positive or negative affective orientation toward particular outcomes. Outcomes desired by an individual are considered positively valent while those that the individual does not desire are negatively valent. Feather (1995) proposes that valences can be conceived as the subjective attractiveness or aversiveness of specific objects and events within the immediate situation. Moreover, valences are linked to a specific context and to a present time frame According to Lewin (1936), valences refer to the goal properties of potential actions and outcomes as perceived by a person at a given moment. Based upon the Lewinian force theory, Vroom argued that behavior on the part of a person is assumed to be the result of a field of forces, each of which has direction and magnitude However, Vroom also recognized that this poses several restrictions. First, an outcome with high positive or 94

PAGE 110

negative valence will have no effect on the generation of a force unless there is some expectancy that the outcome will be attained by some act. Second, as the strength of an expectancy that an act will lead to an outcome increases, the effect of variations in the valence of the outcome or the force to perform the act will also increase. Third, if a person is indifferent to the outcome, neither the absolute value nor variation in the strength of expectancies of attaining it will have any effect on forces. Vroom's proposition resembles the concept in decision theory that people choose in a way that maximizes subjective expected utility (Vroom 1995). Following his idea, this study suggests that OSS developers' post-action expectancies and valence could be combined in a multiplicative way to predict their work attitudes, in other words, satisfaction with being involved in OSS projects, commitment to OSS development, and intention to continue participating in open source projects. Satisfaction The concept that satisfaction is a function of an initial standard (e.g., expectation) and some perceived discrepancy from the initial reference point is based on prior research in the fields of social and applied psychology (Spector 1956; Watts 1968; Locke 1969; ligen 1971; Weaver and Brickman 1974; Shrauger 1975; Andrews and Withey 1976; Campbell et al. 1976; Locker and Dunt 1978) and has been used to analyze a variety ofhuman conditions. In business settings, it has been used to study job satisfaction (Locke 1969), 95

PAGE 111

consumer satisfaction (Oliver 1980), information systems continuance (Bhattacherjee 2001; Bhattacherjee and Premkumar 2004 ), and satisfaction with application services (Susarla et al. 2003). Locke (1976) defines satisfaction in the context of job performance as a pleasurable or positive emotional state resulting from the appraisal of one's job. The study of job satisfaction is a major research area in industrial and organizational psychology (Goldstein 1984). A number of empirical studies have confirmed the important role of job satisfaction as a determinant of intention to turnover (Price 1977; Arnold and Feldman 1982; Bluedorn 1982; Michaels and Spector 1982; Baroudi 1985; Dougherty et al. 1985; Abelson 1987). Employees who are satisfied are more likely to stay on the job and engage in citizenship behaviors such as helping coworkers or customers and doing extra work while those who are dissatisfied are more likely to quit the job, be absent, file grievances, join unions, and go on strike (Organ 1987). Additionally, satisfaction can encourage continuance behavior (Anderson and Sullivan 1993; Oliver 1980, 1993; Wu et al. 2007). Studies suggest that job satisfaction has a direct effect on both organizational commitment and turnover intention (Mobley et al. 1979; Michaels and Spector 1982; Mowday et al. 1982; Boomsma 1985; Dougherty et al. 1985; Dailey 1986; Ward 1988; Mak and Socke12001). However, Williams and Hazer ( 1986) argued that job satisfaction predicts organizational commitment but has a mediated effect on turnover intention. Intriguingly, research on information technology (IT) workers' turnover found that job 96

PAGE 112

satisfaction had a direct effect on turnover while it also had an indirect effect through organizational commitment on turnover (lgbaria and Greenhaus 1992; Igbaria and Guimaraes 1993; Thatcher et al. 2002-3). According to Mobley ( 1982), actually quitting a job is mediated by fonning an affinnative intention to leave, although staying or quitting the job seems to be most consistently related to the degree of job satisfaction. According to Mobley and Locke (1970), the degree of satisfaction is a joint function of the degree of fulfillment of the value and its importance to the person. Thus, in the setting of OSS development, developers' satisfaction level should depend on weighing their post-participation expectancies against their perception of valence for their perfonnance (i.e., their emotional reaction to their own motivations). A satisfactory participation experience including pleasure and/or tangible benefits would strengthen developers' commitment and continuance intention. Open Source Commitment Commitment represents a duty or obligation to engage in future action and arises from frequent interaction (Coleman 1990). Salancik and Pfeffer (1978) indicated that commitment to a course of action may detennine subsequent attitudes Several experimental and field studies have confinned that commitment initiates a rationalizing process through which individuals make sense of their current situation by developing attitudes that are consistent with their commitment (Kiesler 1971; Salancik 1977) 97

PAGE 113

Organizational commitment is defined as the identification with a particular organization, willingness to exert considerable effort on behalf of the organization, and the desire to maintain membership therein (Porter et a!. 1974; Mowday eta!. 1982; lgbaria eta!. 1991; Igbaria and Greenhaus 1992; Igbaria and Guimaraes 1993; Thatcher eta!. 2002-3). It measures the relative strength of an individual's identification or involvement with an organization (Mowday eta!. 1982). Organizational commitment has been regarded as a strong predictor of turnover (Griffeth eta!. 2000). Theory suggests that organizational commitment negatively affects turnover intention (Porter et a!. I 976; Guimaraes and Igbaria 1982; Williams and Hazer 1986; Igbaria and Greenhaus 1992; Igbaria and Guimaraes 1993; Griffeth eta!. 2000; Thatcher eta!. 2002-3). In other words, employees who are highly committed to their employer or organization are less likely to tum over than their less committed counterparts (Thatcher eta!. 2002-3). OSS developers' commitment to OSS development is derived from the concept of organizational commitment. Therefore, developers' commitment is defined as (I) the acceptance of the goals and values of the open source community, (2) willingness to exert effort on behalf of the open source community, and (3) a desire to continue working for OSS development. Continuance Intention Unlike most previous research developed for predicting turnover or discontinuance intentions, this study attempts to measure OSS developers' 98

PAGE 114

intentions to continually engage in future open source projects, since it takes into account three perspectives: (1) the setting ofOSS development is extremely different from traditional working environments, because OSS development is conducted in virtual communities and the majority of developers are voluntary for contributing their work without any financial compensation; (2) understanding developers' continuance intention will be more helpful to estimate the long-term sustainability of OSS development; and (3) researchers argue that the direct and most often attributable effect of job satisfaction lies in recruitment and retention for an organization (Kramer and Schmal en berg 1991; Cavanagh 1992; Snarr and Krochalk 1996; Cum bey and Alexander 1998). In the setting of OSS development, the process by which developers form a continuance intention to engage in future projects can be described by EVT. First, there are the expectations of receiving a specific benefit from participating in OSS projects (i.e expectancies) and the subjective desirability of the benefit (i.e., valence). The benefit might be in the form of enjoyment from writing programs, recognition from other developers, a better paying job, enhanced knowledge and skills, or the obtainment of specialized software. Second, there is the decision (e.g behavior) to participate in OSS development with the expectation of receiving those benefits. Third, individuals are generally influenced by their experience which can potentially lead to the perceptions of whether they have obtained their desired benefits (i.e confirmation of expectancies). Fourth, based upon the extent of 99

PAGE 115

expectation and post-participation perception, the OSS developer experiences a level of satisfaction and simultaneously develops a certain degree of commitment to open source development. Satisfaction and commitment will influence a developer's intention to continue/discontinue participating in OSS projects. Availability According to Miller et al. ( 1990), discussions of volunteer turnover have probably emphasized attitudinal factors because volunteers, unlike paid employees, can quit without first securing other positions In an attempt to predict volunteers' intention to remain at a job, Cnaan and Goldberg-Glen ( 1991) and GalindoKuhn and Guzley (200 1) found that the convenience of the work schedule was a significant determinant when they correlated several motivational factors to a number of prescriptive volunteer expectation and satisfaction factors. Ralston and Rhoden (2005) revealed that many volunteers claimed that they had limited spare time to dedicate to volunteer activities because of existing personal and professional commitments Likewise, Rotolo and Wilson (2006) indicated that more flexible work schedules make it easier for volunteers to commit to volunteer positions. In this study, the convenience of a work schedule is renamed availability and it represents the convenience ofOSS developers' schedules for participating in OSS projects. 100

PAGE 116

Expertise For the purposes of this study, expertise is defined as the extent to which an open source project requires a variety of different skills, abilities, and talents from OSS developers, as adapted from Hackman and Oldham's (1975) skill variety. According to Hackman and Oldham'sjob characteristics model, skill variety should only lead to positive outcomes to the extent that this increase results in a corresponding increase in experienced meaningfulness of the work (Fox and Feldman 1988). If an increase in variety does not result in increased feelings of meaningfulness of cause, it is reasonable to hypothesize that this would result in a negative or negligible change in work attitudes. Literature on expectancy theory indicates that perceptions of skills and abilities influence expectancy perceptions (Katzell and Thompson 1990; Rasch and Tosi 1992). Individuals choose tasks and put forth effort on the basis of their expectancies. Additionally, people avoid activities that they believe exceed their abilities, while choosing those activities they feel capable ofhandling (Bandura 1982). Olson and others (1996) suggested that individuals persist longer and put more effort on tasks in which they expect to succeed. Therefore, persons who believe that their skill and ability set is adequate for achieving success with a new venture are motivated to exert the necessary effort (Douglas and Shepherd 2000; Shaver et al. 2001 ). After all, individuals who expect to perform well do so (Oettingen 2000). 101

PAGE 117

The Influential Factors of Motivation An additional purpose ofthis study is to explore the impact of project characteristics (commercial viability), years ofOSS experience, and participation status (volunteer or paid) on motivation among OSS developers. Because extrinsic monetary reward is lacking in the context of OSS development, motivational processes underlying developers' behavior are likely to be at variance with those in traditional paid work organizations. Thus, this study suggests that these factors may have direct effects on their motivation. Project Characteristics According to volunteerism theory, volunteers do not randomly select tasks, but base a substantial portion of their task selection on motive satisfaction. Individuals may have more positive volunteer experiences when allowed to choose volunteer tasks that will meet their motives. Volunteer who performed tasks that met their motives did report more positive volunteer experiences and intended to continue volunteering in the future (Clary et at. 1998; Houle et at. 2005) In this study, project characteristics refer to whether an OSS product is commercially viable. As more and more OSS products are market driven, they adopt a professional approach to achieving value by establishing a profitable business venture for which customers are willing to pay the going rate (Fitzgerald 2006) In this scenario, both customers and developers need 102

PAGE 118

to perceive value for money With the new OSS era, Fitzgerald (2006) argued that free-that is, zero cost-is replaced by a value-for-money concern and developers will not perceive an "itch worth scratching" as before Thus, project characteristics were brought into this study in order to examine the association of volunteer developers' perceptions of project commercial viability with their motivation. Years of OSS Experience Years of OSS experience represents the length of developers' experience on OSS development. According to work motivation literature, differential length of service would lead to different expectancy and valence perceptions (Brief et al. 1979; Hackett and Betz 1981; Singer and Coffin 1996) The findings from the volunteerism literature indicate that over time, volunteers may experience declining intrinsic motivation (Pearce 1993) and external forces can undermine intrinsic motivation (Clary and Miller 1986). Participation Status Participation status refers to whether OSS developers are volunteers or paid to participate in OSS development. According to the volunteerism literature volunteers were primarily motivated by intrinsic reasons (Sherman and Smith 1984; Singer and Coffin 1996). However, much empirical evidence indicates that although a volunteer's decision to start volunteering may be driven by strong intrinsic motivation such as altruistic goals his or her 103

PAGE 119

decision to continue is more likely based on an evaluation of costs and rewards than it is on altruistic motives (Philips 1982). Thus, in the context of OSS development, it is important to explore whether volunteers' motivation significantly differs from that of paid developers. Summary This study applies EVT, human resource theory, and volunteerism to the analysis ofOSS developers' motivation, work attitudes and continuance intention. EVT and volunteerism are considered helpful in measuring OSS developers' motivation since the majority of developers are volunteers; moreover, based upon different motivations, they may participate in OSS development at varying degrees due to the variation in their expected outcomes. Prior studies on the work attitudes and turnover intentions of IT professionals and those of volunteers can be largely found in the areas of human resource management and volunteer behavior. Therefore, the human resource theory and volunteerism are used to investigate the relevant issues in the context of OSS development. The research design presented in this study assumes that the variables (i.e., motivation, satisfaction, commitment, continuation, availability of schedule, expertise) used to study motivations and work attitudes for IT workers and volunteers should also be associated with OSS developers. Whether the variables used to study traditional workers also operate in a manner similar to those for OSS developers must first be empirically tested. 104

PAGE 120

CHAPTER 5 RESEARCH MODELS To investigate OSS developers' continuance intention and effort, two theoretical models are proposed: the continuation model and the effort model. The continuation model attempts to measure developers' motivation, to identify the antecedents of developers' satisfaction with participation in open source projects and commitment to OSS development, and to analyze the relationships of these constructs and future availability with continuance intention. In the effort model, the impact of developers' motivation on effort is at issue. Moreover, developers' availability for OSS development and their expertise are taken into consideration in the model. For each model, the hypothesis is developed to address the corresponding research questions in this study. Continuation Model The continuation model for this study is based on the EVT, human resource theories, and volunteerism. The model depicted in Figure 5 1 is devised for measuring OSS developers' satisfaction, open source commitment and continuance intention. 105

PAGE 121

Intrinsic Motivation Helping Enjoyment Peer Recognition Extrinsic Motivation Human Capital Career Advancement Personal Needs Vroom 1964 Ill Future Availability -_.,. ,' ,': Miller et al. 1990 + 810 I I I Vroom 1964 Continuance Intention + HI -Satisfaction Williams and Hazer 1986; Miller et al. 1990; lgbaria and Greenhaus 1992; Thather et al. 2002-2003 I , Open Source Commitment ' ' ' \ I I I I I Miller et al. 1990; ' \ , \ \ \ \ I I I I I I I I I I I Igbaria and Greenhaus 1992; Wu et al. 2007 Figure 5 1 Research Model ofOSS Developers' Continuance Intention It predicts that an OSS developer's intention to make future contributions is influenced by his motivation, level of satisfaction, commitment, and future availability for OSS development. Developers' satisfaction mediates the influence of motivations on commitment and continuance intention. Open 106

PAGE 122

source commitment is an intervening variable between satisfaction and continuance intention. Table 5.1 presents the operational definitions of constructs used in this model. Table 5.1 Construct definitions for continuation model Construct Definition Motivation on helping OSS developer's valence and expectancy for altruism and reciprocity Motivation on enjoyment OSS developer's valence and expectancy for enjoying OSS development Motivation on peer OSS developer's valence and expectancy recognition for being recognized by others within the community Motivation on enhancing OSS developer's valence and expectancy human capital for skills and abilities enhancement Motivation on career OSS developer's valence and expectancy advancement for career OHOrtunity_ Motivation on satisfying OSS developer's valence and expectancy personal needs for software modification or source code ac_guisition Satisfaction OSS developer's overall contentment with OSS develo_Q_ment Open source commitment OSS developer's dedication to OSS development OSS continuance intention OSS developer's intention to further engage in future open source projects Future availability The convenience ofOSS developer's future participation in OSS development The continuation model suggests that an OSS developer's intention to make future contributions is influenced by his level of satisfaction. Job satisfaction is, effectively, negatively related turnover (Boomsma 1985; Ward 1988). Locke and Latham ( 1990) argued that satisfied people will be more likely to stay with the organization and accept new challenges that they are 107

PAGE 123

offered. Cnaan and Goldberg-Glen (1991) revealed that people will continue to volunteer as long as the experience, as a whole, satisfies their unique needs. Thus, in the context of OSS development, satisfactory participation experience, in conjunction with psychological pleasure and/or tangible benefits, would greatly influence OSS participants' later continuance decision. Otherwise, dissatisfied participants may cease their involvement in OSS development. This leads to the first hypothesis. Hypothesis 1: Developers' satisfaction with participating in open source projects will directly and indirectly affect their intentions to contribute to future OSS development. In this study, OSS developers' satisfaction is considered a key attitudinal variable leading to their commitment to OSS development. Satisfaction is strongly related to subjective reports of organizational commitment (Williams and Hazer 1986; Lee and Mowday 1987) and is regarded a causal antecedent to organizational commitment (Mowday et al. 1982; Mak and Sockel 2001 ). Steers ( 1977) argued that workers whose needs are satisfied by an organization are likely to develop commitment to this organization. In contrast, when an organization fails to meet individuals' needs, employees' commitment level decreases. Likewise, Reilly (2005) suggested that low job satisfaction can lead to disruptive behavior and sabotage while high job satisfaction can enhance organizational commitment and citizenship. According to Dailey ( 1986), job satisfaction played a critical role in understanding the commitment 108

PAGE 124

of volunteers and it was a strong predictor of their devotion to volunteer activities. This leads to the second hypothesis. Hypothesis 2: Developers' satisfaction with participating in open source projects will have a direct and positive effect on their commitment to OSS development. The research on organizational commitment indicates that the construct has a central role in job scope and turnover literature, and it is strongly related to retention and turnover behavior (Bateman and Strasser 1984; Mowday et al. 1984; Dailey 1986; Williams and Hazer 1986). Prevailing models of turnover among information technology (IT) and volunteer workers have also confirmed that organizational commitment affects behavioral intentions to quit, thereby affecting turnover (Miller et al. 1990; Igbaria and Greenhaus 1992; Thatcher et al. 2002-3). Therefore, this study argues that developers who are more committed to their community are more likely to continue their contributions to future OSS development than those who are relatively uncommitted. This leads to the third hypothesis Hypothesis 3: Developers' commitment to OSS development will have a direct and positive effect on their intentions to contribute to future OSS development. Lerner and Tirole (2002) suggested that OSS developers are willing to continually contribute their efforts if they can derive expected benefits. Thus, once they recognize that those expectations were attained and met their 109

