Gender and management

Material Information

Gender and management men, women, and decision-making in public organizations
Alternate title:
Men, women and decision-making in public organizations
King, Cheryl Simrell
Publication Date:
Physical Description:
ix, 130 leaves : forms ; 29 cm


Subjects / Keywords:
Women government executives ( lcsh )
Women government executives -- Colorado ( lcsh )
Sex discrimination in employment ( lcsh )
Sex role in the work environment ( lcsh )
Sex discrimination in employment ( fast )
Sex role in the work environment ( fast )
Women government executives ( fast )
Colorado ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references.
General Note:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Public Administration.
General Note:
School of Public Affairs
Statement of Responsibility:
by Cheryl Simrell King.

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:
28246848 ( OCLC )
LD1190.P86 1992d .K56 ( lcc )

Full Text
Cheryl Simrell King
B.A., University of Texas, Permian Basin, 1981
M.A., University of Colorado at Denver, 1987
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Public Administration

This thesis for the Doctor of Philosophy
degree by
Cheryl Simrell King
has been approved for the
Graduate School of
Public Affairs

King, Cheryl Simrell (Ph.D., Public Administration)
Gender and Management: Men, Women and Decision-Making in Public Organizations
Thesis directed by Associate Professor E. Sam Overman
Because of rapid changes in the workforce and the ensuing demands for
diversity in the management ranks, most organizations will soon have to account for the
lack of balance between women and men in top management positions. State
governments, particularly, will be called to account because of the-increasing desire
among citizens, leaders and academics for a representative bureaucracy that is socially
Using a sample of executive managers within the State of Colorado
administrative departments, this study investigated the relationships between sex-role
identity, decision-making style and institutional discrimination as determinants of
representation of women in upper management positions. The population (506) of
Colorado, upper level career civil service managers were asked to respond to this
study. In total, 308 managers responded; 238 male and 70 female.
The results indicate that sex-role identity and decision-making style are not
adequate determinants of representation of women at top levels of management
Furthermore, these two constructs did not measure institutional discrimination. The
variables that were most predictive were personal demographics (age, education, etc.)
as well as indications of the function of the organization. Women were most likely to
have greater representation in those departments that performed functions that have
typically been performed by women. In addition, women were more likely than men to
identify with a masculine sex-role. Further research is recommended to compare these
data with other states in an attempt to determine if there is no relationship between sex-
roles/decision styles and the representation of women in top management in state
government or if the environment in Colorado is unique.
The form and content of this abstract are approved. I recommend its publication

This is dedicated to Kate who, with the patience only a child can muster, tolerated me
during this long haul.

The people and groups that made this thesis possible are the managers of the State of
Colorado who so willingly complied with this study; the Colorado Department of
Personnel for their unfailing support in providing the population data and in reviewing
the survey instruments; the Graduate School of Public Affairs (GSPA) at CU-Denver
for fellowships, monetary support and secretarial support; the GSPA Graduate Student
Association for assistance with mailing costs; Roger Carver for data entry and
programming assistance; my thesis support group for the constant presence; Andrea
Ziegert for reviewing my first draft; and E. Sam Overman for showing me my
limitations and my capabilities.
- v -

Figures...................................................... viii
1. INTRODUCTION........................................... 1
The Nature of the Problem........................... 1
The Nature of the Current Research on the Barriers.. 6
Purpose of the Study................................. 12
Limitations of the Research.......................... 14
Validity...................................... 14
Cultural Limitations.......................... 15
Limitations with Respect to Minority
Representation................................ 15
Contribution......................................... 16
2. LITERATURE REVIEW....................................... 18
Personal Discrimination Theories..................... 18
Sex-Role Identification....................... 18
Management/Decision-making Style in Organizations.28
Structural Discrimination Theories................... 37
Research Model....................................... 40
Research Hypotheses.................................. 42
3. METHOD................................................. 44
Design............................................... 44
Variables Under Study................................ 44
Data Collection...................................... 50
Survey Instrument Scoring............................ 53
The Bern Sex-Role Inventory (BSRI)............ 53
The Decision-Style Inventory (DSI)............ 54
Data Analysis Techniques............................. 55
- vi -

4 RESULTS................................................... 57
Descriptive Findings.................................... 57
Comparing Women and Men Managers........................ 66
Comparing Organizations................................. 75
Testing the Hypotheses.................................. 78
Summary of Hypotheses Testing........................... 90
5 CONCLUSIONS............................................... 92
Interpretation of Results: Overall Snapshot of Managers
Within the State of Colorado............................ 92
Interpretations: Sex-Role Identity and Decision Style...97
Sex-Role Identity............................... 97
Decision Style.................................. 100
Predicting Representation....................... 102
Alternative Rival Hypotheses............................ 103
Theoretical and Practical Implications:
Implications for Future Research........................ 106
The State of the State of Colorado.............. 107
The Composition of the Management Workforce..... 107
Dominant Management Styles or Cultural
Masculinity..................................... 108
Relationship Between Decision Style and
Sex-Role Identity............................... 109
Explaining the Representation of Women.......... 110
Some Thoughts on the Future of Public
Administration in the U.S............................... 112
BIBLIOGRAPHY........................................ 125
- vu -

2.1 Research Model............................................. 40
3.1 Interpreting the BSRI..................................... 54
3.2 Range of Scores on the DSI.............................. 55
- vm -

3.1 Gender Breakdown of Top Managers in Colorado
Administrative Departments...................................47
3.2 Breakdown of Colorado Departments By Dependent
Variables................................................... 49
3.3 Response Rates.............................................. 51
3.4 Sample Representativeness................................. 52
4.1 Overall Organizational Characteristics...................... 58
4.2 Overall Individual Demographics............................ 60
4.3 Overall Sex-Role Identity.................................. 62
4.4 Overall Decision Style................................... 63
4.5 Perceptions of Gender/Race Issues in State Government....... 65
4.6 Women and Men Organizational Characteristics...............67
4.7 Women and Men Individual Demographics................. 69
4.8 Women and Men Sex Role Identity..................... 71
4.9 Women and Men Decision Style.........................72
4.10 Women and Men Perceptions of Gender/Race Issues
In State Government......................................... 74
4.11 Select Organizational Characteristics by
Representation at Top....................................... 76
4.12 Select Organizational Characteristics by
Representation Ratio........................................ 77
4.13 Sex-Role Identity by Representation at Top................. 79
4.14 Sex-Role Identity by Representation Ratio...................79
4.15 Decision Style by Representation at Top.................... 81
4.16 Decision Style by Representation Ratio..................... 82
4.17 Sex-Role Identity and Decision Style by Department......... 84
4.18 Relationship between the BSRI and the DSI.................. 85
4.19 Results of Discriminant Analysis
(excluding organizational function)......................... 88
4.20 Results of Discriminant Analysis
(including organizational function)......................... 89
- ix -

The Nature of the Problem
As the world becomes a more complex place and as organizations struggle to
adjust to the chaos and complexity in this country and in the greater global village,
organizations, both public and private, will need to examine the manner in which they
manage and lead. To be competitive in an increasingly global market, either in a public
or private sphere, organizations will be required to be adaptive, flexible and innovative
and will need skills to manage diverse environments (Kanter, 1989; Thomas, 1990).
Thomas (1990) believes that, in order to achieve the levels of flexibility needed in this
economy, organizations will have to make better use of the diverse talent of the
increasingly changing workforce. Thomas and others (Morrison & Von Glinow,
1990) believe.that organizational survival in the future will be predicated upon going
beyond the first generation of affirmative action which is characterized by a focus on
numbers in compliance to governmental regulations, to the next generation of
affirmative action where organizations recognize and appreciate diversity and where
power, influence and resources are distributed without regard to race or sex.
This research study examines some of the factors that may be keeping
organizations from moving from a model of strict affirmative action where quotas are
met at an aggregate level in compliance with regulation, to a model where individuals
are regarded for the value they bring to the organization, regardless of their belonging
to a particular identity group that has traditionally not been represented at all levels in
organizations. Specifically, this study focuses upon women in public organizations
and some of the existing barriers for women who may want to advance to executive
management positions.
- 1 -

Federal affirmative action policies are the result of the Civil Rights Act of 1964,
Title VII which "prohibits discrimination on the basis of sex, race, color, religion or
national origin in any employment condition, including hiring, firing, promotion,
transfer, compensation and admission to training programs" (Title VII Civil Rights Act
of 1964, 3701 et seq., as amended, 42 U.S.C.A. 2000e et seq.). The law authorized
affirmative action as the following:
If the court finds the respondent has intentionally engaged in an unlawful
employment practice...the court may order such affirmative action as may be
appropriate, (in Kelly, 1991, p. 14)
Under affirmative action law, organizations are required to be affirmative, or
proactive, in their employment practices regarding women and people of color. To this
end, quotas were set that required organizations to match the labor force participation
proportion of women and people of color in their organizations. Compliance to these
quotas is required for public organizations. Private organizations voluntarily comply
with the threat of adjudication, or the desire to employ a more diverse workforce, as
their motivations.
Although affirmative action has met its original objectives of increasing the
number of women in organizations, affirmative action has not been effective at all
organizational levels. The proportion or numbers of women across all levels of
organizations is a measure of block equality. The proportion of women at particular
levels of the organization is a measure of segmented equality (Frederickson, 1990,
1980). Simply put, organizations have been effective in ensuring block equality but not
as effective in ensuring segmented equality. In the 1970s and 1980s, more women
moved into the labor force and the pressures to incorporate women at all levels,
segments and strata increased significantly (Kelly, 1991). In 1988, women made up
45% of the labor force (Kelly, 1991). Equal opportunity in terms of access to the
labor force may no longer be an issue for women. Equal opportunity in regard to
access to all jobs in the labor force is still a problem for women.
One segment that is of major concern to organizations and to women is the
executive level. Women have not been entering executive offices at the same rates that
they are entering lower levels of management. Linda Keller Brown (1988) reported
- 2 -

that in 19.47, 5% of all working women were administrators and managers; in 1982,
7.4% of all working women were administrators or managers, a change of only 2.4
percentage points over thirty-five years. Brown notes that although the current
workforce is comprised of approximately 50% women, women have not made great
strides over the years in entering the areas of management and administration. Powell
(1988) states that 38% of all management, executive or administration positions in the
U.S. were held by women. However, at that time, only 2% of the top management
positions were held by women. Paradis and Sloss (1989) report that in 1988, 90% of
the chief executive officers (CEOs) in the nearly 7,000 hospitals in the U.S. were men,
10% were women. In contrast, 85% of the healthcare workforce in 1988 were women.
In the private sector, over 50% of the Fortune 1000 companies still have no women
directors. In 1989, women held 3-4% of the total Fortune 1000 directorships and 2%
of officerships, although almost 50% of the employees of these companies were
women. (Fryxell & Lemer, 1989)
In the public sector as well, women increasingly hold a greater proportion of the
total jobs but continue to have low representation at top levels. Wise (1990) reports
that although women represent 42% of the workforce of the U.S. federal civil service
(1988 data), they hold only 10% of the executive positions (Senior Executive Service).
In a study of five state governments in western United States, a team of researchers
found that women make up almost 50% of the state government workforce in each state
studied, but only hold from 13-26% of the top jobs in state government (Hale & Kelly,
1989; Kelly, Guy & Bayes, et al., 1991). According to Kelly et al. (1991), variation in
the proportion of women in top management is due, in part, to varied efforts of
governors to appoint women to top level positions.
The inequitable distribution of women in top government positions, especially
when compared to the total women in the workforce, presents a special problem for
government administration. Because government has a social or moral responsibility to
ensure representativeness, the need to "fix" the distribution problem may be more
crucial for public sector enterprises than it is for private sector enterprises. Government
is not only compelled to meet current standards of affirmative action, the very nature of
bureaucratic, democratic government demands that employees be representative of the
- 3 -