PAGE 125

valences, their positive continuance intention may also instinctively and correspondingly strengthen. In other words, developers' continuance intention would be a positive consequence of their motivations. This leads to the subsequent hypotheses. Hypothesis 4a: Developers' motivation associated with helping will have a direct and positive impact on their intentions to contribute to future OSS development. Hypothesis 4b: Developers' motivation associated with enjoyment will have a direct and positive impact on their intentions to contribute to future OSS development. Hypothesis 4c: Developers' motivation associated with peer recognition will have a direct and positive impact on their intentions to contribute to future OSS development. Hypothesis Sa: Developers' motivation associated with enhancing their skills and abilities will have a positive impact on their intentions to contribute to future OSS development. Hypothesis 5b: Developers' motivation associated with career advancement will have a direct and positive impact on their intentions to contribute to future OSS development. Hypothesis 5c: Developers' motivation associated with satisfying personal needs will have a direct and positive impact on their intentions to contribute to future OSS development. 110

PAGE 126

Organizational commitment refers to an individual s identification or involvement with the organization. Identification provides a context for pro-social behavior by raising the concern for collective interests which merge with the individual s own interests (O'Reilly and Chatman 1986; Johnson et al. 1999). 0 'Reilly and Caldwell ( 1981) indicated that certain aspects of the initial decision to join an organization are related to subsequent commitment. Volunteerism literature suggests that if volunteers expectations are met and they stay with the organization, they may opt to expand and diversify their duties and contribution which would in tum, increase their commitment (Ralston and Rhoden 2005). Following the concept of organizational commitment, OSS developers motivation would demonstrate a positive relationship with their commitment to OSS development. This leads to another set of hypotheses Hypothesis 6a: Developers' motivation associated with helping will have a positive impact on their commitments to OSS development. Hypothesis 6b: Developers' motivations associated with enjoyment will have a direct and positive impact on their commitments to OSS development. Hypothesis 6c: Developers' motivations associated with peer recognition will have a direct and positive impact on their commitments to OSS development. Ill

PAGE 127

Hypothesis 7a: Developers' motivation associated with enhancing their skills and abilities will have a positive impact on their commitments to OSS development. Hypothesis 7b: Developers' motivations associated with career advancement will have a direct and positive impact on their commitments to OSS development. Hypothesis 7c: Developers' motivations associated with satisfying personal needs will have a direct and positive impact on their commitments to OSS development. Based on Vroom (1964), if a person believes that an object can cause desired or prevent undesired consequences, then the person would have a positive attitude toward it, and vice versa. Locke and Latham ( 1990) also argued that consequences which correspond to a person's wants or values generate job satisfaction; similarly, those that thwart or do not correspond to one's wants generate dissatisfaction. Therefore, based on Vroom (1964) and Locke and Latham (1990), this study suggests that OSS developers' motivation would have a positive impact on their satisfaction level. This leads to the following hypotheses Hypothesis 8a: Developers' motivation associated with helping will have a positive impact on their satisfaction with participating in open source projects. 112

PAGE 128

Hypothesis 8b: Developers' motivation associated with enjoyment will have a positive impact on their satisfaction with participating in open source projects Hypothesis Be: Developers' motivation associated with peer recognition will have a positive impact on their satisfaction with participating in open source projects. Hypothesis 9a: Developers' motivation associated with enhancing their skills and abilities will have a positive impact on their satisfaction with participating in open source projects. Hypothesis 9b: Developers' motivation associated with career advancement will have a positive impact on their satisfaction with participating in open source projects. Hypothesis 9c: Developers motivation associated with satisfying personal needs will have a positive impact on their satisfaction with participating in open source projects In this study, availability refers to the convenience of schedule for OSS development. According to Miller (1985), convenience of schedule was negatively related to both the intention to leave and turnover Since the continuation model has been developed specifically to predict developers intentions to contribute to future OSS development the construct of availability is labeled as "future availability". Future availability is assumed to be a significant predictor of developers' continuance intention. In other words the more available developers are with their future schedules the more 113

PAGE 129

likely they are to voluntarily contribute their time and effort to OSS projects. Thus, this leads to the tenth hypothesis. Hypothesis 10: Developers' future availability will have a direct and positive impact on their continuance intention. Effort Model The effort model shown in Figure 5.2 is devised for measuring the effort that OSS developers contributed to OSS development. This model attempts to test the influence of OSS developers' motivation, current availability, and expertise on effort. Intrinsic Motivation Helping Enjoyment Peer Recognition Extrinsic Motivation Human Capital Career Advancement Personal Needs Current Availability Expertise Vroom 1964 Effort ',, Miller et al. 1990; Rotolo and Wilson 2006 Hackman and Oldham 1975; '' Bandura 1982; Olson et al. 1996 Fieure 5 2 Research Model ofOSS Develooers' Effort 114

PAGE 130

Effort is represented as hours per week that an individual developer spends on OSS development. This model predicts that OSS developers' effort is directly influenced by both their intrinsic and extrinsic motivators. Developers' expertise and current availability for OSS development also have direct effects on their effort. The operational definitions for these constructs are shown in Table 5.2. The proposed relationships among the constructs and the hypotheses are described shortly thereafter. Table 5.2 Construct definitions for effort model Construct Definition Motivation on helping OSS developer's valence and expectancy for altruism and reciprocity Motivation on enjoyment OSS developer's valence and expectancy for enjoying OSS development Motivation on peer recognition OSS developer's valence and expectancy for being recognized by others within the community Motivation on enhancing OSS developer's valence and expectancy human capital for skills and abilities enhancement Motivation on career OSS developer's valence and expectancy advancement for career opportunity Motivation on satisfying OSS developer's valence and expectancy personal needs for software modification or source code acquisition Current Availability OSS developer's current availability for OSS development Expertise The degree to which an open source project requires a variety of different activities in carrying out the work, involving the use of a number of different skills, abilities, and talents of a developer Effort The amount of hours per week an OSS developer puts into the project 115

PAGE 131

According to Vroom (1964), people must believe that exerting a given amount of effort can result in the achievement of a particular level of outcome (the effort-outcome relationship) Moreover the outcome must be attractive in order for people to be motivated to attain it (the valence-personal goals relationship) Thus, based on EVT, OSS developers' motivation could likely have a significant impact on their effort contributed to OSS development. Thus, this leads to the following hypotheses Hypothesis la: Developers' motivation associated with helping will have a positive impact on their effort. Hypothesis lb: Developers' motivation associated with enjoyment will have a positive impact on their effort. Hypothesis lc: Developers motivation associated with peer recognition will have a positive impact on their effort Hypothesis 2a: Developers' motivation associated with enhancing their skills and abilities will have a positive impact on their effort. Hypothesis 2b: Developers' motivation associated with career advancement will have a positive impact on their effort. Hypothesis 2c: Developers' motivation associated with satisfying personal needs will have a positive impact on their effort. OSS developers' availability has been discussed in the continuation model. In the effort model, developers' availability denotes the convenience of their current schedule for OSS development, which refers to the time they 116

PAGE 132

currently spend developing OSS projects. The effort model suggests that the more available the developers' current schedule the more likely they are willing to voluntarily contribute their effort to OSS projects This leads to the third hypothesis Hypothesis 3: Developers' current availability will have a positive impact on their effort. Literature suggests that individuals' perceptions of expertise impact their effort because they would exert more efforts on things they feel capable of handling (Bandura 1982; Olson et al. 1996; Douglas and Shepherd 2000; Shaver et al. 2001). Thus in the OSS setting developers are assumed to put more effort on project development if they perceive that their skills and abilities are useful and valuable for the projects and that this in tum, can fulfill their expectations. This leads to the fourth hypothesis Hypothesis 4: Developers' expertise will have a positive impact on their effort. 117

PAGE 133

CHAPTER6 METHODOLOGY In this study, the structural equation modeling (SEM) method with a cross sectional design was used to address the research questions. Prior to proceeding with the empirical research design, the identification of the research models was conducted along with the development of the survey instrument. Then, a number of developers were chosen from an OSS community to evaluate the instrument via a pre-tested survey before the formal collection of research data. Data collection across a variety of open source projects was carried out via an online survey. Next, the survey data was used to validate the proposed research models and subsequently test the hypotheses. Moreover, the influence of determinants of motivation and the motivation difference between volunteers and paid developers were also evaluated. The limitations of this research design were also discussed in this chapter. Identification of the Research Models Model identification refers to whether a research model is theoretically possible to derive a unique value for each and every free parameter from the observed data (Ullman 1996; Byrne 200 I; Kline 2005). If a unique solution for the value of the structural parameters can be found, the model is 118

PAGE 134

considered to be identified. As a result, the parameters are considered to be estimable and the model therefore testable (Byrne 2001 ). If a model is not identified, then it remains so regardless of the sample size and consequently it should be re-specified; otherwise, attempt to analyze them may be fruitless (Kline 2005). Additionally, any estimate of an unidentified parameter computed by a SEM program is arbitrary and should not be relied on (Raykov and Marcoulides 2000). Structural models may be classified as being just-identified, overidentified, or underidentified. Typically, models need to be overidentified in order to be estimated and in order to test hypotheses about relationships among variables (Ullman 1996). Byrne (200 l) also points out that the aim in SEM is to specify a model such that it meets the criterion of overidentification. A model is called overidentified if it has fewer parameters than observations (i.e., positive degrees of freedom) and it is also identified (Ullman 1996; Raykov and Marcoulides 2000; Byrne 2001; Kline 2005). In addition to overidentification, a structural model needs to satisfy another condition for identificationthat is, every unobserved (latent) variable must be assigned a scale (metric) For identification of the research models, the number of observations and the number of free parameters have to be determined at the very first. According to literature, the number of observations equals p(p+ 1)12, where p is the number of observed variables. The counting of free parameters is based on the following six rules (Bentler 1995). 119

PAGE 135

Rule 1: All variances of independent variables are model parameters. Rule 2: All covariances between independent variables are model parameters. Rule 3: All factor loadings connecting the latent variables with their indicators are model parameters. Rule 4: All regression coefficients (i.e., direct effects) between observed or latent variables are model parameters. Rule 5: The variances and covariances between dependent variables and the covariances between dependent and independent variables are never model parameters. Rule 6: For each latent variable included in a model, the metric of its latent scale needs to be set. Continuation Model Presented in Figure 6.1 is the measurement model of OSS developers' continuation model. It includes observed variables, measurement errors, latent variables and their corresponding factor loadings as well as the covariance of latent variables This model satisfies the necessary conditions for model identification, since there were more observations than free parameters. With 18 observed variables, there are (18 X 19) I 2 = 171 observations available to estimate a total of 75 parameters, i e 10 latent variables 12 measurement errors, 45 covariances, and 8 factor loadings. Thus, the model is overidentified with 96 degrees of freedom. 120

PAGE 136

Figure 6.1 Measurement Model of OSS Developers Continuation Model 121 ' '

PAGE 137

Originally, this measurement model should have 18 measurement errors. Since six motivational constructs (helping, enjoyment, peer recognition, human capital, career advancement, and personal needs) have only a single indicator variable and their respective error variance is set to 0, there are 12 measurement errors. Summarized in Table 6.1 are the numbers and types of free parameters for the research model. Table 6.1 Types and number of parameters for the continuation model Type Variable Total Helping, Enjoyment, Peer Recognition, Latent Variable Human Capital, Career Advancement, 10 Personal Needs, Future Availability, Satisfaction, Commitment, Continuance Measurement Error ev1,ev2,es1,es2,es3,es4,eo1,eo2,eo3, 12 ei 1, ei2, ei3 INMH.,.... INME, INMR, EXMK, EXMC, EXMN, FAV, SA, OSC, CI INME.,.... INMR, EXMK, EXMC, EXMN, FAV, SA, OSC, CI INMR .,.... EXMK, EXMC, EXMN, FAY, SA, OSC, CI Covariance EXMK.,.... EXMC, EXMN, FAV, SA, OSC, 45 CI EXMC.,.... EXMN, FAV, SA, OSC, CI EXMN.,.... FAV, SA, OSC, CI FAV.,.... SA, OSC, CI SA.,.... OSC, CI OSC.,.... CI 122

PAGE 138

Table 6.1 (Con't.) Type Variable Total FAV-FAV2 Factor Loadings SASA2 SA3, SA4 8 OSC OSC2, OSC3 CICI2, CB Effort Model The measurement model ofOSS developers' effort model is shown in Figure 6.2. Likewise, this model satisfies the necessary conditions for model identification, since there were more observations than free parameters With 12 observed variables, there are (12 X 13) I 2 = 78 observations available to estimate a total of 53 parameters which contain 9 latent variables, 5 measurem ent errors, 36 covariances, and 3 factor loadings. Thus, the model is overidentified with 25 degrees of freedom. This measurement model initially should have 12 measurement errors. Since six motivational constructs (helping, enjoyment, peer recognition, human capital, career advancement, personal needs) and the item "effort" are univariate and their corresponding residual terms are set to 0, thus there are 5 measurement errors. 123

PAGE 139

Figure 6 2 Measurement Model of OSS Developers Effort Model The numbers and types of free parameters for the effort model are shown in Table 6.2 124

PAGE 140

Table 6.2 Types and number of parameters for the effort model Type Variable Total Helping, Enjoyment, Peer Recognition, Latent Variable Human Capital, Career Advancement, 9 Personal Needs, Current Availability, Expertise, Effort Measurement Error evl,ev2,epl,ep2,ep3 5 INMH +-+ INME, INMR, EXMK, EXMC, EXMN, AV, EP, EFF INME +-+ INMR, EXMK, EXMC, EXMN, AV, EP, EFF INMR +-+ EXMK, EXMC, EXMN, AV, EP Covariance EFF 36 EXMK +-+ EXMC, EXMN, AV, EP, EFF EXMC +-+ EXMN, AV, EP, EFF EXMN +-+ AV, EP, EFF AV +-+ EP, EFF EP +-+ EFF AV -+AV2 Factor Loadings 3 EP --+ EP2, EP3 Instrumentation An online survey was administered to collect data from existing OSS participants across a variety of open source projects. Prior to conducting the survey, an application was submitted to Human Subjects Research to obtain approval. A copy of this approval is provided in Appendix A. The survey was comprised of two parts. One part solicited respondents to provide their 125

PAGE 141

demographic data such as, gender, age nationality, residency, number of years in IT field, years of contributing to OSS, the number of projects involved, the name of the project currently involved with, the current position in the project, time spent on the project, and the participation status (volunteer or paid to participation). The demographic part was used to collect characteristic information from OSS developers. The second part was the instrument itself. In this part, the questions were adapted and developed based upon pre-validated measures in the related literature Twelve sets of pair-wised questions related to expectancy and valence were used to measure respondents' intrinsic and extrinsic motivations for participating in open source projects. Intrinsic motivations consisting of helping, enjoyment, and peer recognition, were assessed by the scales developed based upon Cheney (1983), Faber and O'Guinn 1992, Wasko & Faraj (2000, 2005), and Hars & Ou (2002). Extrinsic motivations containing enhancing human capital, career advancement, and satisfying personal needs, were measured by the scales adapted and extended from JDI (Smith et al. 1969), Igbaria & Greenhaus (1992), Ghosh et al. (2002), Hars & Ou (2002), and Hertel et al. (2003). OSS developers' satisfaction was assessed using seven-point semantic differential scales from Spreng et al. 's study (1996). Satisfaction was measured with the overall perspective rather than specific facets, since empirical evidences show that in many cases the general measures are as reliable as the sum of facet measures and are also more inclusive measures 126

PAGE 142

(Scarpello and Campbell 1983). Open source commitment was operationalized using the items from the Organizational Commitment Questionnaire (Mowday et al. 1982) to measure the extent to which an OSS developer identifies with OSS development goals Continuance Intention was assessed by using three scale items extended from Mathieson's ( 1991) behavioral intent i on scale to ask respondents whether they intend to continue their contributions to future OSS development. OSS developers' availability of current and future schedule for OSS project were measured with a two-item scale adapted from Miller et al. (1990) The scale items for expertise were developed based upon the measurement of skill variety from Hackman and Oldham's ( 1980) Job Diagnostic Survey (JDS) In addition, there were two scale items devised for survey respondents to determine whether the project they are currently working on is commercially viable All instrument and measurement sources were shown in Appendix B. Pilot Test The purpose of the pilot test is to make sure content validity and construct reliability The instrument was pre-tested with 58 OSS developers selected from Debian.org in order to verify the psychometric properties of the scales based upon Straub's ( 1989) suggestions. Deb ian developed by more than a thousand open source programmers, is a free operating system that uses 127