Affirmative action policies were instituted to rid the workplace of hiring biases
and to correct procedures that limited women and minorities' employment
opportunities. In addition, for government, affirmative action is a policy tool to ensure
that the administrative branch of government is representative and provides an equal
opportunity for all groups in society to influence governmental action. This
opportunity is undermined when managerial ranks are occupied by advantaged groups
while lower ranks are occupied by traditionally disadvantaged groups (Wise, 1990).
Equal opportunity to influence governmental action, or social equity, (Frederickson,
1990,1980) has been presented as a key building block of government that should be
considered in all areas of governance:
Social equity is a phrase that comprehends an array of value preferences,
organizations, design preferences and management style preferences...Social
equity emphasizes responsibility for decisions and program implementation for
public managers. Social equity emphasizes change in public management
Social equity emphasizes responsiveness to the needs of citizens rather than the
needs of public organizations. (Frederickson, 1980, p. 37)
The ideal of what government should look like is defined by issues of social equity or
equality; the reality of what government is today is limited by government not fully
embracing the basic tenets of social equity.
Frederickson in his Compound Theory of Social Equity differentiates
segmented equality from block equality and emphasizes that governments are
responsible to ensure that integration be examined within segments (segmented
equality) as well as across segments (block equality). Ensuring representativeness and
social equity across all strata of government should be as crucial an element of
governance as is providing public services.
Wise (1990) states that social equity (both block and segmented equity) is
especially important in government because government jobs offer material rewards,
some of which are unique to public sector employment that affect individual living
standards. In addition, participation in the public bureaucracy provides an opportunity
for substantive political representation (Wise, 1990, p. 567). As a result, democratic
governments should be compelled to ensure that access to all jobs at all levels in
government is equitable for all groups.
- 4 -

In defining representation, Wise (1990) differentiates between passive and
active representation. Passive representation is, in essence, reflected in current
affirmative action policies. People within the bureaucracy take a passive role by
allowing the leaders to pursue their individual interests, by relying on policy to "police"
the leaders' actions, by assuming that the leaders have everyone's best interest in mind
and letting the "chips fall where they may." Active representation, alternatively,
happens when members of the bureaucracy assume responsibility for democratic
Expanding access to leadership positions in the bureaucracy advances the
opportunity for diverse preferences to be considered in efforts to determine
what constitutes the public interest and how it can be best addressed. Equity is
not served when others attempt to represent the interests of those who are
excluded from power regardless of how effective they might be. (Wise, 1990,
The notion of social equity in public management, or a representative or
participatory bureaucracy, is not a new concept invented during the last decade to deal
with problems of equity in relation to the management of government and the
management of the services that governments provide. The concept of representative
bureaucracy, "the idea that the bureaucracy should in various ways reflect the society of
which it is apart" (Kranz, 1976, p. 67), has its roots in British Colonialism. Kingsley
(1944 in Kranz, 1976) first coined the phrase representative bureaucracy in his analysis
of the British civil service.
Historically, the desire for participation in government has evolved over the last
five decades from an initial emphasis on mirroring in the bureaucracy socio-economic
and political party distributions in the population, to the current focus of representing all
significant groups in the population, especially those groups who have, traditionally,
been disadvantaged in the decision-making process (Kranz, 1976; Krislow, 1974).
Indeed, some authors (Stivers, in press) take the notion of representative
bureaucracy one step further and claim that public administration, as a discipline and
profession, will not be capable of claiming legitimacy unless great strides are made in
increasing representation. In addition, efforts must be made to "correct" the current
arguments of legitimacy based upon competency:
- 5 -

...Defending the legitimacy of public administrators on the basis of their
competence is problematic because the image of the public
administrator...privileges masculine over feminine characteristics and depends
for its coherence on maintaining women in a position of inequality with respect
to life chances and resources. (Stivers, in press, p. 24)
Clearly there continue to be problems with representation at top levels of
management in all sectors of the population of the United States. Because of the need
for social equity in government, the need to have more women represented at top levels
of management in government is, perhaps, even more crucial than the need for equal
representation of women at all levels of other organizations. In order to meet the need
for greater representativeness, greater knowledge is needed about the causal factors that
keep women out of the top levels of management Indeed, greater knowledge is needed
about the barriers that are keeping governments from achieving acceptable levels of
social equity in regard to the representation of women at top levels of management. It
is only through the dissolution of these barriers that government can become
representative. This study was designed specifically to investigate these barriers.
The Nature of the Current Research on the Barriers
This significant lack of balance between the number of women in lower and
middle levels of management as compared to the number of women in upper levels of
management in both the public and private sectors has recently been a key focus of
management research in academia and a key focus in the popular press. Scores of
studies and articles have been published in the last five years investigating this
phenomenon. The term "the glass ceiling" was coined in 1987 as a metaphor for
women's inability to break into executive offices despite the proliferation of women at
other levels in organizations (Morrison, White & Velsor, 1987). The glass ceiling
describes a "barrier so subtle that it is transparent, yet so strong that it prevents women
and minorities from moving up in the management hierarchy" (Morrison & Von
Glinow, 1990, p. 200).
- 6 -

Generally, research investigating the problems of representation in top levels of
management yields results that can be classified under one of the following types of
theories (Morrison & Von Glinow, 1990):
Difference Theories focus on trait differences between men and women due
to socialization and developmental variables. This theoretical area emphasizes
that true or absolute "deficiencies" in underrepresented groups are largely
responsible for differential treatment in the workplace. Included in this category
is the economics-based human capital theory of labor market participation (Blau
& Ferber, 1987) as well as studies of trait differences between men and women
managers (Donnell & Hall, 1980). Much of the earlier work in the field
focused on this model to explain gender differences, although the evidence for
clear differences is inconclusive (Bern 1974; Lever, 1976; Maccoby & Jacklin,
1974; Sherman, 1976).
Individual Discrimination Theories posit that differences in representation are
the result of biases and stereotyping on the part of individuals in the dominant
group (white males) which lead organizations to hire women and minorities at a
substantial discount (Morrison & Von Glinow, 1990). Included in this
category are the economics-based labor market discrimination theories (Blau &
Ferber, 1987) and rational bias theory (Larwood, Szwajkowski & Rose, 1988).
Discrimination theories also include gender and management sex-role
stereotyping theories which hold that traditionally masculine management
behaviors are chosen as superior to traditionally feminine behaviors (Kanter,
1977b; Powell & Butterfield, 1979).
Structural Discrimination Theories seek to explain differences through
structural and systems models which conclude that the imbalance is a function
of the organization or the system within which the organization exists (Alderfer,
1986; Powell, 1988; Thurow, 1969). Included in this category is the
economics-based dual labor market theory, or occupational segregation (Kelly,
1991; Thurow, 1969), intergroup theory (Alderfer, 1986) and organizational
structure theory (Kanter, 1977a, 1977b). Essentially these theories state that
"widespread policies and practices in the social system perpetuate
- 7 -

discriminatory treatment" of underrepresented groups (Morrison & Von
Glinow, 1990, p.201).
Empirically, difference theories, which test true physiological and psychological
differences between men and women, yield inconclusive results. Studies ranging from
the classic literature review of sex differences by Maccoby and Jacklin in 1974 to more
recent research on sex differences in management (Morrison, et al., 1987; Donnell &
Hall, 1980; Powell, 1988) consistently report that there are few, if any, true differences
between men and women aside from the obvious biological determinants. Specifically,
the research indicates that there are no real differences between men and women that
would predispose one or the other to be "better" managers.
In economics, scholars have focused on the human capital theory of labor
market participation as an explanation for the differences between men and women in
the marketplace (Blau & Ferber, 1987). This theory explains the continued differences
by suggesting that individuals are rewarded differently for investment in education,
training and job experience. Essentially, people should chose the area within which
they want to work and invest in their own "human capital" in order to be competitive in
that market. This theory assumes that equal investments, regardless of whether one is
male, female, black or white, will pay-off equally (Blau & Ferber, 1987). Recent
studies have shown, however, that investment in human capital has a higher payoff for
White men than it does for any other group (Larwood, et al., 1988). According to
these studies, discrimination cannot be explained by differential investment in human
The lack of strong, consistent findings of true sex differences between men and
women and the lack of support for the human capital investment theory has led scholars
to focus on discrimination as the primary causal factor in the lack of representation of
women across all levels of management As indicated above, discrimination can occur
at both a personal and organizational level.
Personal or individual discrimination theories focus upon the role of the
individual in the discrimination process. Labor market discrimination theory assumes
that decision makers have discriminatory tastes in employing members of minority
- 8 -

groups, even when the minority is a perfect economic substitute for a member of the
dominant group. To compensate for the loss of utility, or discomfort, associated with
employing members of minority groups, decision makers hire minorities at a substantial
wage or status discount.
Rational bias theory (Larwood et al., 1988) explains discrimination as a
function of the individual within the context of his/her organization. Rational bias
theory states that discrimination makes sense when the cultural norms, or the informal
structure, of the organization supports discrimination regardless of policy or regulation.
A decision-maker chooses to discriminate when his/her behavior will be rewarded,
overtly or covertly, by his/her peers or superiors, regardless of what the formal policy
states about discrimination (Larwood et al., 1988). Rational bias theory makes
particular sense in organizations where the dominant group is fighting to maintain
control and thus resorts to behaviors that ensure the survival of that control.
Other personal or individual discrimination theories focus on sex-roles and sex-
role stereotyping in organizations. These theories posit that discrimination occurs
because of the belief by the dominant group that members of the "out-group" (Larwood
et al., 1988) are less suited for management than are the dominant group. Differential
treatment is not due to actual performance in management jobs but to perceptions or
stereotypes of how group members will perform in management jobs (Powell, 1988).
In differential treatment, the stereotypical perceptions exist even when no real
differences or no real evidence for differences exists because of the cultural and
historical foundations of sex-roles in society.
Lastly, structural discrimination theories focus upon the institutionalized
discrimination that is a function of organizational or societal norms and practices.
Intergroup theory focuses upon the tensions that result when one's identity group
membership (race, ethnicity, sex, age) is at odds with one's organizational group
membership (Alderfer, 1986). Because, generally, individuals tend to surround
themselves with like individuals, organizational groups tend to be organized around
identity groups. Extreme tension exists, for all parties involved, if one belongs to an
organizational group that is primarily composed of people who hold memberships in a
different identity group.
- 9 -

Kanter's (1977a, 1977b) organizational structure theory is quite similar to
Alderfer's (1986) identity group theory. Kanter states that membership within groups
and one's status as a minority member affects the dynamics of the group. This often
impacts the stability of the group. As a result, groups will tend to try to maintain
balance by ensuring that the membership of the group remains homogeneous (Kanter,
1977a, 1977b.)
In economics, dual/segmented market labor theory also attempts to explain
discrimination as a systemic problem (Bayes, 1988; Thurow, 1969). Dual/segmented
market labor theory provides the underpinnings for occupational segregation theory.
Essentially, this theory states that there are two labor markets, the primary market
which consists of the better jobs both in terms of status and pay, and the secondary
market which consists of the least desirable jobs. Little to no movement exists between
the two markets (Thurow, 1969). In management, the secondary jobs not only include
those at lower levels, but also those in staff (versus line) functions where women and
minorities are often the dominant group (Morrison & Von Glinow, 1990).
Dual/segmented market labor theory provides evidence for the "degradation of
competition" hypothesis which explains how primary market jobs become secondary
market jobs as soon as women and minorities become the dominant group (Bayes,
In addition to the difference and discrimination theories, some recent research
has focused on examining the effects of social, biological and cultural roles on
women's access to management. The fact that women are still primarily responsible
for activities related to the domestic sphere (Hochschild, 1989) and the fact that many
women choose to take time off of their careers to birth and raise children (the "Mommy
Track" see, for example, Schwartz, 1989) often are cited as explanations for the
relatively low representation of women in top management positions. Career track
studies indicate that the career of any individual who aspires to upper management
positions is challenging and demanding. Studies have shown that, even in today's
more egalitarian families, women still hold primary responsibility for household
activities and continue to have a difficult time managing the conflicting demands of
career and home, especially when the career is particularly demanding (Hochschild,
1989; Schwartz, 1989). As a result, many women who may desire and may be capable
- 10 -

of holding an executive level position either take time off of their career to tend to their
family needs, in turn seriously impairing their advancement, or choose to opt out of an
executive track early in their career. Schwartz (1989), in her controversial article that
spawned the discussion of "The Mommy Track" in the popular press, suggests that
many organizations are losing women who have the potential to be executive managers
because organizations do not offer career options to women who choose to combine
career and family that allow for the interruptions that result, without permanently
impairing promotional abilities.
The presence of multiple theoretical approaches to understanding the
representation problem of women in top management allow researchers in this area
many avenues from which to approach understanding this problem, However, the
majority of the current scholarly literature investigating the barriers to executive
management for women is primarily focused upon personal and structural
discrimination theories. Many scholars recently have even abandoned all difference
theory approaches to understanding discrimination in the workplace and, to correct for
an overfocus on difference theory in earlier research, have focused almost exclusively
upon a structural/systems approach:
Career-path studies have led researchers to conclude that women's careers [and
men's] are affected less by individual traits (e.g., education credentials) than by
structural variables (e.g., the percent of women managers within the
organization), which affect how women are perceived and responded to (e.g.,
the dynamics of the organizational environment, especially issues such as
access to the power structure). In the United States, the trend in research on
women and men in management has moved away from intrapsychic/individual
explanation toward a social structure paradigm that focuses on the systemic
factors inhibiting upward mobility of women managers. (Brown, 1988, p. 271)
Morrison and Von Glinow (1990) challenge researchers to take a broad
approach to the problem, moving systemic theory one step further to focus on the
interaction of situational factors (structural discrimination) with person-centered
characteristics (personal discrimination) as the main elements accounting for the
differential treatment of women and minorities in organizations. Morrison and Von
Glinow call for interactionist research in organizations using actual managers and
multiple methods so that results will reflect realistic situations.
- 11 -

Recent work investigating the glass ceiling in state government (Hale & Kelly,
1989) employed an interactionist model which included organizational, individual, and
familial variables. This ongoing research provided a formative model for the study
performed here. Hale and Kelly (1989) postulate that the three main categories of
barriers to the top for women are:
internal barriers (sex-role socialization, self-concept, role prejudice, etc.);
support availability barriers (financial resources, education/training, presence
of strong mentors, family variables, domestic responsibilities, etc.); and
structural barriers (organizational variables such as discrimination, sexual
harassment, pay inequities, etc.).
Hale and Kelly (1989) conclude that whether or not one makes it to the top is
influenced by an interaction of these factors, and cannot be explained by any one
variable or one category of variables. Hale and Kelly found that women who were
most successful were so as a result of a variety of factors including their own self
concept, their primary family origins and patterns, the amount of support available to
them both at home and at the workplace and the nature of the work environment itself.
In short, success was a function of a number of related personal, support and
organizational factors (Hale & Kelly, 1989).
Purpose of the Study
This study attempts to take the investigation of the glass ceiling one step further
than previous research by using an approach that focuses upon the interaction of
intergroup dynamics and rational bias actions to attempt to explain the glass ceiling
effect in state government This study measures the interaction of identity group
dynamics and the related discriminatory acts associated with them. In other words, this
study investigates if belonging to a particular identity group, defined here by sex-role
stereotypes and management style instead of strictly by gender, is related to
representation of women at the top and, then,, allows for continued discrimination
- 12 -

against women to continue in state government Specifically, this study investigates the
following question:
Why do some agencies have a greater proportion of women in top management
than others? Is there a dominant style of management, or dominant group, as
defined by cognitive style and sex-role identification, that may keep women,
and men who are different from the dominant group, out of the top?
In a simplistic sense, this study attempts to quantify the notion of the "good-old
boys" network using intergroup theory (Alderfer, 1986), sex-role identity and the use
of management style stereotypes in structural discrimination (Powell, 1989). In this
study, belonging to a particular identity group is measured by personal and cultural
characteristics including sex-role identification (cultural masculinity, femininity,
androgyny) and cognitive or decision-making style.
All major administrative departments of Colorado state government were
investigated to attempt to explain inter-organizational differences in gender
representation. The key variables studied include:
Individual cognitive or decision-making style a measure of management
style of processing information and making decisions based upon Jung's
(1923) model of cognitive style.
Individual sex-role identification and its relation to gender a measure of
cultural sex-role (masculinity, femininity) based upon Bern's (1974) model of
sex-role identification.
The representation ratio of top management in the organizations the
proportion of women to men in top management of surveyed departments.
The relationship between women in top management and women at other
levels of management a ratio of women at the top as compared to women at
other levels to measure for segmented equality (Frederickson, 1991).
- 13 -

Limitations of the Research
As with any research, this study has several limitations that limit the validity of
the findings and the ability to generalize these findings to other situations.
Validity. The design for this study is passive correlational in nature. Because
one cannot readily assign individuals to organizations and organizations to styles,
existing organizations and their members were surveyed to attempt to isolate the
relationships between the independent variables (management style as defined by sex-
roles and cognitive style) and the dependent variables (representation of women at the
top). Because of this limitation, neither direct causal relationships nor the direction of
this causality can be inferred. As with all descriptive research, one can only make
inferences about the state of events at the time of the data collection, not about any
precedents or antecedents. This limits the interpretations of the results of this study.
The use of a mail survey technique meant there was no control over the
environment within which the surveys were completed. Although respondents were
asked to complete the instruments themselves, in one sitting, one cannot know exactly
how or who filled out the questionnaires. This also potentially limits the interpretations
of the findings of this study.
Although attempts were made to mask the hypotheses of the study, keeping face
validity low (knowing that the study was an investigation of sex-roles and differences
between men and women could have influenced the respondents' responses), the effect
of testing as a threat to internal validity could be a problem here. Some respondents
may have guessed the nature of the study and shaped their responses according. Other
threats to internal validity for quasi-experimental studies (e.g., history, maturation,
regression; see Campbell and Stanley, 1963) do not directly apply here because this
study is descriptive and not quasi-experimental.
The ability to generalize these findings is limited to Colorado and to state
governments similar to Colorado. External validity is also constrained because only
administrative departments were investigated. However, a number of other studies
- 14 -

focused on women and men in executive positions in state government (Hale & Kelly,
1989; Kelly & Guy, 1991) do allow a point of comparison for this study. It is
expected that the results will be generalizable to other similar governments. In addition,
the results will be generalizable to similar federal administrative agencies and,
potentially, some not-for-profit organizations. Also, in the tradition of generalizing
private sector studies to the public sector, some limited generalizations could also be
drawn to private sector executives.
Cultural Limitations. The majority of the research that provides the framework
for this study is work that has been done in the United States and with white or Anglo
respondents. The literature reviewed here refers to research which explores masculine
and feminine styles as though those gender differences were culture-universal and not
culture-specific. The result is to omit minorities and other cultures in the literature
much the same way that theorists, in the past, have omitted women from their research
(see Kohlberg, 1981; Perry, 1970). This omission is not intentional and reflects no
overt move to omit other world-views. However, the omission limits the
generalizability of the findings to primarily anglo cultures in the United States.
Limitations with Respect to Minority Representation. Finally, this study is
limited because it investigates only the question of the representation of women in top
management and ignores the representation of minorities in top management. As many
recent studies have indicated (Fergusen, 1984; Morrison & Von Glinow, 1990), the
glass ceiling is no longer only a woman's issue, rather it is an issue for all traditionally
disadvantaged groups. Although minority managers were included in the sample
frame, no attempt was made to isolate minority managers and to incorporate minority
status as a variable in the analysis. This was partially due to the low numbers of
minorities in the sample and partially due to a desire to limit the scope of the study. As
a result, these results cannot be generalized to all traditionally disadvantaged
management or identity groups.
- 15 -

It was hoped that this research would help further the understanding of the
factors that influence women's upward mobility and would be helpful as women
attempt to gain greater equality and equity across all areas of society. Specifically, it
was hoped that this study would be useful in helping public organizations understand
the barriers that exist today that keep organizations from achieving goals of social
As forecasts and trends indicate, women will continue to be a significant
element in the workforce and, as time-in-career ceases to be a mitigating factor,
organizations, both public and private, will need to account for the lack of
representativeness in top management positions. In addition, there will be a greater call
for government to be more representative not only in blocks (all women in government)
but also in segments (at all strata). A greater understanding of the barriers that may be
keeping women out of executive offices will help both women and men overcome the
obstacles that are keeping the glass ceiling suspended.
It was expected, based upon previous research, that there will be nominal
differences between men and women in top management in state government It
followed, then, that these top level managers would be very similar in their personal
and managerial styles. It was expected that those departments who have few women in
top management and/or who are "unbalanced" (fewer women at the top as compared to
women managers at other levels) would be more traditionally "masculine" departments.
This cultural masculinity may create a barrier to advancement for women in these
departments. It was expected that those departments that have more women in top
management, and/or who are more "balanced" (more or relatively equal women at the
top as compared to women at other levels) would be less traditionally "masculine" in
style and may be environments within which the barriers to advancement for women
are less severe.
- 16 -