PAGE 143

Linux as its kernel but most of the basic software utilities come from the GNU Survey participants were requested to complete the instrument and asked to provide their comments on wording, length, and structure The respondents' remarks were used to improve the instrument's format, questions, and scales as necessary. The participants included 55 males and 3 females. Among them, 53 participants were volunteers and 3 were paid employees. Their average age was 30 years old, with an average of 7 years of OSS development experience, 7 projects involved, 7 5 years ofiT experience, and 9 hours per week working for the Debian project. The reliability of each construct was verified with the Cronbach 's alpha. The value ofCronbach's alpha above 0.6 was used to verify that the construct was the internal consistency ofthe indicators. Table 6.3 demonstrated the descriptive statistics of each individual construct and their respective Cronbach 's alphs value. The project type was not listed, since Deb ian is not a commercialized product and the participants in the pilot test were not asked to answer the corresponding questions. 10 http: // www .gnu.org-The GNU project was launched in 1984 to develop a complete Unix style operating system which is free software: the GNU system. (GNU is a recursive acronym for "GNU's Not Unix") 128

PAGE 144

Table 6.3 Descriptive statistics and Cronbach's alpha for the pilot test Standard Cronbach's Construct Mean Deviation Alpha Helping 5.666 0.141 0.771 Enjoyment 5.893 0.179 0.751 Peer Recognition 4.876 0.055 0.927 Human Capital 5.434 0.230 0.934 Career Advancement 4.266 0.032 0.868 Personal Needs 6.021 0.110 0.646 Satisfaction 5.500 0.354 0.538 Open Source Commitment 5.885 0.724 0.781 Continuance Intention 6.374 0.063 0.880 Future Availability 4.422 0.063 0.908 Current Availability 4.293 0.342 0.749 Expertise 5.115 0.235 0.666 Based upon descriptive data, each construct revealed a reasonable distribution across the ranges. Except satisfaction, the Cronbach 's alpha of all other scales was acceptable. Since the alpha value of satisfaction was below the commonly acceptable value 0.6 for exploratory research, this study employed the scale of satisfaction based upon Wu et al. 's (2007) study on OSS developers instead of the original scale items adapted from Hackman's and Oldham's (1980) measurement on job satisfaction of traditional workers. 129

PAGE 145

Sampling and Data Collection The subject groups of this study are OSS developers who are identified by the open source projects that maintain a presence on the Internet. The sample was drawn from developers listed for open source projects hosted on SourceForge.net. SourceForge.net is the world's largest OSS development web site, providing support for one million members and 100,000 projects. The survey respondents were randomly selected from currently active projects including communication utilities, database, desktop, enterprise systems, networking, financial, games, multimedia, scientific, engineering, religion and philosophy, format and protocol, etc Survey participation was solicited via an invitation email that describes the purpose of the study, a request for their voluntary participation, and a statement of potential risk for the participant (See Appendix C). Additionally, participants were informed that completion of the survey implies informed consent. The survey was anonymous, and no individual developers were excluded from participation on the basis of race, color, sex, national origin, religion, handicap, sexual orientation, or any other extraneous consideration. A hyperlink to online survey form was provided for the respondents in the email. If a subject chose to reply the survey, he/she was redirected to a different web site that contains the online questionnaire. The survey period ran 30 days with no follow-up. 130

PAGE 146

The expected number of responses was determined by the number of constructs in the proposed research models According to Newton and Rudestam (1999), the target number of responses is calculated as 50 + (8 X the number of constructs) for estimating R2 Therefore, for the continuation model, a minimum of 130 usable responses is required and the sample size should be above 1300 if a response rate of I 0% is expected. For the effort model, a minimum of 122 responses is expected, which means that the sample size should be over 1220 However, to increase the reliability of the research models, a larger number are necessary. During the survey period, 2,500 invitation emails for survey participation were sent off. Fallowing a single round of data collection with a response rate of 10%, 214 usable responses were gathered The 214 OSS participants' position in their own project is listed in Table 6.4. The sample contained 211 males and 3 females Among them, 181 OSS participants are volunteers and 33 are paid employees. The respondents spread all over the world, of which 115 respondents are from North America, 5 from South America, 83 from Europe, 8 from Oceania (Australia and New Zealand), 2 from Africa, and I from Asia; while 115 persons living in North America, 4 in South America, 78 in Europe, 9 in Oceania, 2 in Africa, and 6 in Asia. The respondents' average age is 30 years old, with an average of 5 years of experience on OSS development, 5 projects involved, 8 years of IT experience, and 9 hours per week working for the project(s) they are involved 131

PAGE 147

m. Almost all subjects report roles that are central to the development of OSS. Hence, these subjects collectively are referred as developers. Table 6.4 Survey respondents' position in OSS projects hosted by SourceF orge Position Frequency Percent Developer 128 59.8 Maintainer 30 14.0 Packager 25 11.7 Project Manager 8 3.7 Tester 7 3.3 Designer 6 2.8 Other 5 2.3 Bug Reporter 3 1.4 Bug Fixer 2 0.9 Total 214 100.0 In the proposed models, each individual motivational construct has two indicator variables, and each indicator variable is formed by calculating the product of "expectancy" item and "valence" item. Based upon Vroom ( 1964 ), the value of each motivational construct is equal to the sum of the values of their individual indicator variables Indicator variables' descriptive statistics for the continuation model are shown in Table 6.5. All items reveal a reasonable distribution across the ranges. 132

PAGE 148

Table 6.5 Descriptive statistics of indicator variables for the continuation model Item Mean Standard Deviation INMH1 5 524 0.847 INMH2 5.613 0.947 INME1 6.019 0.684 INME2 5.942 1.023 INMR1 4.898 1.116 INMR2 4.961 1.059 EXMK1 5.619 1.001 EXMK2 5 813 0.934 EXMC1 4.458 1.298 EXMC2 4.556 1.246 EXMN1 6 035 0.814 EXMN2 6.057 0.811 FAV1 4 350 1.377 FAV2 4.410 1.362 S1 5.590 0.918 S2 5.620 0.846 S3 4.850 1.436 S4 5.370 1.065 OSC1 5.230 1.261 OSC2 6.340 0.877 OSC3 6.340 0.878 Cll 6.340 0.804 Cl2 6.130 0.895 CI3 6.270 0.918 PJI 4.810 1.684 PJ2 4.090 1.739 Legend: CI = OSS continuance intention; OSC =Open source commitment; S = Satisfaction; FAY= Future availability; PJ =Project type; INMH =Motivation on helping; INME = Motivation on enjoyment; INMR = Motivation on peer recognition; EXMK = Motivation on increasing human capital; EXMC =Motivation on career advancement; EXMN = Motivation on satisfying personal needs. Likewise, in the effort model, only volunteers' data were used to estimate the proposed associations. Table 6.6 illustrates descriptive statistics of three 133

PAGE 149

constructs' (availability, expertise, effort) indicator variables, since the descriptive statistics of all motivational constructs are listed in Table 6.5. Except the effort construct, all other items in the effort model reveal a reasonable distribution across the ranges as well. The item "effort" showed a very high standard deviation because the range of hours per week (effort) spent on OSS development was from 0 to 46. Table 6.6 Descriptive statistics of indicator variables for the effort model Item Mean Standard Deviation AVl 4 370 1.550 AV2 3.860 1.636 EP1 5.260 1.301 EP2 5.410 0.994 EP3 4.910 1.381 EFF 7.280 7.276 Legend: EFF = Effort; EP = Expertise ; AV =Current availability. Data Analyses In assessing the proposed research models, construct reliability, validity, and hypothesis testing were measured using LISREL (linear structural relationships). LISREL accounts for all of the covariance in the data and so allows for the examination of all of the correlations, shared variances, and paths in the model when estimating the significance level and coefficient of the paths (Bollen 1989). It also enables unidimensionality analysis, an examination not possible using a principal components factor analysis or Cronbach 's alpha reliability tests (Gerbing and Anderson 1988; Segars 1997). 134

PAGE 150

Based upon recommendations by Gerbing and Anderson (1988), data analysis for this study was performed to test first the validity of the measurement model before making any attempt to evaluate the full structural model. In essence, if a measurement model is operating adequately, one can then have more confidence in findings related to the assessment of the hypothesized structural model (Byrne 2001, p.147). Confirmatory factor analysis (CFA) was used to assess the validity of the indicator items. CFA is more appropriate than alternative approaches such as exploratory factor analysis in areas with strong a priori theory and pre-validated measurement scales (Bagozzi and Phillips 1982), as in this study Reliability Reliability is the consistency or the tendency in a respondent to respond in the same or in a very similar manner to an identical or near-identical question (Bums and Bush 2002, p. 269). Theoretically, there are two measures of reliability using CFA: the indicator reliability (Long 1983) and the composite reliability (F om ell and Larcker 1981 ). The indicator reliability reflects the percent of variation that is explained by the construct it measures In contrast, the composite reliability reflects the internal consistency of the indicators (Werts et al. 1974) In this study, the composite reliability based upon the Cronbach 's alpha was used to measure the consistency of responses. Theory suggests that the Cronbach's alpha greater than .70 is considered acceptable in confirmatory 135

PAGE 151

research and the value should exceed .60 for exploratory research (Nunnally 1967, 1978; Peter 1979; Fornell and Larcker 1981; Nunnally and Berstein 1994; Gefen et al. 2000) Since this study was confirmatory, a composite construct reliability of. 70 was used as the criterion to indicate a reliable set of questions. The effort construct was not evaluated, because it had only a single indicator variable. Table 6. 7 demonstrates the Cronbach 's alpha for both continuation and effort models. Table 6.7 Reliability measurements for constructs Construct Cronbach's Alpha Motivation on helping 0.678 Motivation on enjoyment 0.643 Motivation on peer recognition 0.898 Motivation on enhancing human capital 0.918 Motivation on career advancement 0.869 Motivation on satisfying personal needs 0.757 Satisfaction 0.810 Open source commitment 0.812 Continuance intention 0.861 Future Availability 0.853 Availability 0.764 Expertise 0.709 Project Type 0.757 Except helping and enjoyment, the alpha values for peer recognition, human capital career advancement, personal needs, satisfaction, open source commitment, continuance, future availability, current availability, expertise and project type exceeded 0.70, indicating adequate reliability. Since both helping and enjoyment have two indicator variables, the inter-item correlation 136

PAGE 152

is used as appropriate checks for their reliability The correlation between the two indicators of helping was 0.516 and enjoyment's was 0.512, which demonstrated moderate correlation. Therefore, the two constructs remained unchanged in both motivation and effort models in the interest of content validity. Measurement Model Assessment In the measurement model, each indicator variable used to measure a hypothesized latent variable was modeled as a reflective indicator To evaluate convergent validity for the measurement model, selected goodness-of-fit statistics related to the CFA and standardized indicator factor loadings were examined. Adequate model fits, and high factor loadings for each construct and the lack of noticeable cross loadings are required to meet convergent validity for the measurement model. The assessment of discriminant validity was done by examining each indicator's factor loadings (Chin 1998) and comparing the average variance extracted (AVE) against the correlation of constructs. To meet discriminant validity, all indicator factor loadings should be significant and indicators should load higher on the construct of interest than on any other variable (Chin 1998). In addition, the AVE for each construct should exceed the squared correlation between that and any other construct for adequate discriminant validity (F om ell and Larcker 1981 ). 137

PAGE 153

Continuation Model The analysis gave a chi-square (x2 ) of 189.90 with 96 degrees of freedom yielding the ratio 1.978: I (p = 0.00000), which was well within the recommended range of3:111 The goodness offit index (GFI) was 0.9, adjusted goodness of fit index (AGFI) was 0.81, comparative fit index (CFI) was 0.94, incremental index of fit (IFI) was 0.95, Normed fit index (NFI) was 0.91, root mean square residual (RMR) was 0.052, and root mean square error of approximation (RMSEA) was 0.074. Those values were within the commonly acceptable benchmarks12, suggesting adequate model fit. Thus, it was reasonable to conclude that the continuation model represented an acceptable fit to the data. Each construct's indicator variables and their corresponding factor loadings in the continuation model are listed in Table 6.8. All indicator factor loadings were significant (p < 0.001) and exceeded 0.65. Therefore, convergent validity for the continuation model was met by adequate model fit and high factor loadings. 11 With reference to Gefen et al. (2000), the ratio of r! to degrees of freedom should be between I and 2 (Hair et at., 1998), but the IS literature has been recommending just a x2 as small as possible (Segars and Grover, 1993) and demonstrating a ratio of x2 to degrees of freedom smaller than 3: I (Chin and Todd, 1995). 12 In a well-fitting model, GFI above 0.90, AGFI above 0.80, CFI above 0.95, NFI above 0.90, RMR and RMSEA below 0.50 are expected. IFI developed by Bollen (I 989) addresses the issues of model parsimony and sample size IFI's value consistent with that ofCFI reflects a well-fitting model. RMSEA's values as high as 0.08 represent reasonable errors of approximation in the population (Browne and Cudeck, 1993). MacCallum et at. (1996) elaborated on these cutpoints and noted that values ranging from 0.08 to 0.10 indicate mediocre fit, and those greater than 0.10 indicate poor fit. However, Hu and Bentler ( 1999) suggested a value of 0.06 to be indicative of good bit between the hypothesize model and the observed data. 138

PAGE 154

Table 6.8 Factor loadings for the continuation model Items JNMH lNME lNMR EXMK EXMC EXMN FAY s osc CI INMHI .784 .114 .199 .107 242 .057 .010 .103 .254 -.014 INMH2 .710 .354 .035 267 .089 .063 -.083 .020 .126 .143 INMEI .089 853 .017 .185 .123 .233 065 .000 109 .002 INME2 .290 .760 .201 018 058 108 007 .073 .052 254 !NMRI 084 .046 .926 .073 092 043 .069 .053 .117 .041 lNMR2 097 .119 .907 074 .145 022 -.013 014 .095 .130 EXMKI .072 .096 .068 .900 217 .001 .135 -.088 .022 .093 EXMK2 .217 .124 .100 .867 .241 .037 .091 .013 .075 .134 EXMCI -.019 -.098 .129 .269 .835 -.032 .012 .186 -.021 -.124 EXMC2 090 023 .127 .178 .911 032 .011 074 .061 026 EXMNI .061 .081 -.078 .024 .093 .884 .158 097 038 042 EXMN2 016 .036 097 006 -.085 .880 .076 .125 080 .081 FAVI -.042 .020 .029 .125 038 047 .872 .125 .182 .130 FAV2 .003 .032 .027 .071 018 .030 .908 .071 .158 .122 SI -.146 -.050 057 033 088 020 .090 .834 .093 -.038 S2 .094 016 020 -.038 .113 .135 .053 .875 -.009 084 S3 078 .167 008 .107 -.066 .128 -.071 .742 .084 360 S4 .342 017 047 086 .108 -.042 .141 .782 -.063 049 OSCI .039 -.022 230 .094 .114 074 .132 .041 .800 .107 OSC2 .134 .106 .027 -.019 .010 -.007 .162 .041 .805 .290 OSC3 .187 .106 -.014 .000 -.063 .091 .124 .057 .769 .361 Cll .175 .094 .048 .105 -.049 -.075 .093 .132 .460 .735 Cl2 .034 .145 .095 099 029 106 .304 .037 .351 .733 Cl3 -.076 .101 .171 176 169 203 054 025 .409 676 I Legend: CI = OSS continuance intention; OSC =Open source commitment; S = Satisfaction; FAY= Future ava i lability; INMH =Motivation on helping; INME = Motivation on enjoyment; INMR = Motivation on peer recognition; EXMK = Motivation on increasing human capital; EXMC = Motivation on career advancement; EXMN = Motivation on satisfying personal needs. 2 Extraction Method: Principal Component Analysis 3. Rotation Method: Varimax with Kaiser Normalization 139

PAGE 155

Likewise, according to Table 6.8, the high factor loadings within the corresponding construct and the lack of noticeable cross loadings supported discriminant validity for the measurement model. Table 6.9 contains the AVE for each construct and their correlation matrix. Results showed that the square root of the AVE for a construct was greater than the correlation between that construct and the other ones Hence it satisfied the criterion of discriminant validity as well. Table 6.9 AVE and correlation of constructs for the continuation model Construct Cl osc s FAY INMH IN ME INMR EXMK EXMC EXMN Cl 0 90 osc 0 78 0.81 s 0 08 0.11 0.76 FAV 0.44 0.40 0.20 0.86 INMH 0.34 0.38 0 .15 0.09 1.00 INME 0.38 0.32 0 09 0 .15 0.52 1.00 INMR 0 .25 0 22 0.08 0 .11 0.25 0 26 1.00 EXMK 0.30 0.21 O.QJ 0 27 0.37 0.33 0 .23 1.00 EXMC 0 08 0 02 0 27 0.06 0.02 0.06 0.22 0 38 1.00 EXMN 0 .19 0 .17 O.o7 0 .14 0.16 0.15 0.04 0 09 O.DI 1.00 I Legend: CI = OSS continuance intention; OSC =Open source commitment; S = Satisfaction; FAY= Future availability; TNMH =Motivation on helping; TNME = Motivation on enjoyment; TNMR = Motivation on peer recognition; EXMK = Motivation on increasing human capital; EXMC = Motivation on career advancement; EXMN = Motivation on satisfying personal needs. 2 Diagonal elements (in bold) are the square root of the average variance extracted 13. Effort Model A chi-square (x2 ) of 67.13 with 25 degrees of freedom generated the ratio 2.685:1 (p = 0.00001), which was within the recommended range of3:1. The GFI was 0.94, AGFI was 0.82, CFI was 0.91, IFI was 0 92, and RMR was 13 AVE is calculated as: :E,1? !(:E,1? + :EVar(e )), (Fomell and Larcker, 1981; Gefen et al., 2000). 140