Finally, it was expected that a model could be built using decision-style, sex-
role identification and other demographic information that can adequately explain some
of the variation in representation of women in top management in state government.
As the current theories in leadership focus on the new, less traditional, styles
that are needed to manage organizations in flux, the need to shift the composition of top
management becomes more acute (Harding, 1990; Cleveland, 1985; Loden, 1985;
Ouchi, 1981; McGregor, 1960). As it happens, styles that are described as traditionally
feminine (cooperative, connected, people-oriented) are representative of these new
leadership styles. However, a more feminine style is not necessarily found only in
women and may only be a gender issue in the United States because of cultural and
social variables (Hofstede, 1984). Whether or not cultural masculinity is an
overarching variable that affects all areas of executive management in the U.S. was not
tested here. The results of this research, however, may lead to an expanded view of the
glass ceiling in state government that goes beyond the measure of sex as a contributing
factor to underrepresentativness to a measure of the stereotyped roles associated with
men and women and the impact of these roles upon discriminatory practices..
As indicated earlier the need to understand the barriers to upper management in
the public sector is, to some degree, more crucial than in other sectors because of the
need for social equity in government When we have more representation at all levels,
individuals benefit because of greater opportunities for economic advancement, political
power and social prestige. The consumers of the products and services the government
supplies benefit because government organizations are more likely to be responsive to
their needs. Members of organizations benefit because of greater goodwill and equity
within the organization. And, especially, organizations benefit because of increasing
internal democracy, improved decision-making at all levels, improved responsiveness,
more efficient use of human resources, domestic tranquility and stability and increased
institutional legitimacy (Kranz, 1976; Stivers, in press). The need for these benefits
drives the need for research that will help us understand the barriers to
representativeness in government. A greater understanding may help break down these
barriers and help pave the way to greater administrative representation at all levels, in all
- 17 -

This chapter provides a more detailed review of the current literature in the two
general theoretical areas under study: personal discrimination and structural
Personal Discrimination Theories
Sex-Role Identification. One of the most common aspects of personal
discrimination in the United States is bias and stereotyping based on sex-role
identification. Whether one identifies as a masculine personality, a feminine personality
or a mix of both (androgynous) is due to childhood socialization and expectations about
roles as adults. In fact, when it comes to general behavior, sex-role identity, or gender
roles, may be more powerful determinants of behavior than are biologically determined
sex-related characteristics.
In understanding the development of sex-role identity, one must focus initially
upon human developmental theory and sex-role identity formation theory. The
foundations for sex-role identity, as well as most other behavioral elements, are built at
early ages through interactions with parents and peers. Sex-role identity begins early in
life and continues to adjust and change throughout development For example,
Chodorow (1974) states that from early on, "feminine personality comes to define
itself in relation and connection to other people more than masculine personality does"
(pp. 43-44). Because they usually are parented by a same-gender parent girls come to
experience themselves as less differentiated than boys. As a result feminine
personalities see themselves in relation to others around them, not separate.
- 18 -

Although with most complex and psychological issues it is difficult to determine
whether biological or sociological factors are most influential, gender or sex-role
identity is thought to be primarily a factor of socialization. Therefore, it is important to
differentiate between sex-related characteristics which are biologically determined, and
sex-role characteristics which are culturally influenced and determined:
The words masculine and feminine do not refer in any simple way to
fundamental traits of personality, but to learned styles of interpersonal
interactions which are deemed to be socially appropriate to specific social
contexts, and which are imposed upon, and sustain and extend, the sexual
dichotomy. (Newson, Newson, Richardson & Scaife, 1978, p. 28)
Traditionally, feminine traits have been associated with girls and women while
masculine traits have been associated with boys and men. This is especially true in
cultures which place a high value upon masculinity and men, and a lower value upon
femininity and women (Hofstede, 1984). Although sex-roles vary across cultures,
clear, dominant sex-role patterns emerge cross-culturally which reflect an association of
assertiveness and aggression with masculinity and nurturance and submission with
Male behavior is associated with autonomy, aggression, exhibition and
dominance; female behavior with nurturance, affiliation, helpfulness and
humility. However, the fact that an active feminist movement exists in a
number of countries (albeit predominantly the wealthy ones) shows that some
women, at least (and some men too), no longer take the traditional pattern of
male dominance for granted and try to develop alternative role distributions.
(Hofstede, 1984, p. 178)
Historically, it was thought that one either identifies with a masculine identity or
a feminine identity depending upon the socialization factors in one's youth. Sex-role
identity was seen as being a bi-polar continuum with varying degrees of polarity
represented. Theoretical concepts of sex-role identification, however, changed in the
early 1980s to include the concept of androgyny:
- 19 -

Femininity and masculinity have long been conceptualized as opposite ends of a
singular bi-polar dimension. More recently, scholars in a number of disciplines
have begun to concern themselves with the concept of psychological
androgyny, a term that denotes the integration of femininity and masculinity
within a single individual. The concept of psychological androgyny implies that
it is possible for an individual to be both compassionate and assertive, both
expressive and instrumental, both feminine and masculine depending upon the
situational appropriateness of these various modalities. And it further implies
that an individual may even blend these complementary modalities in a single
act, such as the ability to fire an employee, if the circumstances warrant it, but
with sensitivity for the human emotion that such an act inevitably produces.
(Bern, 1981, p. 4)
The consequences of identifying with a particular gender identity, be it
feminine, masculine or androgynous, have a significant effect upon all aspects of life.
Specifically in relation to management, the consequences of identifying with a feminine
personality can be especially problematic (Powell, 1988).
Part of the importance of identification is that sex-role identity has been found to
be related to a number of other personality constructs. For example, sex-role identity
has been found to be related to perceptions of competency. Sherman (1976) found
that, in a study of female competency, the goals of traditional femininity and
competence are diametrically opposed. Feminine personalities are socialized to be
passive and dependent Feminine personalities are not socialized to be independent in
problem solving and to achieve. Sherman went on to posit that this socialized passivity
and dependence assures that women will not fit in traditionally logical-positivist
organizations where the rational model of decision-making is adhered to and highly
In a classic study examining the differences in children's play, Lever (1976)
reported that girls were more likely to avoid conflict in play than were boys. Boys
were able to resolve disputes more effectively whereas girls avoided disputes; a dispute
during a girls' game would most likely end the game. Lever concluded that boys are
better at handling conflict than girls, a trait which she assumes transfers into adulthood.
Gilligan (1982) evaluates Lever's findings as being bounded by traditional models for
success. Gilligan states:
- 20 -

... [the] assumption that shapes her [Lever's] discussion of results is that the
male model is the better one since it fits the requirements for modem corporate
success. In contrast, the sensitivity and care for the feelings of others that girls
develop through their play have little market value and can even impede
professional success, (p. 10)
Throughout the socialization process, girls and women find that their people-oriented,
relational decision-making and behavior are not valued as critical success factors in
U.S. society.
Recently, attention has turned away from sex-specific studies of sex-role
identity to sex-related studies of sex-role identity. In other words, researchers are
beginning to look at sex-role identification separate from sex in the hope of
understanding the relationships (or lack of) between sex-roles and sex. For example, a
man who has strong feminine characteristics will approach a situation very differently
than a man who has strong masculine characteristics. Similarly, a woman who has
strong masculine characteristics will approach a situation differently than a woman with
strong feminine characteristics. The people most likely to approach the situation
similarly are the people who share the same sex-role identity or style (masculine
man/woman; feminine man/woman; androgynous man/woman) rather than the same
In her classic study of women's attitudes towards success, Homer (1970)
claimed to find that women have a fear of, or a need to avoid, success. However,
recent research found that Homer's work was, in fact, biased by the strategies of
success given by the subjects (Simmons, King, Tucker & Wehner, 1986). What the
women in Homer's original model were reacting to was not success in itself, but rather
success achieved through competitive strategies. When presented with stories that
described both competitive and cooperative success strategies, no gender differences
were found by Simmons et al. Rather, all subjects regardless of gender rated the
competitive strategies more negatively than the cooperative strategies. Cooperative
success strategies (people-oriented, relational) where people work together to achieve
personal success are seen as being more positive by both males and females than are
competitive strategies. The sex-role orientations of the college-student subjects in this
- 21 -

study were not measured, but one could hypothesize that these subjects shared style
preferences more than they shared sexual characteristics.
In a study that challenges the stereotypes of traditional roles, Lyons (1983), in
evaluating individual identity formation found that there are two orientations of identity
that do not necessarily split along gender lines:
1) responsibility orientation where the concept of self is rooted in connection
and relation to others (value cooperation); and
2) rights orientation where the concept of self is rooted in separation and
autonomy (values competition).
Women tend, more than men, to identify with the responsibility orientation of
the self, one where people and events are evaluated in regard to their connections. Men
tend, more than women, to identify with the rights orientation of self. These tendencies
are, most likely, not related to sex but to sex-role socialization.
Desjardins (1989) took Lyon's model and developed a two-dimensional theory
of leadership based upon identity and moral orientations. Her theory encompasses a
"Morality of Rights" dimension, which is traditionally masculine, and a "Morality of
Response" dimension, which is traditionally feminine. The "Rights" dimension
embodies a leader who has a moral orientation toward justice, has an analytical learning
style, interacts best in a competitive mode and fears oppression. The "Response"
dimension has a moral orientation towards relationships, has a synthesizing/cooperative
learning style and interaction pattern and fears abandonment
To test her theory, Desjardins interviewed 36 male and 36 female community
college executives. Although the interviewees exhibited behavior within both
dimensions, the majority of women CEOs could be classified on the "Response"
dimension while the majority of men CEOs could be classified on the "Rights"
dimension. The ability to move back and forth between dimensions as well as the links
between sex and type are both significant findings. The finding that both women and
men in Desjardins' sample were capable of displaying either traditionally masculine or
traditionally feminine characteristics, depending upon the needs of the situation, may
- 22 -

provide some evidence that sex-role related patterns of style are learned through
socialization and are linked more to enculturated identity than to biologically determined
sexual characteristics.
Sex-role identity has been used directly to investigate gender differences in
management by Powell and Butterfield (1979). Powell (1988) claims that gender
differences in management are primarily a function of an interaction between one's
gender, sex-role stereotypes and managerial stereotypes. In a study where respondents
listed the traits that made a manager effective, Powell and Butterfield (1979) found that,
in the majority, respondents believe that an effective manager exhibits traits that are
traditionally considered to be masculine. In fact, exhibiting traits that are traditionally
feminine or androgynous detracts from perceptions of managerial effectiveness. In
addition, these perceptions have not significantly improved over time. Powell and
Butterfield found that in 1977,60% of respondents viewed a good manager as
masculine; in 1985,66% of respondents viewed a good manager as masculine. These
stereotypes continue to be held even in the face of research that shows, in terms of real
performance, the better (more effective, efficient) managers tend to be more
androgynous (combining masculine and feminine tendencies, see Bern, 1974; Sargent,
1983). Powell and Butterfield's results indicate that, regardless of the strides made by
women in management over the last twenty years, and regardless of research that
indicates that better managers tend to be more androgynous through measuring actual
performance, stereotypes or perceptions of good management traits have not
substantively changed over time.
Some history may help illuminate this confusion between perceptions of good
management behavior and the actual characteristics that describe good management
behavior. Powell (1989) states that in the management literature, three major theories
or schools of thought are used to explain differences in management behavior. These
theories make reference to management characteristics which could be classified as both
traditionally masculine and traditionally feminine and include: great-man theories which
were developed in the belief that knowledge about great leaders of Western civilization
provide insight into leadership capabilities; trait theories which focus on individual traits
and characteristics; and behavioral theories which focus on the "specific behaviors used
by managers to influence their subordinates actions" (Powell, 1989, p. 11). Of these
- 23 -

three theoretical categories, only the behavioral theories have survived the test of time
and still receive attention in the management literature.
Two general types of management behavior have been identified and subscribed
to by most theorists: task-oriented behavior and people-oriented behavior (Blake &
Mouton, 1964). Task-oriented behaviors include initiating structure, setting goals and
making decisions and are usually associated with a masculine management stereotype.
People-related behaviors include showing consideration for people, soliciting
subordinates' ideas and demonstrating concern for subordinates' satisfaction and are
usually associated with a feminine management stereotype (Powell, 1989).
Androgynous management is described as a combination of task-oriented and people-
oriented behaviors (Sargent, 1983), or as Bern (1981) described it, the ability to couple
a task orientation with sensitivity to people.
Donnell and Hall (1980) found that good managers successfully integrate both
task and people concerns, average managers concentrate on the task at the expense of
people, and poor managers show little concern for either. Donnell and Hall
hypothesized that the proliferation of masculine management style in most U.S.
organizations could be a possible explanation for management problems many
organizations face today. A dominant style that rewards task over people may lead to a
proliferation of merely average managers in the U.S. (Donnell and Hall, 1980).
Research has shown that the majority of top managers have a masculine
approach to management, that is, an approach that focuses upon tasks rather than
people (Powell, 1988). This approach is considered to be the most stereotypically
desirable management style even though a more androgynous approach has been
shown to be a more effective style. Less masculine personalities may attempt to make it
to the top but may be thwarted by the enormity of the differences between themselves
and their peers. Indeed, these differences become even more pronounced when one is
the token member of a non-dominant group.
In her study which was comprised of interviews with executive women, Milwid
(1987) states:
- 24 -

As a sense of mutual understanding evolves among the members, a group
becomes a safe refuge from the plurality and confusion of the world at large.
To the degree that the group remains homogeneous, its members feel
comfortable and protected. But as soon as someone different joins, the safety
factor disappears.. Given this dynamic, it is not surprising that all male
management teams subtly or overtly resist the presence of women, (pp. 188-
One of the executive women interviewed by Milwid (1987) put it another way:
What is really going on with all of this chain of command stuff is that it
perpetuates itself....They promote people who look the way that they do, who
act the way they do, who play the same sports they do. It's not that they
[purposively] exclude women or any other group -- it's more of a consolidation
of the one given group, (p. 190)
Some women believe there is a negative relationship between being a woman,
or woman-identified, and being an executive. Kelly (1991) describes the masculine
nature of bureaucracies and indicates that feminists that study bureaucracies,
particularly public bureaucracies, assert that bureaucracies have a masculine orientation
and bias because they were developed by men:
...women who become managers and have successful careers must become like
men not only in their training but also in their behavior...fewer women will
have the power, achievement and leadership motives (necessary to advance),
given sex-role socialization.and ideologies, but that those who do have these
motives will behave similarly to men. (Kelly, 1991, pp. 97-100)
In a study that combined survey and interviewing techniques, Gilson and Kane
(1989) found that many of the private sector executive women they surveyed struggled
daily with their roles as women and as an executive. Their survey measured sex-role
characteristics similar to the survey used here and found:
- 25 -

On the surface, the anima (feminine side) of these women seems stronger than
the animus (masculine side)....But, lest we believe that they openly or
consistently display this side to themselves, the following facts quickly dispel
that notion. Only about half manifest their tender, caring sides. On the
quintessentially feminine traits, the numbers dipped even lower. Most telling,
[only 38%] say they are feminine.
The women's dissociation from femininity lies in their belief that acting female
impedes a woman's career. Women in business don't want to be seen as
frivolous or incompetent....Femininity also means flightiness and spaciness and
in business you have to be calculating and rational. (Gilson & Kane, 1989, p.
The authors' conclusions can be heard echoing in the words of the women
interviewed as they spoke of their conflicts between socialized roles and expected roles
within executive management The authors sum up the path of women in management
over time:
When women began migrating into the workforce in the late 1960s they had
few female role models to show them how to retain their feminine attributes and
add to these the professional qualities they saw in successful men.
Metaphorically speaking, the women transferred all of their eggs from one
basket to another; they relinquished their femininity and became as businesslike
as they could. (Gilson and Kane, 1989, p. 106)
Gilson and Kane (1989) go one step further and hypothesize that women in
business over-compensate for being women by denying their identification with their
sex and taking on roles that are clearly different from traditionally female roles:
If women dont want to be women-identified, they will not just curb their
warmth and liveliness, they will steer clear of women's issues and of women
themselves. These women are not comfortable being women-identified. They
are trying to tell me they are like men. [They are trying to make] an alliance
with the men who still, for the most part, control their destinies, (p. 180)
Other interviews with female executive managers communicate the same
message that being feminine is counterproductive to being an executive manager:
- 26 -

When someone is being considered for top management at this company, they
look at a lot of things other than competencies and performance. They look at
style issues and deportment and whom the individual knows in the community.
Having a female style doesn't get you into senior management. In the short
term, being a woman pays off; in the long term, it certainly doesn't (Milwid,
1987, p. 83)
Stivers (in press) believes that the negative relationship between femininity and
executive management also exists in public administration:
...women who pursue careers as professional public administrators are faced
with a dilemma the fundamental dissonance between what is expected of them
as professional women and what is expected of them as professional experts....
Thus, it is fraudulent to offer women an [opportunity] rise through the
ranks of bureaucracy while the requirements and exemplary qualities for that
sort of career remain inconsistent with what is expected of them as women. To
point out that a number of women have done it successfully is to miss the point.
They have virtually never done it without constant effort to manage their
femaleness on the job and without continuing struggle to balance work and
home responsibilities. (Stivers, in press, p. 24)
Perhaps Fierman (1990) summed it up succinctly in her popular press article on
the glass ceiling by advising women in top management to "look like a lady; act like a
man; work like a dog."
To summarize, many theorists believe that the obstacles to the top for women
can be explained by gender stereotypes. Management continues to be a masculine
domain, especially at upper management levels. Further, psychological type theories
indicate that type may be only gender-related, not gender-specific, and that particular
types that are more masculine in nature are more likely to be in top management
positions as a result of dominant stereotypes. Stereotyping of a preferred management
style may be one of the barriers to upper management that is keeping the glass ceiling
suspended for women, particularly if the stereotype leads to a dominant group of
masculine upper-level managers.. This is especially true if the structure or the system
of the organization allows the dominant group to retain its control over the keys to the
executive suite.
- 27 -

Management/Decision-Making Style in Organizations. Although people and
organizations can be seen as discriminating on a number of characteristics related to
sex-roles and sex-role stereotypes, many of these characteristics are related to
information processing, cognitive functioning and decision-making. Furthermore,
many discrimination theorists see cognitive functioning or decision-making as integral
characteristics upon which discrimination is based, especially discrimination in
organizations (Alderfer, 1987; Kanter, 1977a).
Human decision-making in organizations is a complex topic. In the middle of
the 1980s, in response to a systems approach to management that swept the field of
management and organizational behavior, much attention in management and in
management research was focused upon decision-making in organizations.
Decision-making can be defined as the means by which individuals and
organizations arrive at a course of action for a particular problem or situation. It is the
complex process by which alternative courses of action are identified and analyzed in
preparation for a final actionable solution to a problem (Gortner, Mahler & Nicholson,
Traditionally, decision-making has been investigated from three different, but
overlapping, perspectives. The organizational and management sciences investigate the
decision-making of organizations or of particular key individuals within an organization
(e.g., managers or leaders). Social and cognitive psychologists investigate the
decision-making of individuals irrespective of their contextual bounds and focus more
specifically on the cognitive processes involved in the decision itself. Psychologists
also focus upon human problem-solving as a separate area of investigation that is linked
to decision-making. Problem-solving is investigated as a human information
processing model as an element or part of the greater whole of human decision-making.
In organizational decision-making, the unit of analysis is the organization itself. In
human decision-making or problem-solving, the unit of analysis is the individual.
The heritage of organizational decision-making can be traced as far back as
1938 when Chester Barnard published his important treatise on the function of the
executive in organizations. Simon's work on re-evaluating economic theories (Simon,
- 28 -

1947) and the work of March and his colleagues on defining organizational ambiguity
and the role of the individual (Cyert & March, 1963; March & Olson, 1982) helped to
set the nature of the current theory in organizational decision-making. The nature of
this inquiry has developed and blossomed over time into a wide-ranging field of study.
The study of organizational decision-making has evolved from a rational-
bounded approach to one that reflects a multidimensional or interdisciplinary approach.
Inherent in this interdisciplinary approach is the notion that organizational decision-
making can be less rational and more intuitive, depending upon the positions of the key
actors in the decision, or the individuals involved in the decision process (Agor, 1989,
1986). Hunt, Kryzystofiak, Meindl and Yoursy (1989) present a model of the
structure of decision-making in organizations in which the outcome of the decision is
based upon the decision-maker (the individual), the decision task (what kind of
decision is it, e.g., programmed versus non-programmed decision, see Simon, 1960),
the decision situation (the organization) and the decision process. According to this
model, it is shortsighted to study organizational decision-making without also studying
the individual decision-maker and the elements that he/she brings to the decision
The study of individual decision-making processes has focused on the cognitive
processes of the individual, use of judgmental heuristics and problem-solving style.
The general thesis is that individuals use simple methods that are serial and sequential
which are bounded and anchored in previous experience or learning (Einhom &
Horgath, 1982). In this model, decisions are made to optimize the survivability of the
The study of individual decision-making has followed a course similar to
organizational decision-making. The roots of individual decision-making are in
statistical decision-making and economic theories, with recent research reflecting the
trend toward studying the cognitive processes and the social/contextual and cultural
aspects of decision-making. Most models reflect an input-output-feedback engineering
model within which the elements of search, information acquisition, mental processing
(e.g., matching with memory), evaluation and action are important (Einhom &
Horgath, 1982). The effect of the group (Janis, 1971), as well as cultural (Allison,
- 29 -

1969) and personality factors have also been investigated and related to individual
Recent work on human decision-making in organizations focuses upon
cognitive style, or the methods by which individuals arrive at decisions. Perhaps the
most influential work in the area of understanding gender differences in cognitive style
and information processing is Carol Gilligan's (1982) work on moral development in
women. Although the generalizability of Gilligan's results is limited due to her
methodological dependence upon qualitative techniques, she hypothesized that the
development of morality in women is organized around responsibility, care and
connections (Gilligan, 1982). Her exploration of this topic was predicated on her belief
that, because the leading models of moral development (Kohlberg, 1981; Kohlberg &
Kramer, 1960) traditionally place women at lower levels of moral development, these
models ignore the more feminine approach to morality which focuses upon
responsibility, care and connections.
In Kohlberg's (1981) model, the highest level of moral development represents
adherence to blind-justice or the subordination of everything to abstract and universal
laws and principles. Lower levels of moral development represent an organization
around connections and contextual evidence. Kohlberg used only male subjects in
developing his model and assumed that women, when tested with the model, were
simply less morally developed that men. Gilligan (1982), in contrast to Kohlberg
(1981) found that women have an understanding of the abstract laws and principles but
apply them with an appreciation of the context of the moral dilemma; they evaluate the
impact of the moral situation upon the affected individuals in addition to evaluating the
universal principles. This does not make women less morally developed than men,
only different. Gilligan writes:
Kohlberg and Kramer imply that only if women enter the traditional arena of
male activity will they recognize the inadequacy of this moral perspective and
progress, like men, toward higher stages where relationships are subordinated
to rules, and rules to universal principles, (p. 18)
Gilligan's work spawned many other evaluations of feminine tendencies to
organize events and information around connections. Foremost among these is
- 30 -