PAGE 156

0.056. Those values were within the commonly acceptable benchmarks, suggesting adequate model fit. RMSEA at 0.087 was relatively high, although it indicated a mediocre fit (MacCallum et al. 1996) NFI at 0.88 was slightly below the 0.90 threshold But, NFI is sensitive to sample size and may indicate poor fit with small samples, even when the model is correct and thus, is not a reliable indicator of model fit (Bentler and Bonnett 1980). Thus, the effort model exhibited an acceptable fit to the data. Table 6.10 shows each individual construct's indicator variables and their corresponding factor loadings for the effort model. All indicator factor loadings were significant (p < 0 001) and exceeded 0 65. Thus, convergent validity for the measurement model was met by adequate model fit and high factor loadings. Table 6.10 Factor loadings for the effort model Items INMH lNME INMR EXMK EXMC EXMN AV EP EFF INMHI .862 150 .190 .082 -.162 .058 -.015 -.073 .118 INMH2 .685 .411 .005 245 092 083 .053 .192 .107 INMEI .174 750 .033 146 -.115 237 .070 079 .190 INME2 227 .833 .208 .055 -.013 -.108 .119 .014 .031 INMRI .111 046 .934 064 .107 046 039 069 047 INMR2 .071 .162 910 106 .124 -.004 -.070 .058 .091 EXMKI .047 098 .077 .923 .180 .001 .094 .145 -.030 EXMK2 222 .138 .116 .874 .244 .031 .068 029 000 EXMCI 005 -.094 .116 224 .903 -.035 -.041 -.031 015 EXMC2 095 .035 .119 .163 904 .040 .045 .155 .012 EXMNI 036 .025 -.064 .051 034 881 088 .161 006 EXMN2 062 .079 .107 .023 -.028 879 014 .146 -.071 141

PAGE 157

Table 6.10 (Con't.) Items INMH INME INMR EXMK EXMC EXMN AV EP EFF AVI -.069 .052 .004 .119 -.074 .220 .860 .112 -.098 AV2 .085 .094 -.029 .024 -.010 -.088 .905 015 .149 EPI -.120 .436 .001 .326 .110 .159 .136 .706 .185 EP2 .066 012 003 .167 -.035 .148 .156 846 -.125 EP3 .025 007 .137 076 .116 -.183 .030 .835 .080 EFF .165 -.084 .127 -.019 .031 -.061 .058 .022 .944 I Legend: EFF = Effort; EP = Expertise; AV = Current availability; TNMH = Motivation on helping; INME = Motivation on enjoyment; TNMR = Motivation on peer recognition; EXMK =Motivation on increasing human capital; EXMC = Motivation on career advancement; EXMN = Motivation on satisfying personal needs. 2. Extraction Method: Principal Component Analysis 3. Rotation Method : Varimax with Kaiser Nonnalization To evaluate discriminant validity, factor loading and each construct's AVE were examined. Table 6.10 shows that the high factor loadings within the corresponding construct and the lack of noticeable cross loadings supported discriminant validity for the management model. Table 6.11 exhibits the AVE for each construct and their correlation matrix. Results showed that the square root of the AVE for a construct was greater than the correlation between that construct and the other ones. Hence, the criterion of discriminant validity was met. 142

PAGE 158

Table 6.11 AVE and correlation of constructs for the effort model Construct EFF EP AV INMH IN ME INMR EXMK EXMC EXMN EFF 1.00 EP 0 03 0 64 AV 0 .05 0 08 0.83 INMH 0.22 0.25 0.08 1.00 INME 0 06 0 34 0 .18 0.52 1.00 INMR 0.20 0.16 O.Dl 0.25 0.26 1.00 EXMK 0 00 0 .41 0.18 0 37 0 .33 0 .23 1.00 EXMC 0 04 0 .21 0.02 0.02 0 06 0.22 0.38 1.00 EXMN 0 .10 0 .15 0.25 0.16 0 .15 0.04 0.09 0 .01 1.00 I. Legend: EFF = Effort; EP = Expertise; AV = Current availability; TNMH = Motivation on helping; TNME = Motivation on enjoyment; TNMR = Motivation on peer recognition; EXMK =Motivation on increasing human capital; EXMC = Motivation on career advancement; EXMN = Motivation on satisfying personal needs 2. Diagonal elements (in bold) are the square root of the average variance extracted Structural Model Assessment To test the hypotheses, the structural models for both continuation and effort models that specified the hypothesized associations were evaluated based upon their corresponding measurement model. Structural path estimates were measured for significance by examining the statistical significance of all structural parameter estimates. Statistically significant values at p < 0 .05 were considered to be acceptable in this study. Continuation Model Estimation of the model generated an overall x2c1o9 ) value of212.68 with the x2 to degrees of freedom ratio of 1:1.951. The AGFI, CFI IFI, NFI values were 0.82, 0.94, 0.94, and 0.90, respectively. The RMR value was 143

PAGE 159

0.076, and RMSEA was 0 073. Although GFI at 0.88 was slightly lower than the 0.90 threshold, the overall model fit indexes indicated an acceptable fit. In the interest of parsimony, a final model was determined by the removal of the fifteen insignificant associations. A nested x2 of22.78 with thirteen degrees offreedom (p = 0.044) demonstrated no violation of data by eliminating the paths. Figure 6.3 shows the standardized path coefficients and levels of significance for the significant associations as well as the overall fit indexes. Based on the structural model, continuance intention (p < O.OOI) was well predicted by open source commitment ({J= 0.74), satisfaction ({J= 0.23) and future availability ({J = 0.23). Thus, Hypothesis I, 3, and I 0 were supported. Motivation on helping ({J= 0.38) had a positive and statistically significant effect on commitment (p < 0.00 I). It implies that volunteer developers' commitment to OSS development increased as their level of motivation on helping rose, supporting Hypothesis 6a Additionally, motivation on helping ({J= 0.22,p < O.OOI) had an indirect effect on continuance intention through both satisfaction and commitment. 144

PAGE 160

Model fit:-/= 212 68 (df = 109), GFI = 0.88, AGFI = 0 82, NFI = 0 90, CFI = 0 94, IFI = 0 94 RMR = 0 076 RMSEA = 0.073 R2=0.14 significant at p < .001 significant at p < .01 significant at p < 05 Figure 6.3 Standardized LISREL Estimations of the Continuation Model Motivation on helping (/3= 0.27,p < 0.01) and career advancement (/3= 0 36, p < 0.001) showed direct and statistically significant affects on satisfaction, supporting Hypothesis 8a and 9b In other words, developers who reported higher levels of motivation on helping and career advancement 145

PAGE 161

were more likely to report high levels of satisfaction. Although motivation on enhancing human capital had a statistically significant effect on satisfaction ({J = -0.24, p < 0.05), the relationship was negative. That is, developers with high levels of motivation on enhancing human capital were not satisfied with their experience. Thus, Hypothesis 9a was not supported. Both human capital ({J= 0.05,p < 0.05) and career advancement ({J= -0.08,p < 0.01) had an indirect effect on continuance intention through satisfaction. Effort Model The estimations gave a chi-square (:x,2 ) of 71.92 with 29 degrees of freedom yielding the ratio 1:2.480 (p = 0.00002). The GFI, AGFI, CFI, and IFI values were 0.94, 0.83, 0 92, and 0.92, respectively. The RMR value was 0.056, and RMSEA was 0.081. Although the RMSEA was moderately high and NFI (0.87) value was lower than the 0 90 threshold, the overall model fit indexes indicated an acceptable fit. In the interest of parsimony, a final model was determined by the removal of the four insignificant associations. A nested :x,2 of 4.79 with four degrees of freedom (p = 0.309) demonstrated no violation of data by eliminating the paths Figure 6.4 shows the standardized path coefficients and levels of significance for the structural model as well as the overall fit indexes. 146

PAGE 162

Model fit: -/ = 71. 92 (df = 29), GFI = 0.94 AGFI = 0.83, NFI = 0.87 CFI = 0.92 IFI = 0.92 RMR = 0 056 RMSEA = 0 .081 significant at p < .001 sign i ficant at p < .01 significant at p < 05 Figure 6.4 Standardized LISREL Estimations of the Effort Model According to the model, OSS developers' effort was predicted primarily b y helping (/3 = 0 29) at p < 0 00 I, secondarily by peer recognition (13 = 0.21 p < 0.01) and current availability (13 = 0.20, p < 0.05), thus supporting Hypotheses I a, lc and 3 Although motivation on enjoyment had a statistically significant effect on effort (/3= -0.31, p < 0 .001 ), the relationship 147

PAGE 163

was negative. This result did not support Hypothesis 1 b. Therefore, the analyses revealed that volunteer developers' effort was dependent upon their motivation on helping, peer recognition, and current availability. The Analysis of Influential Factors of Motivation The impact of project characteristics (commercial viability) and years of OSS experience on developers' motivations were analyzed by the corresponding structural models. To measure the impact of the participation status on OSS developers' motivations, testing for invariance of factor loadings and in variance of latent variable means were conducted. All of these analyses were performed using LISREL. Project Characteristics A structural model was developed to investigate the impact of project characteristics on developers' motivators. Figure 6.5 demonstrates the outcomes of LISREL estimations. The analysis indicated that a chi-square (x2 ) of 27.50 with 16 degrees of freedom yielding the ratio 1:1.719 (p = 0.03627), which was well within the recommended range. The GFI at 0.96, AGFI at 0.92, CFI at 0.96, IFI at 0.96, NFI at 0.91, were within the commonly acceptable thresholds. RMR at 0.054 and RMSEA at 0.063 were slightly high, but they were still within the acceptable range. Thus, these indexes suggested an adequate model fit. 148

PAGE 164

Model fit:/= 27 50 (df = 16), GFI = 0 96, AGFI = 0 92 NFI = 0.91 CFI = 0 96, IFI = 0 96 RMR = 0 054, RMSEA = 0.063 R2=0.90 R2 = 0 60 R2 = 0.08 R2=0.2 0 R2 =0.00 significant at p < .001 significant at p < .05 ns insignificant at p < .05 Figure 6 5 LISREL Estimations of Project Characteristics to Motivations According to the estimations, the project characteristics significantly influenced all three intrinsic motivators : helping(/)= 0.95,p < 0 001), enjoyment(/)= 0 .75,p < 0.001), and peer recognition(/)= 0.29,p < 0.001). In other words, if developers perceive that an open source project as having high commercial viability, they are more likely to report high levels of intrinsic motivation In contrast, the project commercial viability influenced only two extrinsic motivators, human capital(/)= 0.45,p < 0 001) and personal needs(/)= 0.18,p < 0.05). Due to very low R2 values for career advancement and personal needs the project characteristics were not 149

PAGE 165

considered as a good predictor of developers' extrinsic motivation. Similarly, the project characteristics did not strongly predict peer recognition because of low R2 value (0.08). Years of OSS Experience Similarly, a structural model was developed to test the influence of the length of OSS experience on developers' motivators. The standardized path coefficients and levels of significance for the structural model as well as the overall fit indexes are shown in Figure 6.6. The analysis gave a chi-square (x2 ) of 30.27 with 13 degrees of freedom yielding the ratio 1:2.328 (p = 0.004), which was within the recommended range The GFI at 0 95, AGFI at 0.90, CFI at 0.91, and IFI at 0.91, were all within the commonly acceptable thresholds. However, NFI at 0.85 was lower than the 0.90 value. RMR at 0.076 and RMSEA at 0.086 appeared to be fairly high. Nevertheless, they were acceptable, since MacCallum et al. (1996) suggested that the value ranging from 0.08 to 0.10 indicates mediocre fit. Overall, these indexes suggested an adequate model fit. ISO

PAGE 166

Model fit: / = 30.49 (df = 13), GFI = 0 95, AGFI = 0 90, NFI = 0 85, CFI = 0.91, IFI = 0.91 RMR = O.D76, RMSEA = 0 086 R2 = 0 .57 R2 = 0 49 R2=0.10 R2 = 0.23 R '=o.oo R2=0.05 significant at p < .001 significant at p < 05 ns insignificant at p < 05 Figure 6.6 LISREL Estimations of OSS Experience to Motivations The results revealed that the length of OSS developers' experience significantly influenced motivation on helping (/3 = 0. 76, p < 0.001 ), enjoyment W = 0 70, p < 0 001 ), peer recognition (p = 0.32, p < 0.001 ), human capital (p = 0.48,p < 0.001), and personal needs (/3= 0.22 p < 0.05). Thus, developers with more experience reported high levels of motivation on helping, enjoyment, peer recognition, human capital, and personal needs However, due to extremely low R2 values for personal needs, years of OSS experience was not a strong predictor of developers' extrinsic motivation 151

PAGE 167

apparently. Similarly, the project characteristics did not strongly predict peer recognition because of low R2 value (0.1 0). Participation Status Participation status was assessed by testing for the equivalence of the factor loading pattern and testing for latent mean differences. Thirty-three paid developers' data was included to analyze the motivation difference between volunteers and paid workers. The equivalence of the factor loading pattern was used to examine whether the measurement model was invariant across volunteers and paid developers The results of the chi-square difference test for the intrinsic and extrinsic motivators across the two groups are shown in the Table 6.12. Due to the small p-values, the factor loadings of the two groups were significantly different in extrinsic motivation In contrast, the difference in intrinsic motivation across the two groups was not quite significant, since it was significant at p < 0.05 but insignificant at p < 0.0 1. Table 6.12 Testing for the equivalence of factor loadings across two groups Motivation Hypothesis x2 df p value Equal(HOl 30.92 18 0.0294 Intrinsic Unequal (H 1) 22.36 15 0.0987 Difference 8.56 3 0.0358 Equal (HO) 64.24 23 9.108E-06 Extrinsic Unequal (H 1) 38.95 20 0.0068 Difference 25.29 3 1.343E-05 152

PAGE 168

With respect to the in variance of latent mean structures, the results shown in Table 6 .13 revealed that volunteers reported the higher level of motivation on helping, enjoyment, enhancing human capital, and satisfying personal needs than paid developers did. In contrast, paid developers demonstrated higher motivation on peer recognition and career advancement than volunteers. However, based on the t statistic, the only significant difference between volunteers and paid developers was personal needs. In other words, the two groups were not indifferent in other motivators. Table 6.13 Invariant mean tests of motivational constructs Construct Parameter Standard t statistic Estimate Error Helping 0 .35 0.19 1.84 EnjoY!!!_ent 0.26 0.19 1.39 Peer recognition -0.07 0.19 -0.36 Human capital 0.19 0.19 0.97 Career advancement -0. 35 0.19 -1.78 Personal needs 0.98 0.22 4.42 I. Negative numbers represent the latent mean of volunteers lower than that of paid developers 2. represents significant at < 0.00 I Limitations The proposed research had several limitations. First, given a modest response rate, the results may be influenced by non-response bias. Theoretically, in order to test for non-response bias, it is desirable to collect data from non-respondents and compare it to the data supplied willingly (Keams and Lederer 2004). However, operational constraints impede further !53

PAGE 169

investigation of non-respondents to test for this bias. Second, the empirical analysis only considered those survey participants who were willing to respond, but did not take non-respondents and open source discontinuers into consideration. Third, since the survey was online and anonymous, the researcher was not able to observe or control the environment where the survey was conducted. Moreover, the proposed research encountered several problems in design and analysis contexts due to the limitations of structural equation modeling. Based upon Tomarken and Waller (2005), the major limitations are listed as follows: 1. SEM is still a large-sample technique. That is, although there are some recent suggestions in the literature about the analysis of structural equation models in smaller samples, most application of SEM still require large samples (Kline 2005; Tomarken and Waller 2005). 2. Structural models are typically only approximations of reality (e.g., Cudeck and Henley 1991; Browne and Cudeck 1993; MacCallum and Austin 2000; Raykov and Marcoulides 2000; Meehl and Waller 2002; MacCallum 2003). One way that SEM models are approximations is by omitting variables that are implicated in the causal processes or other features of a model. Such omissions present a misleading picture of the measurement and/or causal structure, and commonly result in biased parameter estimates and inaccurate estimates of standard errors (e.g., Kaplan 1989; Mauro 1990; Reichardt 2002). 154

PAGE 170

3 Although measures of global model fit test the validity of model-imposed restrictions on the covariance matrix, they do not directly test what might be considered lower-order components of a model (Tomarken and Waller 2003). Such components include specific model parameters (e g., path coefficients) and relevant quantities that can be derived from such parameters, such as the proportion of variance in an endogenous variable that is accounted for by the specified predictors in the model (Tomarken and Waller 2005). 4 SEM analyses generally focus on summary statistics; that is, covariances and, in some cases, means. The reliance on summary statistics and measures of global fit may lead researchers to ignore the issue of how well models fit at the level of the individual participant (Tomarken and Waller 2005). Raykov and Marcoulides (2000) also point out that an important limitation in SEM is the infrequent evaluation of estimated individual-case residuals. Individual-case residuals are routinely used in applications of regression analysis because they help researchers with model evaluation and modification. 5. The estimators most commonly used in empirical applications ofSEM (e.g., maximum likelihood) use all available information in the covariance matrix of the observed variables to generate parameter estimates. This feature allows the effects of a misspecified parameter to be propagated beyond the specific equation in which it occurs (e g., Kaplan 1988, 1989). Thus, an omitted path from a given latent variable to another latent variable could 155