Belenky, Clinchy, Goldberger and Tarule's (1986) work on women's styles of
knowledge acquisition. Using Perry's (1970) model of intellectual and ethical
development of college-age men as their base model, they posit that, like Kohlberg's
model, Perry's model fails to include ways of knowing that are specific to women.
Like Kohlberg, Perry also failed to include women in his developmental sample and,
also like Kohlberg, women tend not to reach higher stages of intellectual development
in the Perry model (Perry, 1970).
Belenky et al. (1986), using extensive qualitative and quantitative methods,
developed five epistomological perspectives from which women view reality and draw
conclusions about truth, knowledge and authority:
1) Silence experience self as mindless and voiceless.
2) Received Knowledge knowledge is received from authorities.
3) Subjective Knowledge truth and knowledge is personal and private,
subjectively known or intuited.
4) Procedural Knowledge analytical and objective procedures are used for
attaining and communicating knowledge.
5) Constructed Knowledge knowledge is both subjective and objective,
knowledge is contextual.
In their sample, the authors found that the most common perspective for women
is the subjective knowledge perspective (50% of their sample fell into this category).
The least common are procedural knowledge and constructed knowledge, those
perspectives that would be most valued by a masculine model of cognitive style. They
In a world that emphasizes rationalism and scientific thought, there are bound to
be personal and social costs of a subjectivist epistomology. Women
subjectivists are at a special disadvantage when they go about learning and
working in the public domain. (Belenky et al., 1986, p. 55)
Gilligan and Belenky et al.'s work focused the attention of moral and cognitive
development theorists on the limitations of the traditional models. More recent research
on this topic has made use of the Myers-Briggs Type Indicator (MBTI, 1985; Briggs &
- 31 -

Myers, 1957). The MBTI is a test of psychological type developed by Irene Myers and
Katherine Briggs based upon Carl Jungs (1923) theory of psychological type. Jung
concluded that there are a limited number of fundamental ways in which people react
psychologically to various situations. Two of his four dimensions of psychological
type represent cognitive functions, or functions related to the acquisition and evaluation
of information. The perception functions divide the acquisition of information into
either a sensing function (acquiring information from external sources) or the intuitive
function (acquiring information from internal gut level sources). The judgment
functions divide the evaluation of information into either a thinking function (the use of
logic and analysis) or the feeling function (the use of emotional, spontaneous
processes). Of the four Jungian functions, the judgment functions are the only
functions that show any sex differences as measured by the MBTI. More women tend
to be feeling information evaluators and more men tend to be thinking evaluators
(Myers & Briggs, 1985).
Otis and Quenk (1989) used the MBTI to investigate psychological type and
moral development. Their study represents the significant attention that researchers are
currently paying to sex differences in moral development as an attempt to quell the
morass of assumptions that women are not as good as men in moral and cognitive
functioning. Otis and Quenk's findings question sex-related differences in moral
reasoning and cognitive processing. They found that gender differences in moral
reasoning (like gender differences in sex-role related variables) are not necessarily
related exclusively to sex, but rather are related to the individual's style (preference for
thinking or feeling as measured by the MBTI).
In the Otis and Quenk study, thinking evaluators were more likely to use a
masculine justice model for evaluating moral dilemmas; feeling evaluators were more
likely to use a more feminine justice model. The feminine justice model is similar to
Gilligan's (1982) model of care considerations and Desjardins' (1989) response
dimension. Otis postulates that these information processing differences are not a
function of sex, but, rather, are a function of type as measured by the MBTI. Type is
related to sex but is not a direct sex-linked function. In other words, type is socially or
culturally determined rather than biologically determined and is more related to whether
one identifies with a masculine or feminine sex-role.
- 32 -

Rowe and Mason (1987) have taken the Myers-Briggs typology of decision
style one step further in their development and subsequent extensive testing of their
theory and measurement of decision style. Rowe and Mason, using the same Jungian
psychological antecedents as Myers and Briggs, focus on decision style as a
combination of cognitive complexity and value orientation. Cognitive complexity
relates to the cognitive predilection for decision situations where there is little or great
ambiguity and complexity. Some people prefer a less ambiguous and complex decision
environment, others prefer, or are comfortable with, more ambiguity and complexity.
In addition, individual decisions are also affected by one's value orientation which may
be directed either at human and social concerns or to task and technical concerns.
Combining the cognitive complexity dimension and the value orientation dimension
yields four basic decision styles:
Directive Decision Style: low tolerance for ambiguity/cognitive complexity and a
task/technical value orientation.
Analytical Decision Style: high tolerance for ambiguity/cognitive complexity and
a task/technical value orientation.
Behavioral Decision Style: low tolerance for ambiguity/cognitive complexity
and a people/social value orientation.
Conceptual Decision Style: high tolerance for ambiguity/cognitive complexity
and a people/social value orientation.
There are strong relationships between the MBTT types and the four basic
decision styles as measured by the Decision Style Inventory (DSI- Rowe & Mason,
1987). Essentially, Rowe and Mason (1987) have defined styles that are highly
correlated with the MBTI types on two major dimensions; perception functions
(iNtuiting and Sensing, or how information is evaluated) and judgement functions
(Thinking and Feeling, or how information is acquired). People who have an ST style
preference on the MBTI will tend to use a directive decision style, people who have a
NT style preference on the MBTI will tend to use an analytical decision style.
- 33 -

Alternatively, people who have a SF style preference on the MBTI will tend to use a
behavioral decision style while people who have a NF style preference on the MBTI
will tend to use a conceptual decision style. Results of tests for construct validity
between the MBTI and the DSI indicate that the two methods are measuring similar
constructs (Rowe & Mason, 1987).
Directive types tend to be highly task-related and have little tolerance for
ambiguous situations or situations that require a great deal of cognitive complexity.
Directive types are often fairly control oriented, are realistic and direct when dealing
with problem situations, are pragmatic and decisive. Directive types rely more on
intuitive, or gut-level, information to make decisions than on facts and numbers.
Directive types tend to be oriented more on tasks than on people and, as a result, are
often seen as authoritarian in their leadership style. Many famous military leaders and
bureaucrats are examples of directive types (Rowe & Mason, 1987).
Analytical types are also task-related but are capable of handling situations with
a great deal of ambiguity and cognitive complexity. Analytical types are focused on
problem solving and discovering new solutions to problems. They are disciplined,
enjoy challenges and apply careful analysis to their work. Analytical types rely heavily
on facts and numbers to make their decisions. Analytical types are often found in jobs
that require sophisticated analytical skills (Rowe & Mason, 1987).
Conceptual types also perform well in highly ambiguous situations or situations
that require a great deal of cognitive complexity. However, conceptual types tend to be
more people-oriented than task-oriented. Conceptual types tend to be dreamers and
futurists who rely highly on intuition to make decisions. Conceptual types rely
significantly on their observation of people and their imagination. Decisions are
typically broad and flexible. Following rules is rarely chosen over exploring all
options. Entrepreneurs are often conceptual types (Rowe & Mason, 1987).
Behavioral types are highly people-oriented and prefer low ambiguity and
demands for cognitive complexity. Behavioral types are focused on feelings,
developing others' potential, interactions, acceptance and sensitivity. Teachers and
social workers are often typical behavioral types (Rowe & Mason, 1987).
- 34 -

The DSI is often preferred over the MBTI because it measures decision style
alone, not style along with other dimensions of personality. The DSI is also preferred
because it measures decision style using a rank-order scale rather than a bi-polar,
forced-choice answer. Also, the DSI measures dominant decision styles as well as
back-up styles whereas the MBTI only measures dominant style. Dominant decision
styles are defined as the preferred style that is used in most decisional circumstances.
Back-up styles are used situationally, indicating that decision-style is fluid, not static.
As the MBTI perception functions (Thinking, Feeling) are the only functions of
style that show sex differences (men are more likely to be thinking; women are more
likely to be feeling), one would assume that there are sex differences in decision style.
That is, men should be more likely to use directive and analytical styles while women
should be more likely to use behavioral and conceptual styles. In fact, sex differences
testing with the DSI reveals that women have a greater tendency to have a dominant
behavioral or conceptual style and are more likely to be in careers that value this style.
When sex differences are tested within career fields (e.g., for engineers or top
managers), however, there are no consistent differences between the sexes. In other
words, the identity of the organizational group is more likely to match personal decision
style than is the identity of the individual. Work groups are more likely to have
homogeneous decision styles, even if the make-up of that group is diverse in regard to
other key issues including gender and race.
In addition, upper level managers are likely to have the same decision style as
measured by both the MBTI and the DSI. Results of studies using the MBTI to
measure differences in type among managers at different levels in a mix of
organizations show that the majority of managers, executives and administrators are
thinking types (90% of corporate executives, 75% of public sector executives and 64%
of public administrators are thinking types, Center for the Application of Psychological
Type, 1986), indicating that one of the obstacles to women (and men) who are more
feeling evaluators could be the dominant information evaluation style of top and middle
levels of management
Studies using the DSI to measure decision style in managers at different levels
in a mix of organizations indicate that the majority of top executives and managers are
- 35 -

directive or analytical types (Rowe & Mason, 1987). These types represent a task
orientation and a mix of a more rational approach to decision-making with a more
intuitive approach to decision-making. Although there is little evidence that shows sex-
differences in type as measured by the DSI, no research has been done to investigate
the relationships between decision style and sex-role identity. Directive and analytical
types may be related to a culturally masculine approach whereas conceptual and
behavioral styles may be related to a feminine approach to decision-making and
decision situations. If this is true, then it follows that decision style of the dominant
group may also be a barrier for women to the upper levels of management
Recent leadership research merges task and people orientations with traditional
masculine and feminine problem-solving skills. In the turbulent times of the 1990s,
scholars are calling for multi-faceted leaders and managers. Agor (1986) states that
decision-making at top levels of management is not an area where programmed
decisions abound, but rather new decision situations are faced daily. Agor posits that
intuitive decision-making rather than the traditional models of bounded/rational
decision-making is of greater value at this level. Cleveland (1985) states that the
knowledge revolution is calling for executives who use consensual modes of decision-
making, work cooperatively and balance power and social responsibility. Powell
(1988) states that the new leaders should value teamwork, have a concern for the
individual as well as the task, and should emphasize collaboration and interdependence
rather than competition and independence. Loden (1985) advocates a feminine model
of leadership where cooperation, teamwork, quality and a mix of intuitive and rational
thinking drive the leadership behavior. She claims that a balance of traditionally
masculine and feminine traits will be needed in more dynamic, flexible environments.
If decision-style is related to whether one's type is more thinking or feeling and
the majority of top managers currently are more thinking or more analytical/directive,
then this revolution in leadership will require that top management broaden its base with
respect to decision-making styles to include a mix of types and approaches.
Traditionally, people who take an approach that is different from a masculine approach
have been women or men with non-traditional styles. Because of this tendency to relate
sex to decision style, the dominant decision style of management may be far more
responsible for keeping women and other non-traditional groups (or people who
- 36 -

identify with a feminine or androgynous sex-role identity) out of top management than
are issues directly related to gender or sex differences. To some degree, dominant
decision style in top management is a discrimination issue which could involve sex-role
identification as a key identity group (Alderfer, 1986), and may be keeping non-
dominant groups out of top management
Structural Discrimination Theories
As briefly discussed earlier, structural discrimination theories focus upon the
institutionalized discrimination that results from organizational or societal norms and
practices in dealing with members of certain groups that are in non-dominant positions.
Structural discrimination theories explain how personal discrimination becomes an
integral part of systems and structures.
Alderfer's (1986) intergroup theory addresses the structural discrimination that
is the result of identity group membership (race, ethnicity, sex, age) being at odds with
organizational group membership. Because, generally, individuals tend to surround
themselves with like individuals, organizational groups tend to be organized around
identity groups. Extreme tension exists, for all parties involved, if one belongs to an
organizational group that is primarily composed of people who hold memberships in a
different identity group. Alderfer (1986) defines the idea of a group in organizations as
...with individuals who are interdependent, moves to the sense of the group as
a significant social object whose boundaries are confirmed from inside and
outside (the organization), recognizes that the group as a whole is an interacting
unit through representatives or by collective action, and returns to the individual
members whose thoughts, feelings and actions are determined by forces within
the individual and from both the group members and non-group members, (p.
The properties that are characteristic of intergroup relations within and among
organizations regardless of the group or the setting of the relationships include: 1) the
group boundaries which determine who is a member; 2) power differences which
- 37 -

influence the degree of permeability, or the ease with which group boundaries can be
crossed; 3) affective patterns which influence permeability in regard to the degree of
polarization of feeling among groups; 4) cognitive formations which include language,
perceptions of objective and subjective phenomena and decision-making and are used to
explain the nature of experiences encountered by group members and to influence
relations with other groups; and 5) leadership behavior which reflects the total pattern
of intergroup behavior in any situation (Alderfer, 1986).
Alderfer further explains that the tensions between identity groups (e.g.,
gender, racial, religious identity) and organization groups are inevitable as long as there
are systematic, structural processes that allocate people to organizational groups as a
function of their identity groups:
There is usually enough tension among organizational groups to occupy the
emotional energy of the top group, who have the task of managing group
boundaries and transactions, Thus, unless there are special forces to strengthen
the boundaries of identity groups within organizations (i.e., give them more
authority), the inclination of those in senior positions will be to manage only in
terms of organization groups. The manner in which an organization is
embedded in its environment and the relations among identity groups in that
environment will affect the degree to which management processes respond to
identity and organization groups, or just to organization groups, (p. 207)
Kanter's (1977a; 1977b) research on gender dynamics in organizational groups
work provides evidence that validates and substantiates Alderfer's theory. Kanter
(1977a), in studying small groups, found that the gender ratio of a group had a
significant impact upon the management behavior of group members and, in turn, upon
gender stereotyping. She described four categories of group gender ratio:
1) Uniform all same gender
2) Skewed skewed in one direction from 85:15 to 100:0
3) Tilted less than skewed 65:35 to 85:15
4) Balanced 50:50 split, or equal gender split
In uniform and balanced groups, tension and problems between men and
women are minimal. In tilted groups, one gender becomes the majority while the other
becomes the minority. Effects of gender differences in a tilted group exist but are not
- 38 -

as severe as in skewed groups were the greatest degree of tension exists. In a skewed
situation, one gender becomes dominant, the other becomes token. It is in the skewed
situation that the greatest gender stereotyping and same-gender teaming takes place
(Kan ter, 1977a). Although Kan ter designed her research to measure the effect of
women as tokens, she does point out that the same behaviors hold true no matter whom
is the dominant gender.
In a skewed group, role encapsulation and boundary heightening tend to take
place. In boundary heightening, the dominant group members tend to become more
aware of what they have in common and tend to make the boundaries or differences
between the two groups more apparent. In role encapsulation, the token members are
subject to greater performance pressures and play only limited roles. Due to both
boundary heightening and role encapsulation, the token members can either conform
and try to become like the dominant group members or they can hang on to their
differences and try to succeed. Unfortunately, few will succeed in this latter scenario.
Intergroup theory (Alderfer, 1986) and Kanter's (1977a; 1977b) of gender
balance both may explain why there are few women in upper management positions
and why certain management behaviors that are considered to be traditionally masculine
may be dominant in these positions, regardless of the sex of the manager. According to
intergroup theory (Alderfer, 1986), if managers are allowed to manage only by
organizational group rather than a combination of identity and organizational group,
people who are different from the dominant group will not rise to the top because the
boundaries of the identity group at the top of these organizations are not permeable.
Institutionalized management by boundaries, especially management style differences
as manifested by masculine management and decision-making styles, may be yet
another factor keeping the glass ceiling suspended for women.
- 39 -

Research Model
This study, using a positivist, empirical approach, attempts to explain some of
the factors that are keeping the. glass ceiling suspended for women in state government
As stated throughout this chapter, the current view in the field holds that a majority of
these factors are thought to be related to discrimination, specifically discrimination that
is personal and institutional in nature. As Morrison and Von Glinow (1990) have
stated, it is no longer sufficient to approach the problem of underrepresentation of
women (and minorities) in management from a unilateral or one dimensional approach.
Rather, hypotheses and research, in order to be productive, should take an interactionist
approach, looking for the interaction between personal and systemic variables.
Looking at only those variables that can be explained by structural elements is
not sufficient. In a recent study attempting to model the determinants of integration in
the public workplace (Kellough, 1990), a statistical model of U.S. federal civil service
employees at all levels yielded explained variance as small as 28% and as large as 48%
(depending upon the specifications of the model). The author states:
As much as 50 percent of the variation in the employment of women and
minorities [in the federal civil service system] is apparently explained by
contextual variables that are not ordinarily changed by an agency's EEO office.
(Kellough, 1990, p. 564)
The theory tested here is modeled in Figure 2.1. Essentially, this study
attempts to measure the interaction of personal and structural discrimination variables
with personal and organizational demographics in order to explain the nature of
variation in representation in top management levels in Colorado state government.
- 40 -

Figure 2.1
Research Model
decision style
>ex-role identity
- 41 -

Research Hypotheses
This study investigates the hypothesis that dominant sex-role identity, decision-
making style and demographic/organizational variables interact to partially explain
variations in the representation of women in top management
It is hypothesized that in a masculine organization, that is, task related and
competitive, there would be fewer women in top decision-making positions. In more
feminine organizations, that is, people related and cooperative, there will be more
women in top decision-making positions.
Furthermore, it is hypothesized that the underrepresentation of women in top
management of certain work groups may not be a gender issue, but rather a style issue.
Women may be barred from the executive suite not because of their sex but rather,
because their managerial style does not match the dominant style of the organization.
A key variable in this research is the dominant management style of die
organization. It is assumed that the dominant management style of the organization
determines the composition of the top management. That is, organizations that value
traditionally masculine management behaviors will have top management representation
ratios tilted towards males; organizations that value traditionally feminine or
androgynous management behaviors will be more balanced or only slightly skewed
towards males.
To simplify, the hypotheses are summarized as follows:
H(l): There are clear dominant styles in organizations as measured by the
similarities of top management in regard to sex-role identification and decision
style. Furthermore, these styles are primarily masculine and
- 42 -

H(2): Analytic and directive decision styles are positively related to a
masculine sex-role identity.
H(3): Behavioral and conceptual decision styles are positively related to a
feminine or more androgynous sex-role identity.
H(4) Masculine-analytic/directive organizations, as defined by the majority of
top management, have fewer women in top management both as a function of
total numbers and in comparison to women at other levels of management
Androgynous/feminine-behavioral/conceptual organizations will have more
women in top management and will be more gender balanced.
H(5): Decision style and sex-role identity will discriminate between
organizations that have more women in top management and/or are more gender
balanced and organizations that have more men in top management and/or are
less gender balanced.
Using decision style and sex-role identity as the key personal and structural
predictors of representation of top management, a sample of Colorado executive public
managers was analyzed. These predictive variables, which represent developmental,
social, cultural and institutional influences on managers, provide the framework in this
attempt to understand the barriers that are keeping women out of upper management in
state government.
- 43 -

The proposed design for this study is passive correlational and descriptive in
nature. It entails a one-time survey of top level managers of the State of Colorado.
Variables Under Study
Specifically, the independent variables under study are:
Decision Style as measured by the DSI (1987).
Sex-Role Identity as measured by the Bern Sex Role Inventory
Organizational and Individual Demographic Factors (position, size of
organization, sex, age, race, education, marital status, time in career, job
The dependent variable is representation of women at top management levels in
the departments studied but is measured in two ways:
Proportion of women in top management because of limitations of
using proportions as a dependent variable in a linear model, limitations in the
variance by departments, and the need to aggregate departments due to limited
numbers of women in upper management levels, a dichotomous classification
of departments was obtained by categorizing departments as low representation
or as high representation. In low representation departments, women represent
- 44 -

less than 25% of top management; in high representation departments, women
represent 25% or more of top management. The cutoff of 25% was determined
based upon the range of representation in state governments as measured by
Kelly (Kelly et al., 1991; Hale and Kelly, 1989) and based upon the
distribution of representation in Colorado.
Representation ratio of top management is measured by the ratio of the
proportion of women in top management as compared to the proportion of
women in middle management. This is an important distinction because a
measurement of representation alone is a measure of block equality.
Representation at the top as compared to representation at other management
levels is a measure of segmented equality. For example, a department may have
high block representation but women may be represented in other levels of
management at a much higher proportion. Therefore, segmented representation
is much lower. This is a description of an unbalanced organization and may be
as indicative of representation problems as an organization that has low
Using Kanter's schema of gender ratios (1977a, modified for use here),
a ratio of .9 to >1.00 is a balanced organization, in which the proportion of
female top managers matches the proportion of female middle and lower
managers; a ratio of .5 to .89 is a skewed organization, in which there are
somewhat fewer women at the top than at middle and lower levels; a ratio of
<.5 is a tilted organization, in which there are significantly fewer women at the
top than at middle and lower levels of management Because of the lack of
balanced organizations in the population, balanced and tilted organizations were
combined into one group to create a dichotomous variable.
- 45 -

The State of Colorado was chosen as the sampling frame for this study because
of its proximity to the University of Colorado and because of the availability of data and
information on managers in state government. Only administrative departments were
included to rule out possible mitigating differences between administrative agencies and
legislative/judicial agencies. Administrative departments were chosen specifically to
control for possible differences between career civil servants and elected/appointed
Top managers were defined as department heads, division heads and section
heads of organizations who were either classified employees or appointed executives.
These managers were assumed to have both budgetary and human resource
responsibilities. In the Colorado classification system, these managers have a
management class rating of 99.
There are 506 such managers in the Colorado government representing 21
departments. Table 3.1 lists the departments and the number of top managers by sex.
Because of the relatively low number of managers at this level, it was decided not to draw a
random sample of managers but to use the population as a sample frame. Indeed, it was
hoped that at least half of the managers (250) would respond to ensure a sample large
enough to analyze.
- 46 -