PAGE 171

potentially bias estimates of other structural or measurement parameters that would appear to be far downstream from the misspecified parameter (Tomarken and Waller 2005). 6 In the SEM context, even a completely correct theoretical model (e.g one that includes all necessary variables and paths) can fit poorly and yield highly biased estimates if the study is poorly designed (Tomarken and Waller 2005). Design factors such as the reliability and number of observed indicators and the number of time points assessed in repeated measures studies are significant influences on statistical power and the sensitivity of measures of fit to misspecifications in the SEM context (Matsueda and Bielby 1986; Mandys et al. 1994; Raykov 2000; Tomarken and Waller 2003). 156

PAGE 172

CHAPTER 7 DISCUSSION AND CONCLUSIONS Many issues affect OSS developers' intention to continue their involvement in future open source projects and the effort they put into OSS development. Overall the findings of this study provide insight into OSS developer motivation. This chapter summarizes the results from the tests of the research models and presents the implications from both theoretical and practical perspectives Additionally, extended research on OSS and the contributions of this study are discussed The final section concludes this study. Summary ofResults According to the results, open source commitment has the strongest influence on volunteer developers' intentions to participate in future open source projects, followed by future availability and their satisfaction with participating in open source projects. These associations demonstrate that commitment to OSS development satisfactory experience and recognition of the convenience of future schedule for contributing to open source projects all influence volunteer developers' continuance intention. This finding supports the arguments from volunteerism that individuals who believed that they had satisfied their initial motivations for volunteering were more likely to form 157

PAGE 173

intentions to continue to volunteer. Organizational commitment has been regarded a useful index to predict future levels of volunteer involvement and more flexible work schedules make it easier for volunteers to commit to volunteer work. Volunteer developers' commitment primarily depends on their motivation to help. This finding is also consistent with the volunteerism theory. According to the volunteerism literature, people engage in volunteer work because they want to help others. Moreover, if volunteers' expectations are met and they stay with the organization, then they may wish to expand and diversify their duties and contribution which, in tum, would increase volunteer-organization commitment. OSS developers' satisfaction is influenced by both motivation to help and career advancement. The finding supports the postulation of a gift culture that people will receive help and support, given that previous contributions have created a tacit reciprocity agreement. Also, it reveals that helping indirectly influences continuance intention through satisfaction. Similarly, career advancement opportunities derived from OSS development enhances satisfaction and thus, indirectly influences continuance intention. Career advancement likely refers to future job offers, shares in commercial open source-based companies, and future access to the venture capital market. It might also be realized through signaling. Volunteer developers may regard working on OSS as an effective way to demonstrate (and advertise) their capabilities and skillfulness. The result is consistent 158

PAGE 174

with the assertion that the larger the contribution of an individual to open source projects, the more likely it is that the commercial software vendors will recognize the value of the individual. Intriguingly, enhancing human capital revealed a negative correlation to satisfaction This finding is inconsistent with OSS and volunteerism literature which contradicts not only that learning by doing is of essential importance to the success of OSS projects, but also that a positive volunteer experience can be achieved when a volunteer experiences new learning opportunities with the potential for personal or professional growth In the context of OSS development, developers who start an open source project are generally master programmers, and they primarily contribute their time and skills to developing the software or solving the related technical problems. Thus, it suggests that enhancing human capital might not be a significant concern for these volunteers. With respect to effort, developers' motivation to help enjoyment, peer recognition and current availability significantly influenced the time they contributed to OSS development. However, enjoyment was found to have a negative effect on effort. In other words, if developers' participation in open source projects is based on enjoyment, they would not dedicate much time to software development. Although existing OSS research has no consensus regarding which motivation has the most dominant impact on individual OSS contributors' effort, this study suggests that developers' effort may be primarily influenced by intrinsic motivation. 159

PAGE 175

Project commercial viability and the length of experience did influence OSS developers' motivation. Both of them influence the motivation to help, enjoyment, peer recognition, enhancing human capital, and personal needs. Developers with more experience were likely to have higher levels of motivation to help, enjoyment, peer recognition, enhancing human capital, and personal needs than those with less experience. Likewise, if developers perceived that an open source project was highly commercially viable, they were likely to reflect this same motivation as was found for years of experience on OSS development. In addition, paid developers primarily differed from volunteers in motivation on personal needs, i.e., they had less personal needs for the software. In conclusion, the results of this study indicate that volunteer developers are motivated by a combination of intrinsic (i.e., helping) and extrinsic (i.e., career advancement) motivators. Although it cannot be determined that volunteers will necessarily make future contributions based on these motives, the research subjects in this study have confirmed that OSS development is instrumental in the motivation to help and in career advancement. In addition, their effort is primarily associated with helping and peer recognition Implications Job attitude constructs of satisfaction and commitment are considered by many to be inherently related to motivation. The continuation model developed here is useful for measuring the influence of both intrinsic and 160

PAGE 176

extrinsic motivators on satisfaction, commitment, and continuance intention. The results of this study show that motivation to help indirectly influences continuance intention through satisfaction and commitment; on the other hand, career advancement indirectly influences continuance intention through satisfaction. Motivation on career advancement was assumed to be an extrinsic motivator, yet it behaves as an intrinsic motivator by influencing continuance through satisfaction. This might suggest that satisfaction with OSS development is increased as a result of the career advancement opportunities the open source experience provided. The ability of the model to distinguish between the effects of extrinsic motivation and intrinsic motivation is important for eventually discovering the full motivational structure of OSS developers. Prior studies of volunteers (outside ofthe open source context) have shown that intrinsic motivation is generally more important than extrinsic. The results of this study support a different premise, that extrinsic motivators are equally significant to volunteer developers. The existence of economic benefits to OSS developers provides a strong indication that the open source community has created an exchange arrangement between developers and organizations that benefit from OSS. Adopters of OSS must employ technicians who are skilled in its use. These organizations are likely to recruit from the very open source community that developed the code because the community produces skilled participants and identifies the best talent. These OSS developers also possess the knowledge 161

PAGE 177

needed to utilize the OSS. By giving away their code for free, those OSS developers benefit from better career opportunities. The open source movement is likely to succeed as long as this economic partnership between OSS developers and OSS adopters holds. This research has several implications for open source project managers. Since the study focused on the estimation of perceived benefits, open source project managers may consider how to create an environment that provides both intrinsic and extrinsic rewards to developers. Secondly, these benefits apply not only to highly visible open source projects like Linux, but apply to other types of open source projects as well. Furthermore, managers of commercial software development should consider applying aspects of OSS development to improving in-house practices, such as managing virtual teams which are common to software outsourcing arrangements. In particular, they should consider incorporating "communities of practice" concepts into their project management approaches Investigating how commercial software development could benefit from OSS development practices and principles is an area of research that begs further exploration. von Hippe! and von Krogh (2003) provide an excellent theoretical treatise of the potential benefits of a mixed strategy of blending proprietary software development with OSS development. Although the results of this study show that adequate motivation exists, the data shows that not all OSS developers agree that they will benefit equally. For example, as indicated by Likert's ratings of neutral or below neutral, 162

PAGE 178

48.6% of the respondents indicated that participating in open source projects will not make finding a better job any easier; further, 44.4% do not expect to derive any career advancement opportunities. Therefore, open source project managers should continue to improve upon their procedures for creating open opportunities to work on projects of the participant's choosing, so as to allow participants the control to develop and hone their skills of choice and to develop software that will be of direct value to them. Since satisfaction with OSS development is greatly influenced by motivation on career advancement, open source project managers should be able to attract and retain the best developers by providing ample opportunities for participants to demonstrate their capabilities in a highly visible way. For example, open source communities might consider using special awards to recognize outstanding accomplishments or introduce additional grade levels to recognize varying levels of performance. Also, open source communities should consider ways in which it can improve networking opportunities for its developers. The network process should enable developers to promote themselves to other members, enlist their support, gather information about job opportunities, and identify sought-after skills; concurrently, it should be able to promote developers' recognition among peers which, in tum, would lead developers to contribute more time to OSS development since peer recognition is one of the determinants of developer effort. This research also has implications for information systems researchers. The work motivation model presented here is an excellent starting point from 163

PAGE 179

which to build more elaborate models. An ideal empirical design for testing OSS developers' continuance intention would include observing novice OSS developers' initial motivations and post-expectations in order to faithfully capture the complex, dynamic interrelationships between involvement and continuance decisions. However, this study was limited to a theoretical model which did not take into account the developers' initial motivations Knowledge ofOSS developers' initial valences and expectancies are needed to understand why some open source members quit. It may be the case that discontinuers start with a set of valences and expectancies that are different from continuers. Another possibility is that discontinuers may be unwilling to wait for the realization of delayed economic benefits. Thus, a longitudinal research or more complex framework may be required. Additional Research The results of this study facilitate further research in analyzing the antecedents of OSS developer satisfaction and commitment-as well as the association thereof-since this research did not show that satisfaction influenced commitment in the context of OSS development. This finding is quite different from the general findings of prior studies on traditional worker behavior. Such an analysis can provide valuable insight not only into the process by which OSS developer satisfaction and commitment are formed, but also the identification of factors that can influence the relationship between the two attitudes. 164

PAGE 180

Unlike organizations who hire employees, open source communities rely upon an unusual combination of intrinsic and extrinsic motivation to attract and retain developers. A number ofOSS researchers (e g., Dinkelacker and Garg 2001; Masum 2001) have explored the extent to which the paradigm of open source movement might influence the future of employment and rewards. Hence, understanding the recruitment and retention of volunteer developers is of great importance not only to OSS organizations, but also to other industries. Can strategies for retaining talented people in the settings of OSS development be applied to traditional working environments? This topic is worth further investigation as it relates to several different perspectives, such as work motivation and industrial and organizational cultures. Contributions Although many advocates of the OSS movement have theorized about the motivations of OSS developers and these developers have been surveyed regarding their motivations and effort, the theories are vague, imprecise, and not thoroughly tested. This study is among the first to apply EVT to understanding the motives of OSS developers and the relationship between motives and effort. Hence, a major contribution of this study is twofold: the initial development of a work motivation model, based on EVT, which studies the motives of OSS developers and an exploratory model which measures the effort that developers dedicate to OSS development. The continuation model improves the understanding of motivation by measuring both the subjective 165

PAGE 181

importance of the motive and a prediction of the outcome expectancy, while the effort model helps identify the impact of developers' motivations on their effort. Additionally, this is one ofthe earliest studies on investigating OSS developers' intention to further volunteer for future open source projects. Therefore, it will pave the way for social science researchers who attempt to explore the impact of psychological cognition on OSS developers' continuance behaviors from the perspective of intrinsic and extrinsic motivations. Generally, OSS development depends upon the willingness of volunteers. In reality, there are a number of factors that might influence their intentions to voluntarily contribute to open source projects. Thus, examining the reasons for future participation is essential for understanding the sustainability of open source movement Conclusions The main purpose ofthis study was to investigate volunteer developers' motivation. It attempted to identify what motivates volunteers to continue making OSS contributions and to expand their effort. To this end, EVT, human resource theories, and volunteerism were adapted to form two research models: a theoretical model of OSS developers' continuation, and an exploratory model of developers' effort. The findings regarding the motivation of developers showed that volunteering can reflect multiple motivations: intrinsic motivation (i.e., 166

PAGE 182

helping) and extrinsic motivation (i.e., career advancement). A positive and satisfactory experience leads to a positive attitude toward retention. Moreover, volunteer developers' commitment primarily depends on motivation to help. These results, taken as a whole, suggest that OSS developers are motivated both by the altruism of helping and the economic incentives for career advancement. This study also provided empirical evidence that volunteers may vary their motivation according to their actual experience and project commercial viability. The length of experience suggests a life cycle effect on motivation; i.e., the longer developers stay, the higher levels of motivation grow. In addition, the more commercially viable an open source project is, the more likely developers are motivated. Developers' effort was primarily influenced by both their motivation to help and peer recognition, which implies that the impact of intrinsic motivators on effort is stronger than that of extrinsic motivators. Enjoyment demonstrated a significantly negative effect on effort, suggesting that developers with high levels of motivation on enjoyment did not spend more time in open source projects than those with less motivation on enjoyment. Finally, scheduling availability is an important factor of predicting continuation and effort. 167

PAGE 183

APPENDIX A. HUMAN SUBJECTS APPROVAL 0 Unlvo,.lty of Colo,.do ot Donvor ond Hoolth Sclencoa Contor : -:.=-=":::*""::S.::.:"IIifi>CU..:;,_.,=:.:R.:... ____ Commitlee ____ -_lnslllutlona ___ I_R_-_ .___ 'd------------Compua Box 120, P.O. Box 173364 Denver, C-.clo 110217-33&4 Phone : 3CJ3.5S6..4060. Fax: 303-5S6-33n DATE: July6, 2006 TO: FROM: Dorothy Yates, HSRC Chair SUBJECT: ChomgGuang Wu Hwnan Subjects Resc:an:h PrO oco1112006 An Empirical Analysis of Open Source Developers' Motivations and Continuance Intentions Using Expectancy Value Theory and the Satisfaction-Commitment Model Your protocol has been approved as ex'mpt under CFR Title 45 Part 46.101.b. This approval is J!ood for up to one year from this date Your responsibilities as a researcher include: If you make changes to your research protocol or design you should contact the HSRC so that we can determine if your exempt status continues You an: responsible for maintaining all documentation of consent. Unless specified differently in your protocol, all data and consents should be maintained for three years If you should encounter adverse human subjects issues, please contact us immediately If your research continues beyond one year from the above date, contact the IISRC for an extension The HSRC may audit your documents at any time Thank you for submitting yolir protocol and good luck with your research Campuses Oowrtown fitzsunons ar t-/lntn and CO;l1r c:tdV 168

PAGE 184

APPENDIX B. SURVEY ITEMS Construct Item Wording and Code Source I intend to continue participating Adapted from in OSS projects rather than Mathieson 1991 Continuance discontinue my involvement (Cil) Intention (CI) I expect to engage in OSS projects Adapted from in the future (CI2) Mathieson 1991 I would be willing to make future Adapted from contributions to OSS projects Baker et al. 1992 (CI3) I am willing to put in a great deal of effort beyond that normally Open Source expected in order to help the Commitment community be successful (OCl) Adapted from (OC) I support the community's value Mowday et al. 1982 (OC2) I really care about the fate of the community (OC3) How do you feel about your overall experience of participating in OSS projects : -Very dissatisfied Very Satisfaction satisfied (SA 1) Adapted from (SA) -Very displeased Very Spreng et al. 1996. pleased (SA2) -Very frustrated Very contented (SA3) -Very terrible Very delighted (SA4) 169

PAGE 185

Construct Item Wording and Code Source In the future, my schedule allows Future for adequate time to work on this Availability project (FAVl) Adapted from In the future, I expect to have Miller et al. 1990 (FAV) sufficient time to work on this project (FAV2) The time that I work on this open Current source project fits my schedule Availability just fine (AV I) Adapted from (AV) Regularly, I am able to find Miller et al. 1990 sufficient time to work on this open source project (AV2) I have a number of complex or high-level skills that are useful on this project (EPI) This open source project is one Adapted from Expertise (EP) where I have the skills and talents Hackman and needed to provide valuable Oldham 1980 knowledge for it (EP2) My expertise enables me to readily make major contribution to this project (EP3) Project In your opinion, to what extent is this product commercially viable? Developed based on Characteristics To what extent are IT vendors Fitzgerald 2006 (PJ) interested in this product? 170

PAGE 186

Intrinsic Motivations Construct Item Wording and Code Source Being able to help other OSS Adapted from Hars developers is tome and Ou 2002 (INMHlV) Participating in OSS projects will Adapted from Hars give me an opportunity to help and Ou 2002 Helping others solve their problems (INMH) (INMHlE) Sharing code and programming Developed based on practice in an OSS community is Wasko and Faraj to me (INMH2V) 2000 Participating in OSS projects will Developed based on give me an opportunity to share Wasko and Faraj my knowledge about programming 2000 (INMH2E) Whether I like the nature of the Adapted from Faber project is tome and O'Guinn 1992 (INMElV) Enjoyment Writing open source code will Adapted from Faber (INME) allow me to write code that I enjoy and O'Guinn 1992 (INMElE) Having fun developing OSS is Adapted from Hars to me (INME2V) and Ou 2002 Participating in OSS projects will Adapted from Hars give me a feeling of fun and Ou 2002 (INME2E) 171

PAGE 187

Construct Item Wording and Code Source Gaining peer recognition in the Adapted from Hars open source community is and Ou 2002 to me (INMR 1 V) I will gain recognition from others Adapted from Hars Peer in the open source community and Ou 2002 Recognition through my contributions (INMR) (INMRlE) Earning respect from other Adapted from developers in the open source Cheney 1983 community is to me (INMR2V) By my OSS contributions, I will Adapted from earn the respect of other Cheney 1983 developers in the community (INMR2E) 172

PAGE 188

Extrinsic Motivations Construct Item Wording and Code Source Improving my performance at software development is to me (EXMK IV) Participating in OSS projects will improve my software development Adapted from Human Capital performance (EXMK1 E) Ghosh et al. 2002; (EXMK) Advancing my skills at software Hars and Ou 2002; development is tome Hertel et al. 2003 (EXMK2V) Participating in OSS projects will advance my skills in developing software (EXMK2E) Improving my job opportunities is to me (EXMC 1 V) Participating in OSS projects will Adapted from Career get me a better job (EXMC1E) Ghosh et al. 2002; Advancement Advancing my career is Hars and Ou 2002; (EXMC) to me (EXMC2V) Hertel et al. 2003 Participating in OSS projects will aid in advancing my career (EXMC2E) Developing software that corresponds to my needs is __ to me (EXMN 1 V) Personal Writing open source code will Adapted from Needs produce software tailored to my Ghosh et al. 2002; (EXMN) needs (EXMN1E) Hars and Ou 2002; I engage in OSS products that I Hertel et al. 2003 intend to use (EXMN2V) I will use the OSS products that I helped develop (EXMN2E) 173

PAGE 189

Demographics 1. Gender 2. Nationality 3. Residency 4.Age 5. Number of Years in IT Field 6. Participation Status [volunteer, paid] 7. Years of contributing to OSS 8. Number of OSS projects involved 9. What is the name of the OSS product that you are currently working on? 10. Current role in the project [coding, debugging, testing, bug reporting, algorithm design, user interface design, documentation, communication, maintainer, packager, project management] 11. How much time do you currently spend on the project (average hours per week)? 174

PAGE 190

APPENDIX C. INFORMED CONSENT Dear OSS contributor, I am Chuck Wu, a PhD student at University of Colorado at Denver. Currently, I am conducting research on open source software development for my doctoral thesis. You are selected for this study because of your involvement in open source projects. As an experienced developer, your responses are essential to developing an understanding of this issue. My goal is to understand the reasons why OSS development is successful and to identify ways to improve ass experience for developers. Therefore, your help is much appreciated and acknowledged. This is a ONE TIME request. The survey will take you about 20 minutes to complete. Please click on the following link http://carbon.cudenver.edu/-jgerlach/OSSks.html to access the survey. If there is anything I can do to assist you, please feel free to email me. This survey is strictly anonymous. Please be assured that any information you provide will be used with the strictest of confidence. Your individual responses to this survey will not be disclosed and will be reported as part of statistical summaries only. Participation in this survey is voluntary, and does involve some risk. If you choose to participate, you have the right to not complete any and all portions of the survey. It's possible that the information you supply might be unintentionally disclosed to the general public, even though I take precautions to prevent this from occurring. Completion of this survey implies acknowledgment ofthis risk and informed consent. Please contact the Human Subjects Research Committee (HSRC) of University of Colorado at Denver regarding your rights as a research subject. Thank you very much for your support of my research. 175

PAGE 191

ChuckWu c 1 wu@ouray.cudenver.edu Tel: 1-303-8610168 Research Administration UCDHSC 1380 Lawrence Street, Suite 300 Campus Box 120, PO Box 173364 Denver, CO 80217-3364 U.S.A Tel: 1-303-5564060 176

PAGE 192

BIBLIOGRAPHY Abelson, M.A. "Examination of Avoidable and Unavoidable Turnover ." Journal of Applied Ps y chology 72 (1987) : 382-386. Ambrose, M.L., and C.T. Kulik. "Old Friends, New Faces: Motivation Research in the 1990s." Journal of Management 25, no.3 (1999): 231-292. Andrews, F M., and S.B. Withey. Social Indicators of Well-Being New York: Plenum Press, 1976. Anderson, J.C., and F.L. Moore "The Motivation to Volunteer." Journal of Voluntary Action Research 7 (1978): 120-129 Anderson, E.W., and M .W. Sullivan. "The Antecedents and Consequences of Customer Satisfaction for Firms, Marketing Science 12, no. 2 (Spring 1993): 125-143. Appel be B. "The Future of Open Source Software." Journal of Research and Practice in Information Technology 35 no. 4 (2003) : 227-236. Arnold, H.J., and D C Feldman A Multivariate Analysis of the Determinants of Job Turnover Journal of Applied Ps y chology 67 (1982) : 350-360 Bagozzi, R.P., and L.W. Phillips "Representing and Testing Organizational Theories : A Holistic Construal." Administrative Science Quarterl y 27 (September 1982): 459-489 177

PAGE 193

Baker, J., M Levy, and D. Grewal. "An Experimental Approach to Making Retail Store Environment Decisions." Journal of Retailing 68 (Winter 1992) : 445-460 Bandura, A. "Self-Efficacy: Toward a Unifying Theory of Behavioral Change." Psychological Review 84 (1977): 191-215. ----------."Self-Efficacy Mechanism in Human Agency American Psychologist 37, no 2 (1982): 122-147 Bar-Tal, D. "American Study of Helping Behavior: What? Why: and Where?" In Development and Maintenance of Prosocial Behavior : International Perspectives on Positive Morality, edited by E. Staub, D. Bar-Tar, J Karylowski, and J. Reykowski 5-27. New York: Plenum, 1984. Barahona, G H.P. Quiros, and T. Bollinger. "A Brief History ofFree Software and Open Source." IEEE Software 16, no. 1 (1999): 32-33. Baroudi, J.J "The Impact of Role Variables on IS Personnel Work Attitudes and Intentions." MIS Quarterly 9, no. 4 (1985): 341-356. Bass, L., P. Clements, and R. Kazman. Software Architecture in Practice. Reading, Massachusetts: Addison-Wesley, 1998. Bateman, T., and S. Strasser. "A Longitudinal Analysis ofthe Antecedents of Organizational Commitment." Academy of Management Journal27 (1984): 95-112. Bates J., C. DiBona, K.R. Lakhani, and B. Wolf. Hacker Survey. The Boston Consulting Group, 2002. Batson, C.D. The Altruism Question: Toward a Social-Psychology Answer. Hillsdale, NJ: Erlbaum, 1991. 178

PAGE 194

Beale, A.V. "Exploring Careers through Volunteerism." The School Counselor 32 (1984): 68-71. Behling, 0 and F .A. Starke "The Postulates of Expectancy Theory Academy ofManagementJournall6, no. 3 (1973): 373-388. Behrman, D.N., and W.O Perreault. "A Role Stress Model of the Performance and Satisfaction of Industrial Salespersons." Journal of Marketing 48 (Falll984): 9-21. Benson, P., J. Dohority L. Garman, E. Hanson, M. Hochschwender, C. Lebold, R. Rohr, and J Sullivan "Intrapersonal Correlates of N onspontaneous Helping Behavior." Journal of Social Psychology II 0 (1980) : 87-95. Bentler, P.M., and D G Bonnett. "Significance Tests and Goodness of Fit in the Analysis of Covariance Structures, Psychological Bulletin 88, no 3 (1980): 588-606. ---------. EQS Structural Equations Program Manual Encino, CA: Multivariate Software, 1995. Bergquist, M., and Ljungberg, J. The Power of Gifts: Organizing Social Relationships in Open Source Communities." Information Systems Journall1, no. 4 (October 2001) : 305-320. Bhattacherjee, A. "Understanding Information Systems Continuance: An Expectation-Confirmation Model." MIS Quarterly 25 no 3 (2001): 351-370. Bitzer, J., and G. Premkumar. "Understanding Changes in Belief and Attitude toward Information Technology Usage: A Theoretical Model and Longitudinal Test." MIS Quarterl y 28, no 2 (2004) : 229-254 179

PAGE 195

----------, S. Wolfram, and P.H. Schroder "Intrinsic Motivation in Open Source Software Development." MIT working paper, 2004, http : //opensource mit,edu/papers/bitzerschrettlschroder.pdf. (accessed August 18, 2006). Blau P M Exchange and Power in Social Life. New York: John Wiley and Sons 1964 Bluedom, A.C. "A Unified Model of Turnover from Organizations." Human Relations 35 (1982): 135-153. Bollen K.A. Structural Equations with Latent Variables. New York : John Wiley and Sons, 1989 Bollinger, T., R. Nelson, K M. Self, and S Turnbull. "Open-Source Methods : Peering through the Culture." IEEE Software 16, no. 4 (July 1999) : 8-11. Bonaccorsi, A., and C Rossi. "Why Open Source Software can Succeed Research Policy 32, no. 7 (2003): 1243-1258 Boomsma, A. "Nonconvergence, Improper Solutions, and Starting Values in LISREL Maximum Likelihood Estimation." Psychometrica 50, no 2 ( 1985): 229-242. Brief, A.P., M. van Sell and R.J Aldag "Vocational Decision Making among Women: Implications for Organizational Behavior Academy of Management Review 4 ( 1979): 521-530. Browne, M.W., and R. Cudeck. "Alternative Ways of Assessing Model Fit." In Testing Structural Equation Models, edited by K.A. Bollen and J.S. Long, 136-192. Newbury Park, California : Sage Publications, Inc., 1993. Bums, A.C., and R F. Bush. Marketing Research 41h ed Englewood Cliffs, NJ: Prentice Hall 2002 180

PAGE 196

Byrne, B M. Structural Equation Modeling with AMOS: Basic Concepts, Applications and Programming. Mahwah, NJ: Lawrence Erlbaum Associates, 200 1. Campbell, A., P.E. Converse, and W.L. Rodgers. The Quality of American Life New York: Russel Sage Foundation, 1976. Cavanagh, S.J. "Job Satisfaction ofNursing Staff Working in Hospitals." Journal of Advanced Nursing 17, no. 6 (1992): 704-711. Cheney G "On the Various and Changing Meanings of Organizational Membership: A Field Study of Organizational Identification." Communication Monographs 50 (1983): 342-362. Chin, W.W and P.A. Todd. "On the Use, Usefulness, and Ease ofUse of Structural Equation Modeling in MIS Research : A Note of Caution." MIS Quarterly 19, no 2 (June 1995): 237-246. ---------. "The Partial Least Squares Approach to Structural Equation Modeling." In Modern Methods for Business Research, edited by G .A. Marcoulides, 295-336. Mahwah, NJ: Lawrence Erlbaum Associates, 1998. Christen, M., G. Iyer, and D. Soberman. Job Satisfaction, Job Performance, and Effort : A Reexamination Using Agency Theory Journal of Marketing 70, no 1 (2006) : 137-150 Clary, E.G., and J. Miller. "Socialization and Situational Influences on Sustained Altrusim Child Development 57 (1986): 1358-1369. ----------, M. Snyder, R.D. Ridge, J. Copeland, A.A. Stukas, J. Haugen, and P Miene. "Understanding and Assessing the Motivations of Volunteers : A 181

PAGE 197

Function Approach Journal of Personality and Social Psychology 74, no. 6 (1998): 1516-1530. Cnaan, A.R., and R .S. Goldberg-Glen. "Measuring Motivation to Volunteer in Human Services." Journal of Applied Behavioral Science 27, no. 3 (September 1991 ): 269-284. Coleman, J.A. Foundations of Social Theory. Cambridge, Massachusetts: Belknap Press, 1990. Comerford, R. "The Path to Open-Source Systems ." IEEE Spectrum 36, no. 5 ( 1999): 25-31. Connors, T.D. The Volunteer Management Handbook. New York: John Wiley and Sons 1995 Csikszentmihalyi, M. Beyond Boredom and Anxiety. San Francisco, California: Jossey-Bass, 1975. Cudeck, R., and S.J Henley. "Model Selection in Covariance Structure Analysis and the Problem of Sample Size: A Clarification." Psychological Bulletin 109 (1991): 512-519. Cumbey, D.A., and J.W Alexander. "The Relationship of Job Satisfaction with Organizational Variables in Public Health Nursing." Journal of Nursing Administration 28 (1998) : 39-46. Curtis, J E., E. Grabb, and D Baer. "Voluntary Association membership in Fifteen Countries: A Comparative Analysis." American Sociological Review 57 (1992): 139-152 Cuskelly, G. The Influence of Committee Functioning on the Organizational Commitment of Volunteer Administrators in Sport." Journal of Sport Behavior 18 (1995): 254-269. 182

PAGE 198

Dailey, R .C. "Understanding Organizational Commitment for Volunteers: Empirical and Managerial Implications Journal of Voluntary Action Research 15 (1986): 19-31. Dalton, D., and W. Todor. "Turnover Turned over : An Expanded and Positive Perspective." Academy of Management Review 9 (1979): 225-235 David, P.A., A. Waterman, and S. Arora. FLOSS-US the Free / Libre/Open Source Software Survey for 2003 http : //www stanford.edu/group/floss-us/report/FLOSS-US-Report pdf. (accessed September 28, 2005). Deci, E.L. Intrinsic Motivation. New York: Plenum Press, 1975. ----------, and R.M. Ryan. Intrinsic Motivation and Self-Determination in Human Behavior. New York: Plenum Press 1985 ----------, R Koestner, and R.M. Ryan. "A Meta-Analytic Review of Experiments Examining the Effects of Extrinsic Rewards on Intrinsic Motivation." Psychological Bulletin 125 (1999): 627-688 Dempsey, B.J., D. Weiss, P. Jones, and J. Greenberg. "Who Is an Open Source Software Developer?" Communications of the ACM 45, no. 2 (February 2002): 67-72. Demsetz, H. "Towards a Theory of Property Rights." American Economic Review 57 (1967): 347-359. Dingelacker, J., and P K. Garg "Corporate Source: Applying Open Source Concepts to a Corporate Environment." In Making Sense of the Bazzar: rr Workshop on Open Source Software Engineering Proceedings of the International Conference on Software Engineering (ICSE '01), Toronto, Ontario, Canada, 200 I. 183

PAGE 199

Dorsch, K.D., H.A. Riemer, V. Sluth, D.M. Paskevich, and P Chelladurai. "What Determines a Volunteer's Effort?" International Year of Volunteers, 2002, http:// www nonprofitscan.calpdf/SR!Dorsch-SR2EnglishWeb. pdf. (accessed April 23, 2007). Dougherty T.W., A.C. Bluedorn, and T.W. Keon "Precursors ofEmployee Turnover: A Multi-Sample Causal Analysis Journal of Occupational Behavior 6 (1985): 259-271. Douglas, E.J., and D A. Shepherd. "Entrepreneurship as a Utility Maximizing Response." Journal of Business Venturing 15 (2000): 231-251. Eisenberg, N. Altruistic Emotion, Cognition and Behavior. Hillsdale, NJ: Erlbaum, 1986. Ellis, S.J., and H.K Noyes. By the People: A History of Americans as Volunteers San Francisco, CA: Jossey-Bass Publishers, 1990. Faber, R.J., and T.C. O'Guinn. "A Clinical Screener for Compulsive Buying." Journal of Consumer Research 19 (December 1992): 459-469. Feather, N.T. Expectations and Actions : Expectancy-Value Models in Psychology. Hillsdale, New Jersey: Erlbaum, 1982. ----------. "Values, Valences, and Choice: The Influence ofValues on the Perceived Attractiveness and Choice of Alternatives." Journal of Personality and Social Psychology 68, no. 6 (1995): 1135-1151. Festinger, L. A Theory of Cognitive Dissonance. Stanford, CA: Stanford University Press, 1957. 184

PAGE 200

Fielding, R. "Shared Leadership in the Apache Project." Communications of the ACM 42, no. 4 (1999): 42-43. Fitzgerald, B. "The Transformation of Open Source Softare." MIS Quarterly 30, no. 3 (September 2006): 587-598. Flowers, V.S., and C.L. Hughes. "Why Employees Stay." Harvard Business Review 51, no. 4 (1973): 49-60 Fornell, C., and D.F. Larcker. "Evaluating Structural Equations with Unobservable Variables and Measurement Error." Journal of Marketing Research 18 (February 1981): 39-50. Forsyth, J. "Volunteer Management Strategies: Balancing Risk and Reward." Nonprofit World 17 (1999): 40-43. Fox, S., and G. Feldman "Attention State and Critical Psychological States as Mediators Between Job Dimensions and Job Outcomes." Human Relation 41 (1988): 229-245. Francies,G.R. "The Volunteer Needs Profile: A Tool for Reducing Turnover." The Journal of Volunteer Administration 1 (1983): 17-33. Gacek, C., and B. Arief. "The Many Meanings of Open Source ." IEEE Software 21, no. 1 (2004) : 34-40. Galindo-Kuhn, R., and R.M. Guzley. "The Volunteer Satisfaction Index: Construct Definition, Measurement, Development, and Validation." Journal of Social Service Research 28, no. 1 (2001): 45-68. Gamm, L., and C. Kassab. "Productivity Assessment of Volunteer Programs in Not-for-Profit Human Services Organizations." Journal of Voluntary Action Research 12, no. 3 (1983): 23-38. 185

PAGE 201

Gefen, D., D.W. Straub, and M. Boudreau. "Structural Equation Modeling and Regression: Guidelines for Research Practice." Communications of the AIS 4, no. 7 (2000): 1-77. George, J., and G. Jones. Understanding and Managing Organizational Behavior. Reading, Massachusetts: Addison-Wesley, 1996. ----------,and----------. Understanding and Managing Organizational Behavior, 2nd ed. New York: Addison-Wesley, 1999. Gerbing, D.W., and J.C. Anderson "An Updated Paradigm for Scale Developmennt Incorporating Unidimensionality and Its Assessment." Journal of Marketing Research 25 (May 1988) : 186-192. Ghosh, R.A., R. Glott, B. Krieger, and G. Robles. "Survey of Developers." Free/Libre and Open Source Software: Survey and Study, FLOSS Final Report, International Institute of lnfonomics, University of Maastricht, The Netherlands, June 2002. http://www. infonomics.nl/FLOSS/report/Final4 htm. (accessed July 16, 2004). Gidron, B. "Volunteer Work and Its Rewards." Volunteer Administration 11 ( 1978): 18-32. ---------. "Source of Job Satisfaction among Service Volunteers." Journal of Voluntary Action Research 12 (1983): 20-35 ----------."Predictors of Retention and Turnover among Service Volunteer Workers." Journal ofSocial Service Research 8, no. 1 (1984): 1-16. Gillespie, D.F., and A.E .O. King I. "Demographic Understanding of Volunteerism." Journal of Sociology and Social Welfare 12 (1985): 789-816. 186

PAGE 202

GNU Project, 2000b, "What is copyleft?" http :/ /www.gnu ai.mit.edu/copyleft/copyleft html (accessed April 12, 2004). Goldstein, D.K. "An Examination of Work-Related Correlates of Job Satisfaction in Programmer/Analysts." MIS Quarterly 8, no. 2 (1984) : 103-115. Griffeth, R.W., P .W. Hom, and S .A. Gaertner. "A Meta-Analysis of Antecedents and Correlates of Employee Turnover: Update, Moderator Tests, and Research Implications for the Next Millennium." Journal of Management 26, no 3 (2000) : 463-488. Grube,J.A and J.A. Piliavin. (2000). "Role Identity, Organizational Experiences, and Volunteer Performance." Personality and Social Psychology Bulletin 26 (2000) : 1108-1119. Guimaraes, T and M. lgbaria. "Determinants of Turnover Intentions: Comparing IC and IS Personnel." Information Systems Research 3, no. 3 (September 1982): 273-303. Hackett, G., and N.E. Betz. "A Self-Efficacy Approach to the Career Development of Women." Journal of Vocational Behavior 18 (1981): 326-339 Hackman, J.R and G.R. Oldham. "Development of the Job Diagnostic Survey." Journal of Applied Psychology 60, no. 2 (1975): 159-170. ----------,and----------. Work Redesign Reading, Massachusetts: Addison-Wesley 1980. Hair, J.F. Jr., R.E. Anderson, R.L. Tatham, and W.C. Black. Multivate Data Analysis with Readings 5th ed, Englewood Cliffs, NJ: Prentice Hall, 1998. 187

PAGE 203

Hall, D.T., and Schneider, B. "Correlations of Organizational Identification as a Function of Career Pattern and Organizational Type." Administrative Science Quarterly 17 (1972): 340-350. Hann, I.H., J. Roberts, S. Slaughter, and R. Fielding. "Why Do Developers Contribute to Open Source Projects? First Evidence of Economic Incentives." Proceedings of the International Conference on Software Engineering (ICSE '02), ACM, Orlando, Florida, 2002. Harhoff, D., J. Henkel, and E. von Hippel. "Profiting from Voluntary Information Spillovers : How Users Benefit from Freely Revealing their Innovations." MIT Sloan Management, Working Paper, 2003. Harrison, D.A. "Volunteer Motivation and Attendance Decisions: Competitive Theory Testing in Multiple Samples from a Homeless Shelter." Journal of Applied Psychology 80, no 3 (1995): 371-385. Hars, A., and S. Ou. "Working for Free?-Motivations of Participating in Open Source Projects." International Journal of Electronic Commerce 6, no. 3 (2002): 25-39. Hertel, G., S. Niedner, and S. Hermann. "Motivation of Software Developers in the Open Source Projects: An Internet-Based Survey of Contributors to the Linux Kernel." Research Policy 32, no. 7 (2003): 1159-1177 Hildreth, P., C. Kimble, and P. Wright. "Communities of Practice in the Distributed International Environment." Journal of Knowledge Management 4, no 1 (2000): 27-37. Holzberg, J.D., H. Gewirtz, and E. Ebner. "Changes in Moral Judgment and Self-Acceptance in College Students as a Function of Companionship with Hospitalized Mental Patients." Journal of Consulting Psychology 28 (1964): 299-303 188

PAGE 204

Houle B.J., B.J. Sagarin, and M.F. Kaplan. "A Functional Approach to Volunteerism: Do Volunteer Motives Predict Task Preference?" Basic and Applied Social Psychology 27, no. 4 (2005) : 337-344 Hu, L.T., and P.M. Bentler. "Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria versus New Alternatives." Structural Equation Modeling : A Multidisciplinary Journal 6 (1999): 1-55. Iaffaldano, M.T., and P.M. Muchinsky. "Job Satisfaction and Job Performance: A Meta-Analysis ." Psychological Bulletin 97 (1985): 251-273. Igbaria, M J.H. Greenhaus, and S. Parasuraman "Career Orientations of MIS Employees: An Empirical Analysis." MIS Quarterly 15, no. 2 (1991): 151-169. ----------,and----------. "Determinants of MIS Employees' Turnover Intentions: A Structural Equation Model." Communication of the ACM 35, no. 2 (1992): 34-49. ----------,and T. Guimaraes. "Antecedents and Consequences of Job Satisfaction among Information Center Employees." Journal of Management Information Systems 9, no 4 (1993): 145-174. ligen, D.R. "Satisfaction with Performance as a Function of the Initial Level of Expected Performance and the Deviation from Expectations." Organizational Behavior and Human Performance 6 (January 1971 ): 345-361 Jamison, I.B. "Turnover and Retention among Volunteers in Human Service Agencies." Review of Public Personnel Administration 23, no. 2 (June 2003): 114-132. 189

PAGE 205

Jenner, J .R. "Participation, Leadership, and the Role ofVolunteerism among Selected Women Volunteers." Journal of Voluntary Action Research II (1982): 27-38. Johnson, M. Voluntary Social Services Oxford : Basil Blackwell and Martin Robertson, l98I. Johnson, W.L., A.M Johnson, and F. Heimberg "A Primary and Second Order Component Analysis of the Organizational Identification Questionnaire." Educational and Psychological Measurement 59, no. I (1999): 159-170. Johnson J P "Open Source Software: Private Provision of a Public Good Journal of Economics and Management Strategy II, no. 4 (2002): 637-662. Kanfer, R "Motivation Theory and Industrial and Organizational Psychology In Handbook of Industrial and Organizational Psychology 2, edited by M D. Dunnette and L. Hough, 75-ISO Palo Alto, CA: Consulting Psychology Press, I990. Kaplan, D. "The Impact of Specification Error on the Estimation, Testing, and Improvement of Structural Equation Models." Multivariate Behavioral Research 23 (1988): 42-57. ----------."A Study of the Sampling Variability and Z-Values of Parameter Estimates from Misspecified Structural Equation Models." Multivariate Behavioral Research 24 (I989): 42-57. Katz, D. "The Functional Approach to the Study of Attitudes." Public Opinion Quarterly 24 (1960): I63-204. Katzell, R.A., and D .E. Thompson. Work Motivation: Theory and Practice." American Psychologist 45 (1990) : 144-I53 190

PAGE 206

Keams, G.S., and A.L. Lederer. "The Impact oflndustry Contextual Factors on IT Focus and the Use ofiT for Competitive Advantage." Information & Management 41, no 7 (2004): 899-919. Kemper T.D. "Altruism and Voluntary Action." In Participation in Social and Political Activities, edited by D.H. Smith and J. Macaulay et al., 306-338. San Francisco, CA: Jossey-Bass Publishers, 1980 Kiesler, C.A., ed. The Psychology of Commitment: Experiments Linking Behavior to Belief New York: Academic Press, 1971. King, M., O.L. Walder, and S. Pavey. "Personality Change as a Function of Volunteer Experience in a Psychiatric Hospital." Journal of Consulting and Clinical Psychology 35 (1970): 423-425. Kline, R.B Principles and Practices of Structural Equation Modeling. 2nd ed. New York : The Guilford Press, 2005. Kramer, M., and Schmalenberg, C. "Job Satisfaction and Retention: Insights for the '90s Part 2 Nursing 21, no. 4 (April 1991): 51-55. Krebs, D.L. "Empathy and Altruism." Journal of Personality and Social Psychology 32, no 6 (1975): 1132-1146 Lakhani, K R., and R.G Wolf. "Why Hackers do What They do : Understanding Motivation Effort in Free/Open Source Software Projects." MIT Sloan Management, Working Paper 4425-03, Sept 2003. ----------,and E. von Hippel. "How Open Source Software Work: Free User-to-User Assistance." Research Policy 32, no 6 (2003): 923-943. Latane, B., and J .M. Darley. The Unresponsive Bystander: Why Doesn't He Help? New York: Appleton-Century-Crofts, 1970. 191

PAGE 207

Latham, G P., and C.C Pinder. "Work Motivation Theory and Research at the Dawn of the Twenty-First Century." The Annual Review of Psychology 56 (2005): 485-516 Lave, J., and E Wenger. Situated Learning: Legitimate Peripheral Participation. Cambridge, UK: Cambridge University Press, 1991. Lee, T.W., and R.T Mowday. "Voluntarily Leaving an Organization: An Empirical Investigation of Steers and Mowday's Model ofTurnover." The Academy of Management Journal30, no 4 (December 1987): 721-743 Lerner, J., and J. Tirole. "Some Simple Economics of Open Source." Journal of Industrial Economics 50, no. 2 (2002): 197-234. Lewin, K. Principles ofTopological Psychology New York: McGraw-Hill, 1936. Locke, E.A. "Toward a Theory of Task Motivation and Incentives." Organizational Behavior and Human Performance 3 ( 1968): 157-189. ----------. "What is Job Satisfaction?" Organizational Behavior and Human Performance 4 (November 1969): 309-336. ----------."The Nature and Causes of Job Satisfaction." In Handbook of Industrial and Organizational Psychology, edited by M.D. Dunnette, 1297-1349. Chicago, IL: Rand McNally, 1976. ----------,and G .P. Latham "Work Motivation and Satisfaction: Light at the End of the Tunnel." Psychological Science 1, no. 4 (1990) : 240-246 Locker, D., and D. Dunt. "Theoretical and Methodological Issues in Sociological Studies of Consumer Satisfaction with Medical Care." Social Science and Medicine 12 (July 1978): 283-192. 192

PAGE 208

Logue, A.C "Training Volunteers." T+D Magazine 55, no. 6 (2001):62-69. Long, J.S Confirmatory Factor Analysis: A Preface to LISREL. Beverley Hills, California: Sage Publications, Inc., 1983. Lusch, R.F., and R.R. Serpkenci. "Personal Differences, Job Tension, Job Outcomes, and Store Performance : A Study of Retail Store Managers." Journal of Marketing 54 (January 1990): 85-101. Lynch, R. "Volunteer Retention and Feelings of Connection." e-Volunteerism: The Electronic Journal of the Volunteer Community 1, no. 1 (2000): Electronic version Lynd-Stevenson, R.M. "Expectancy-Value Theory and Predicting Future Employment Status in the Young Unemployed." Journal of Occupational and Organizational Psychology 72 ( 1999): 1 01-106. MacCallum, R.C., M. W. Browne, and H.M. Sugawara. "Power Analysis and Determination of Sample Size for Covariance Structure Modeling." Psychological Methods 1 (1996): 130-149. ----------,and J.T. Austin "Applications of Structural Equation Modeling in Psychological Research." Annual Review of Psychology 51 (2000): 201-226. ----------."Working with Imperfect Models." Multivariate Behavioral Research 38 (2003): 113-139. Mak, B L., and H. Sockel. "A Confirmatory Factor Analysis of IS Employee Motivation and Retention Information and Management 38, no. 5 (2001): 265-276. 193

PAGE 209

Mandys, F., C.V. Dolan, and P.C.M. Molenaar. "Tow Aspects of the Simplex Model: Goodness of Fit to Linear Growth Curve Structures and the Analysis of Mean Trends." Journal of Educational and Behavioral Statistics 19 (1994): 201-215. Mankiw, G. N Principles of Economics. 3rd ed. Mason, Ohio: Thomson South-Western Press, 2004. Maslow, A.H. Motivation and Personality. 2nd ed. New York: Harper Press, 1970. Masum, H. "Reputation Layers for Open Source Development. In Making Sense of the Bazzar: I'' Workshop on Open Source Software Engineering, Proceedings of the International Conference on Software Engineering (ICSE '01), Toronto, Ontario, Canada, 2001. Mathieson, K. "Predicting User Intentions: Comparing the Technology Acceptance Model with the Theory of Planned Behavior." Iriformation Systems Research 2, no. 3 (1991): 173-191. Matsueda, R.L., and W.T. Bielby. "Statistical Power in Covariance Structure Models." In Sociological Methodology, edited by N.B. Tuma, 120-158. San Francisco, California : Jossey-Bass, 1986. Mauro, R. "Understanding L.O.V E (LeftOut Variables Effor): A Method for Estimating Effects of Omitted Variables." Psychological Bulletin 108 (1990): 314-329. Mauss, M., The Gift The Form and the Reason for Exchange in Archaic Societies London, UK: Routleddge, 1959 McCurley, S., and R Lynch. Volunteer Management: Mobilizing all the Resources of the Community Downers Grove, IL: Heritage Arts Publishing, 1996. 194

PAGE 210

McEvoy, G.M., and W.F. Cascio. "Do Good or Poor Performers Leave? A Meta-Analysis ofthe Relationship between Performance and Turnover." Academy of Management Journal30 (1987): 744-762. McSweeney, P., and D. Alexander. Managing Volunteers Effectively. Brookfield, VT: Ashgate, 1996. Meehl, P.E., and N.G. Waller. "The Path Analysis Controversy: A New Statistical Approach to Strong Appraisal of Verisimilitude." Psychological Methods 7 (2002): 283-300. Michaels, C.E., and P .E. Spector "Causes of Employee Turnover: A Test of the Mobley, Griffeth, Hand, and Meglino Model." Journal of Applied Psychology 67 (1982): 53-59. Miles, I., W. Sullivan, and E. Kuo. "Ecological Restoration Volunteers: the Benefits of Participation." Urban Economics 2 (1998): 27-41. Miller, H.E., R. Katerberg, and C.L. Hulin. "Evaluation of the Mobley, Homer, and Hollingsworth Model of Employee Turnover." Journal of Applied Psychology 64 ( 1979) : 509-517. Miller, L.E. "Understanding the Motivation of Volunteers: An Examination of Personality Differences and Characteristics of Volunteers' Paid Employment." Journal of Voluntary Action Research 14 (1985): 112-122. ----------, G.N. Powell, and J. Seltzer "Determinants of Turnover among Volunteers." Human Relations 43, no. 9 (1990): 901-917. Miller, K.L., and P.R. Monge. "Participation, Satisfaction, and Productivity: A Meta-Analytic Review." Academy of Management Journal29 (1986): 727-753. 195

PAGE 211

Mirvis, P.H., and J.E. Hackett. "Work and Work Force Characteristics in the Non-Profit Sector." Monthly Labor Review 106, no. 4 (1983): 3-12. Mitchell, T .R. "Expectancy-Value Models in Organizational Psychology." In Expectations and Actions, edited by N.T Feather, 293-311. Hillsdale, NJ: Erlbaum, 1980. ----------,and D. Daniels. "Motivation." In Handbook of Psychology 12, edited by C. Borman, D.R. Ilgen, and R .J. Klimoski, 225-254. New York: Wiley, 2003. Mobley, W.H., and E .A. Locke. "The Relationship of Value Importance to Satisfaction." Organizational Behavior and Human Performance 5 (1970): 463-483. ----------, S.O. Homer, and A.T. Hollingsworth. "An Evaluation of Precursors of Hospital Employee Turnover Journal of Applied Psychology 63 (1978) : 408-414. ----------, R.W. Griffeth, H.H. Hand, and B.M. Meglino. "Review and Conceptual Analysis of the Employee Turnover Process." Psychological Bulletin 86, no. 3 (1979): 493-522. ---------. Employee Turnover: Causes, Consequences, and Control. Reading, Massachusetts: Addison-Wesley, 1982. Mockus, A., R.T. Fielding, and J. Herbsleb. "A Case Study of Open Source Software Development: The Apache Server." Proceedings of2000 International Conference on Software Engineering (ICSE '00), 263-272, Limerick, Ireland, 2000. Moon, J.Y., and L. Sproull. "Essence ofDistributed Work: The Case ofLinux Kernel." First Monday 5, no. 11 (November 2000): 381-404. 196

PAGE 212

Mowday, R.T. L.W Porter, and R.M Steers Employe e -Organization Linkages : The Psychology of Commitment and Absenteeism and Turnover. New York: Academic Press, 1982. Mowen, J.C. The 3M Mod e l of Motivation and Personality: Theory and Empirical Applications to Consumer Behavior Boston: Kluwer Academic 2000. ----------, and H. Sujan "Volunteer Behavior : A Hierarchical Model Approach for Investigating Its Trait and Functional Motive Antecedents." Journal of Consumer Psychology 15, no. 2 (2005) : 170-182. Murk, P.J., and F.J. Stephan. "Volunteers: How to Get Them Train Them and Keep Them. Economic Development Review 9 no. 3 (1991): 73-75. Mustonen M Copy left-the Economics of Linux and Other Open Source Software." Information Economics and Polic y 15 (2003): 99-121. Mutchler J.E., J .A Burr, and F .G. Caro "From Paid Worker to Volunteer: Leaving the Paid Workforce and Volunteering in Later Life." Social Forces 81, no 4 (June 2003) : 1267-1293. Nakakoji K., Y. Yamamoto, Y. Nishinaka, K. Kishida, andY. Ye "Evolution Patterns of Open Source Software Systems and Communities." Proceedings of International Workshop on Principles of Software Evolution (IWPSE), 76-85, Orlando, FL USA 2002 Naylor, H. Volunteers Toda y : Finding, Training and Working with Them. New York, NY: Association Press, 1967. Newton, R.R. and K.E Rudestam. Your Statistical Consultan t : Answers to Your Data Analysis Questions. Thousand Oaks, California : Sage Publications Inc., 1999. 197

PAGE 213

Nunnally, J C Psychometric Theory. New York: McGraw-Hill 1967. ----------. Psychometric Theory. 2"d ed. New York: McGraw-Hill, 1978. ----------,and I.H Bernstein. Psychometric Theory. 3rd ed. New York: McGraw-Hill, 1994 Oettingen, G "Expectancy Effects on Behavior Depend on Self-Regulatory Thought." Social Cognition 18, no 2 (2000): 101-129. Oliver, R.L. "A Cognitive Model for the Antecedents and Consequences of Satisfaction." Journal of Marketing Research 17 ( 1980) : 460-469. ----------. "Cognitive, Affective, and Attribute Bases ofthe Satisfaction Response," Journal of Consumer Research 20 (December 1993): 418-430. Olson, J.M., N.J. Roese, and M.P Zanna. "Expectancies." In Social Ps y chology: Handbook of Basic Principles, edited by E.T. Higgins and A.W. Kuglanski, 211-238 New York: Guilford Press 1996. Omoto, A M and M Snyder "Sustained Helping without Obligation: Motivation, Longevity of Service, and Perceived Attitude Change among AIDS Volunteers Journal of Personality and Social Psychology 68 (1995): 671-686. O'Reilly, C. A., III, and D.F Caldwell. "The Commitment and Job Tenure of New Employees: Some Evidence ofPost-Decisional Justification Administrative Science Quarterly 26 (1981): 597-616. ----------,and J. Chatman. "Organizational Commitment and Psychological Attachment: The Effects of Compliance, Identification, and Internalization on Pro social Behavior." Journal of Applied Psychology 71, no. 3 (1986), 492-499. 198

PAGE 214

Organ, D. W. Organizational Citizenship Behavior: The Good Soldier Syndrome. Lexington, Massachusetts: Lexington Books, 1987. Ozinga, J. Altruism. Westport, Connecticut: Praeger Publishers, 1999. Pearce, J.L. "Job Attitude and Motivation Differences between Volunteers and Employees from Comparable Organizations Journal of Applied Psychology 68 (1983): 646-652. ----------. Volunteers: The Organizational Behavior of Unpaid Workers. London, UK: Routledge, 1993 Peter, J.P. "Reliability: A Review of Psychometric Basic and Recent Marketing Practices." Journal of Marketing Research 16 (February 1979): 6-17. Pfeffer, J. Organizations and Organization Theories. Marshfield, MA: Pittman, 1982. Phillips, M.H. "Motivation and Expectation in Successful Volunteerism." Journal of Voluntary Action Research 11 (1982): 118-125 Piliavin, J.A., J.F Dovidio, S.L. Gaertner, and R.D. Clark. Emergency Intervention. Orlando, FL: Academic Press, 1981. ----------, E.D. Evans, and P. Callero. "Learning to 'Give to Un-named Strangers': The Process of Commitment to Regular Blood Donation." In Development and Maintenance of Prosocial Behavior: International Perspectives on Positive Morality, edited by E Staub, D. Bar-Tar, J. Karylowski, and J. Reykowski, 471-492. New York: Plenum, 1984. ----------,and H. Chamg. "Altruism: A Review of Recent Theory and Research." Annual Review of Sociology 16 (1990): 27-65. 199

PAGE 215

Porter, L.W., R.M. Steers, R.T. Mowday, and P.V. Boulian. "Organizational Commitment, Job Satisfaction and Turnover among Psychiatric Technicians." Journal of Applied Psychology 59, no. 5 (1974): 603-609. ----------, W.J. Crampon, and E.J. Smith. "Organizational Commitment and Managerial Turnover: A Longitudinal Study Organizational Behavior and Human Performance 15, no.l (1976): 87-98. Price, J .L. The Study of Turnover. Ames: Iowa State University Press, 1977. Qureshi, H B. Davies, and D. Challis. "Motivations and Rewards of Volunteers and Informal Care Givers ." Journal ofVoluntaryAction Research 8, no. 1-2 (1979): 47-55. Ralston, R., and S. Rhoden. "The Motivations and Expectations of Volunteers on Cycle Trails: The Case of the National Cycle Network, UK." Tourism and Hospitality Planning & Development 2, no. 2 (August 2005): 101-114. Rasch, R.H., and H.L. Tosi. "Factors Affecting Software Developers' Performance: An Integrated Approach ." MIS Quarterly 16 (1992): 395-413. Raykov, T. "On Sensitivity of Structural Equation Modeling to Latent Variable Misspecification." Structural Equation Modeling 7 (2000): 596-607 ----------,and Marcoulides, G.A. A First Course in Structural Equation Modeling. Mahwah, NJ: Lawrence Erlbaum Associates, Inc., 2000. Raymond, E.S. The Cathedral and the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary. Sebastopol, CA: O'Reilly & Associates, Inc., 2001. 200

PAGE 216

Razzak, S.M. "Volunteer Management: A Key Issue for Volunteer Organization." Volunteer 6, no. 6 (December 2001), http:/ / cultural forum virtualactivism.net!V olunteer-6. pdf (accessed April 18, 2007). Reichardt, C.S "The Priority of Just-identified Recursive Models." Psychological Methods 7 (2002): 307-315. Reilly, R.R. "Motivating and Retaining Technical Employees." Stevens Alliance for Technology Management 9, no. 3 (2005): 1-3. Rheingold, H. Virtual Community: Finding Connection in a Computerized World. London, UK: Minerva!, 1994. ----------. The Virtual Community: Homesteading on the Electronic Frontier. Cambridge, Massachusetts: MIT Press, 2000. Robbins, S.P. Organizational Behavior, 81h ed. Upper Saddle River, NJ: Prentice Hall 1998 Roberts, J.A., I.H. Hann, and S.A. Slaughter. "Understanding the Motivations, Participation and Performance of Open Source Software Developers: A Longitudinal Study ofthe Apache Project." Management Science 52, no. 7 (July, 2006) : 984-999. Robinowitz, S., D. Hall and J. Goodale "Job Scope and Individual Differences as Predictors of Job Involvement." Academy of Management Journal20 (1977): 273-281. Ross-Ackerman, S. "Bribes and Gifts." In Economics, Values and Organization, edited by A. Ben-Ner and L. Putterman, 296-328. New York, NY: Cambridge University Press, 1998. 201

PAGE 217

Rotolo, T., and J. Wilson. "Employment Sector and Volunteering: The Contribution ofNonprofit and Public Sector Workers to the Volunteer Labor Force." The Sociological Quarterly 47, no. 1 (2006): 21-40. Rotter, J.B. Social Learning and Clinical Psychology. Eaglewood Cliffs, NJ: Prentice Hall 1954. Ryan, R.M., and E .L. Deci. "Intrinsic and Extrinsic Motivations: Classic Definitions and New Directions." Contemporary Educational Psychology 25 (2000): 54-67. Salancik, G.R. "Commitment and Control of Organizational Behavior and Belief." In New Directions in Organizational Behavior, edited by B Staw and G .R. Salancik, 1-54. Chicago, Illinois : St. Clair Press, 1977. ----------,and J. Pfeffer. "A Social Information Processing Approach to Job Attitudes and Task Design." Administrative Science Quarterly 23, no. 2 ( 1978): 224-23 5. Scarpello, V., and J.P Campbell. "Job Satisfaction: Are All the Parts There?" Personnel Psychology 36, no. 3 (1983) : 577-600 Schmidt, K.M., and M. Schnitzer. "Public Subsidies for Open Source? Some Economic Policy Issue of the Software Market." Harvard Journal of Law and Technology 16, no. 2 (2003): 473-506. Schram, V.R. "Motivating Volunteers to Participate." In Motivating Volunteers : How the Rewards of Unpaid Work Can Meet People's Needs, edited by L.F. Moore, 13-29. Vancouver B.C.: Vancouver Volunteer Centre, 1985. Schroeder, D.A., L.A. Penner, J.F. Dovidio, and J.A. Piliavin. The Psychology of Helping and Altruism: Problems and Puzzles. New York: McGraw-Hill, 1995. 202

PAGE 218

Schwartz, S.H. "Moral Decision Making and Behavior." In Altruism and Helping Behavior, edited by J. Macauley and L. Berkowitz, 127-141. New York: Academic Press, 1970. Segars, A.H., and V. Grover "Re-Examining Perceived Ease of Use and Usefulness: A Confirmatory Factor Analysis MIS Quarterly 17, no 4 (December 1993): 517-525. ----------."Assessing the Unidimensionality of Measurement: A Paradigm and Illustration within the Context of Information Systems Research." Omega 25, no. I (1997) : 107-121. Sharp, E.G. "Citizen Organization in Policing Issues and Crime Prevention: Incentives for Participation." Journal of Voluntary Action Research 7 (1978): 45-58. Shaver, K.G., E.J. Gatewood, and W.B. Gartner. "Differing Expectations: Comparing Nascent Entrepreneurs to Non-Entrepreneurs." Paper presented at the Academy of Management, Washington, D.C., 2001 Sheldon, K M., and A.J. Elliot. "Not All Personal Goals are Personal: Comparing Autonomous and Controlled Reasons for Goals as Predictors of Effort and Attainment." Personality and Social Psychology Bulletin 24, no. 5 (1997): 94-105. Sherman, J.D., and H.L. Smith. "The Influence of Organizational Structure on Intrinsic versus Extrinsic Motivation." Academy of Management Journal 27 (1984): 877-885. Shrauger, J.S. "Responses to Evaluation as a Function oflnitia1 Self-Perceptions." Psychological Bulletin 82 (July 1975): 581-596. 203

PAGE 219

Singer, M.S., and T K Coffin "Cognitive and Volitional Determinants of Job Attitudes in a Voluntary Organization." Journal of Social Behavior and Personality 11, no. 2 (1996): 313-328. Skoglund, A.G. "Do Not Forget about Your Volunteers: A Qualitative Analysis of Factors Influencing Volunteer Turnover Health & Social Work31, no. 3 (August, 2006):217-220. Smith, M. J. Bruner, and R. White. Opinions and Personality New York: Wiley, 1956. Smith, P C., L.M. Kendall and C .L. Hulin. The Measurement of Satisfaction in Work and Retirement : A Strategy for the Study of Attitudes. Chicago, IL: Rand McNally, 1969. Smith, D H "Altruism, Volunteers, and Volunteerism." Journal of Voluntary Action Research I 0, no I ( 1981 ): 21-36 Snarr, C.E and Krochalk, P .C. "Job Satisfaction and Organizational Characteristics: Results of a Nationwide Survey of Baccalaureate Nursing Faculty in the United States." Journal of Advanced Nursing 24, no 2 (1996): 405-412 Snyder, M "Basic Research and Practical Problems : The Promise of a 'Functional' Personality and Social Psychology." Personality and Social Ps y chology Bulletin 19 (1993): 251-264. Spector, A.J. Expectations Fulfillment and Morale." Journal of Abnormal and Social Psychology 52 (January 1956) : 51-56. Spreng, R.A., S.B. MacKenzie, and R.W Olshavsky. "A Reexamination of the Determinants of Consumer Satisfaction." Journal of Marketing 60, no. 3 (1996): 15-32. 204

PAGE 220

Stallman, R M. GNU Emacs Manuall9, no. 33, Boston, MA: Free Software Foundation, 1996. Starns, B.J., and W.W. Wymer, Jr. "Conceptual Foundations and Practical Guidelines for Retaining Volunteers to Serve in Local Nonprofit Organizations." Journal of Nonprofit & Public Sector Marketing 9, no. 112 (2001): 97-118. Staub, E. Positive Social Behavior and Morality: Vol. 1 Social and Personal Influences New York: Academic Press, 1978. Staw, R.M. "The Consequences of Turnover." Journal of Occupational Behavior 1 ( 1980): 253-273. Steers, R.M "Antecedents and Outcomes of Organizational Commitment." Administrative Science Quarterly 22, no. 1 (1977): 45-56. ----------,and R.T. Mowday. "Employee Turnover and Post-Decision Accommodation Processes." In Research in Organizational Behavior 3, edited by B.M. Staw and L.L. Cummings Greenwich, Connecticut: JAI Press, 1981. Stevens, J.M., M.J. Beyer, and M.H. Trice. "Assessing Personal, Role and Organizational Predictors of Managerial Commitment." Academy of Management Journa/21 (1978): 380-396. Straub, D.W "Validating Instruments in MIS Research." MIS Quarterly 13, no. 2 (1989): 147-169. Susarla, A A. Barua, and A.B. Whinston. "Understanding the Service Component of Application Service Provision: An Empirical Analysis of Satisfaction with ASP Services." MIS Quarterly 27, no. 1 (2003): 91-123. 205

PAGE 221

Thatcher, J.B L.P. Stepina, and R.J. Boyle. "Turnover oflnformation Technology Workers: Examining Empirically the Influence of Attitudes, Job Characteristics, and External Markets." Journal of Management Information Systems I9, no. 3 (Winter 2002-3): 23I-261. Tomarken, A.J., and N.G. Waller. "Potential Problems with Well Fitting Models." Journal of Abnormal Psychology 112, no. 4 (2003): 578-598 ----------,and----------. "Structural Equation Modeling: Strengths, Limitations, and Misconceptions." Annual Review of Clinical Psychology I (2005): 31-65. Tschirhart, M., D.J. Mesch, J.L. Perry, T.K. Miller, and G. Lee. "Stipended Volunteers : Their Goals, Experiences, Satisfaction, and Likelihood of Future Service." Nonprofit and Voluntary Sector Quarterly 30, no. 3 (200 I): 422-443. Ullman, J.B. "Structural Equation Modeling." In Using Multivariate Statistics, 3rd ed, edited by B .G. Tabachnick and L.S. Fidell, 709-8I9. New York: HarperCollins College Publishers, I996. Vinokur-Kaplan, D., S. Jayaratne, S., and A .W. Chess. "Job Satisfaction and Retention of Social Workers in Public Agencies, Non-Profit Agencies, and Private Practice: The Impact of Workplace Conditions and Motivators." Administration in Social Work I8, no. 3 ( I994 ): 93-I21. von Hippel, E. The Sources of Innovation. New York, NY : Oxford University Press, 1988. ----------,and G. von Krogh. "Open Source Software and the 'Private--Collective' Innovation Model: Issues for Organization Science." Organization Science I4, no. 2 (March-April2003): 209-223. Vroom, V.H. Work and Motivation. New York: Wiley and Sons, 1964. 206

PAGE 222

----------. Work and Motivation San Francisco, California: Jossey Bass, 1995. Walker, O.C., G .A. Churchill, and N.M. Ford. "Motivation and Performance in Industrial Selling: Present Knowledge and Needed Research." Journal of Marketing 14 (May 1977) : 156-168. Ward, E.A. "Relation of Job Satisfaction and Job Knowledge and Their Effect on Intention to Turnover." Psychological Reports 63, no. 2 (1988): 611-615. Wasko, M.M., and S. Faraj. "It Is What One Does: Why People Participate and Help Others in Electronic Communities of Practice." Journal of Strategic Information Systems 9, no. 2-3 (2000): 155-173. ----------,and---------. "Why Should I Share? Examining Social Capital and Knowledge Contribution in Electronic Networks of Practice." MIS Quarterly 29, no I (2005): 35-57. Watts, W.A. "Predictability and Pleasure: Reactions to the Disconfirmation of Expectancies." In Theories ofCognitive Consistency: A Sourcebook, edited by R.P. Abelson, E. Aronson, W.J. McGuire, T.M. Newcomb, M.J Rosenberg, and P.H. Tannenbaum, Chicago, IL: Rand-McNally, 1968. Watts, L.R., and H.C White. "Assessing Employee Turnover." Pers onnel Administrator 33, no 4 (1988): 80-85 Weaver, D., and P. Brickman. "Expectancy, Feedback and Disconfirmation as Independent Factors in Outcome Satisfaction." Journal of Personality and Social Psychology 20 (March 1974): 420-428. Wenger E C Communities of Practice : Learning, Meaning, and Identity Cambridge University Press, 1998. 207

PAGE 223

----------,and W.M. Snyder. "Communities of Practice: The Organizational Frontier." Harvard Business Review 78, no. 1 (2000): 139-145. Werner, R. Personality Characteristics of Citizen Volunteers in Relations to their Level of Human Relations Skills. Unpublished Doctoral Dissertation, Boston College, 1976 Werts, C.E., R.L. Linn, and K.G. Joreskog. "Interclass Reliability Estimates: Testing Structural Assumptions." Educational and Psychological Measurement 34, no 1 (1974): 25-33. Williams, L.J., and J T. Hazer. "Antecedents and Consequences of Satisfaction and Commitment in Turnover Models: A Reanalysis Using Latent Variable Structural Equation Methods." Journal of Applied Psychology 71, no. 2 (1986): 219-231. Wu, C.G J.H. Gerlach, and C.E. Young. "An Empirical Analysis of Open Source Software Developers' Motivations and Continuance Intentions." Information & Management 44, no. 3 (April 2007): 253-262 Ye, Y., and K. Kishida. "Toward an Understanding of the Motivation of Open Source Software Developers." Proceeding of the 251h International Conference on Software Engineering (ICSE), IEEE Press, 2003. 208