Table 3.1
Gender Breakdown of Top Managers in Colorado
Administrative Departments
Department Total Top Managers Total Female Total Male % Female
Deoartment of Agriculture 7 0 7 0%
Department of Military Affairs 2 0 2 0%
Department of Highways 47 1 46 2%
Department of Corrections 44 3 41 7%
Department of Public Safety 12 1 11 7%
Department of Administration 23 2 21 9%
Department of Natural Resources 58 6 52 19%
Department of Labor and Employment 16 3 13 21%
Department of Revenue 35 7 28 23%
Department of Local Affairs 12 2 19 25%
Department of Institutions 65 16 49 25%
Department of Law 3 1 2 25%
Department of Personnel 8 2 6 25%
Department of Social Services 44 14 30 29%
Department of Regulatory Affairs 23 8 15 33%
Department of Health 42 13 29 33%
Legislative (administration onlvi 13 5 8 36%
Department of Education 25 11 14 44%
Department of State 2 1 1 50%
Office of the Governor 22 15 7 65%
Department of Treasury 3 2 l 75%
Total 506 111 395 22%
- 47 -

The population data on the managers were collected in February, 1990 using the
state's computer database of employees, with the assistance of the Department of
Personnel. The data collected on each class 99 manager included name, contact
information, sex, age, total number of years with the state, and whether the employee
was a classified or an exempt employee. Even exempt administrators at this level
carried a Class 99 rating regardless of their exemption from state classification. Table
3.2 shows the distribution of departments by the dependent variables for this study; 10
departments fall into the low representation category and 11 departments fall into the
high representation category. Twelve departments fall into the tilted representation ratio
category and nine departments fall into the balanced/skewed (B/S) representation ratio
category. Total numbers of managers are also distributed relatively evenly between the
dependent variable measuring proportion (N iow = 256; N high = 250) but less so in
the dependent variable measuring ratio (N tilted = 315; N b/S = 191).
- 48 -

Breakdown of Colorado Departments Bv Dependent
Department Proportion Female DeDtreD* Gender Ratio** Deptrat***
Deoartment of Agriculture 0% Low .00 Tilted
Deoartment of Military Affairs 0% Low .00 Tilted
Deoartment of Highways 2% Low .11 Tilted
Deoartment of Corrections 7% Low .15 Tilted
Deoartment of Public Safetv 7% Low .15 Tilted
Deoartment of Administration 9% Low .15 Tilted
Deoartment of Natural Resources 19% Low .59 B/S
Department of Labor and EmDlovment 21% Low ,39 Tilted
Department of Revenue 23% Low .49 Tilted
Deoartment of Local Affairs 25% Low .48 Tilted
Deoartment of Institutions 25% High .42 Tilted
Deoartment of Law 25% High .54 B/S
Deoartment of Personnel 25% High .32 Tilted
Department of Social Services 29% High .44 Tilted
Deoartment of Regulatorv Affairs 33% High .93 B/S
Department of Health 33% High .60 B/S
Legislative (administration onlvl 36% High .68 B/S
Deoartment of Education 44% High .63 B/S
Deoartment of State 50% High .90 B/S
Office of the Governor 65% High 1.15 B/S
Deoartment of Treasury 75% High - -8.3 , B/S
* Deptrep = classification of representation of women in top management. Less than
25% is low representation, 25% or more is high representation. ** Representation
ratio is calculated by dividing the number of women at top by the number of women at
middle and lower management levels (excluding front-line supervisors). *** Deptrat
= classification of ratios. Less than .50 is tilted, .50 to .89 is skewed, .90- 1.00+ is
balanced (B/S = balanced or skewed).
- 49 -

Data Collection
Data were collected using a mail survey technique. Potential respondents were
sent the survey battery, a cover letter stating the purpose of the survey and a stamped,
return envelope in July, 1991. The battery included Rowe and Mason's (1987)
Decision Style Index to measure decision style, The Bern Sex-Role Inventory (BSRI,
short form; Bern, 1981) to measure sex-role identity, and measures designed by the
investigator to measure other individual and organizational variables (see Appendix A
for survey instruments). All data were scaled on either interval or, when possible, on
ratio scales to simplify data analysis.
Respondents were asked to reply within three weeks of receiving the survey
instruments. Response rates were monitored closely over this time period. Because
response rates were initially low for women overall and for both men and women in
some departments, follow-up phone calls were made to all women in the sample frame
and to men from the departments showing initial low response rates.
Overall, a total of 319 surveys were returned. Eleven were discarded because
of confusion with instructions or because the surveys were not complete, leaving a total
of 308 completed surveys. This represents an overall response rate of 61%. In total,
70 women (63% response rate) and 238 men (60% response rate) returned usable
surveys. The proportion of men to women in the sample (77%/23%) almost exactly
matched the proportion of men and women in the population (78%/22%). Response
rates varied significantly by department as can be seen in Table 3.3.
- 50 -

Table 3.3
Response Rates
Department Total Respondents Male Female Response Rate
Department of Agriculture 4 3 1 57%
Department of Military Affairs 2 2 0 100%
Department of Highways 33 33 0 75%
Department of Corrections 26 24 2 60%
Department of Public Safety 9 8 1 75%
Department of Administration 16 12 4 70%
Department of Natural Resources 41 38 3 57%
Department of Labor and Employment 12 IQ 2 75%
Department of Revenue 19 14 5 54%
Department of Local Affairs 5 5 0 42%
Department of Institutions 35 25 10 54%
Department of Law 2 1 1 67%
Department of Personnel 7 4 3 88%
Department of Social Services 26 19 7 60%
Department of Regulatory Affairs 10 5 4 44%
Department of Health 19 13 6 45%
Legislative (administration onlvl 9 4 5 69%
Department of Education 18 12 6 72%
Department of State 2 1 1 100%
Office of the Governor 12 3 9 55%
Department of Treasury 1 1 0 33%
Total 2JLS 70 61%
- 51 -

Unfortunately, the time lag between collecting the population data (February, 1991)
and collecting the survey data (July, 1991) produced some problems with comparing the
sample to the population. In the five to six month lag, a few managers retired, were
promoted or moved on to other jobs. Therefore, the sample data do not always match the
population data (e.g., the Department of Agriculture had no women in upper levels in
February but one woman responded in July). To check for problems with sample
diffusion, the data were examined to ensure that all respondents were top managers. The
data confirmed that respondents were indeed in top management positions. The personnel
changes were few and did not affect the overall classification of the affected departments
into dependent variable groups and were, therefore, disregarded.
To check for selection bias in the sample, sample statistics were compared with
the available population statistics. As can be seen in Table 3.4, the sample appears to
be representative of the population. Indeed, the high response rate for a mail survey
indicates that the sample should be representative of the population.
Table 3.4 Sample Representativeness
Variable Population Sample
Average Age 47.22 47.54
Average Years with State 14.08 16.32
Percent Male 78% 77%
Percent Female 22% 23%
Classified 84%
Unclassified 16%
- 52 -

Survey Instrument Scoring
The Bem Sex-Role Inventory fBSRD. The BSRI (short form) consists of 30
items; 10 items that are associated with a masculine identity, 10 items associated with a
feminine identity and 10 filler items used to reduce the face validity of the instrument
Respondents rate the degree to which the item describes them using a 1-7 scale.
The BSRI yields three related scores. First, the items in the instrument
associated with a masculine identity are tallied and averaged to arrive at the Masculine
score. Second, the items associated with a feminine identity are tallied and averaged to
arrive at the Feminine score. The filler items are disregarded. Finally, the masculine
score is subtracted from the feminine score to determine the Bemscore or the total scale
Interpreting the BSRI can be done either using normative data developed by the
instrument's author or by using the sample as the normative base for determining where
respondents fall on the dimensions. Because this sample is somewhat different from
Bern's normative sample (consisting largely of college students and the general public
in the 1970s and early 1980s; see Bem, 1981), norms were determined using this
sample. Figure 3.1 indicates the four cell framework that results from using a median
split on both the masculine and feminine scales to create the sex-role identity typology.
The respondents were divided at the weighted median (weighted to give equal weight to
both male and female respondents) on both scales which resulted in a fourfold
classification designating respondents as either masculine, feminine, androgynous or
Interpreting the Bemscore, the total sex-role identity score, can only be done by
comparing derived scores to normative data. Essentially, a negative number indicates
greater masculinity, a positive number indicates greater femininity. Scores falling in the
range of -1 to -3 are considered to be masculine; scores falling between 1- to +1 are
considered to be androgynous and scores falling between +1 and +3 are considered to
be feminine. Bem, however, recommends using the median split typology instead of
- 53 -

this interpretation as other research has discovered problems with interpreting the
instrument using a result that represents a continuum rather than a matrix (Bern, 1981).
Figure 3.1 Interpreting the BSRI
Masculine Score
Below Median = Above
Median 5. 0 Median
Feminine Score
Below Median Undifferentiated Masculine
Median = 4.5
Above Median Feminine Androgynous
Adapted from The Bern Sex-Role Inventory Professional Manual. (Bern, 1981)
The Decision Style Inventory ('DSD. The DSI consists of twenty questions,
each with four responses, that concern typical situations facing executives. Although
each of the four responses to a particular question may appear equally desirable, the
instrument is designed to force individuals to rank or differentiate among them. There
are no right or wrong answers; the scores reflect preferences for different responses.
Each response is ranked either a 1,2,4, or 8, with the highest number indicating the
greatest degree of preference.
Once all the responses had been ranked, the scores in each column were totaled,
yielding four scores which represented the four decision dimensions: directive,
analytical, conceptual or behavioral. These scores were then classified as preferences
for using the style. Using standard deviations from normative data (N=10,000) the
- 54 -

range of scores for four strengths of preferences were determined as indicated in Figure
Figure 3.2 Range of Scores on the DSI
Least Preferred Backup Dominant Verv Dominant
Directive 20 to 67 68 to 81 82 to 89 90 to 160
Analytical 20 to 82 83 to 96 97 to 104 105 to 160
Conceptual 20 to 72 73 to 86 87 to 94 95 to 160
Behavioral 20 to 47 48 to 61 62 to 69 70 to 160
A decision style can be a least preferred style indicating that one rarely or never
uses that style; a backup style, indicating that the style is used as a backup to the
dominant styles if the situation warrants; or a dominant/very dominant style, indicating
that the style is the primary decision style of choice. In the normative samples, 31% of
respondents were classified as least preferred on any given style, 38% as backup, 15%
as dominant and 16% as very dominant (Rowe and Mason, 1987).
Data Analysis Techniques
Data analysis was completed in two steps. First, simple descriptive statistics
were calculated on the independent variables to profile the respondents and to profile
the departments as defined by the two dependent variables (low versus mod/high
representation of women and balanced/skewed or tilted ratio). Simple, first-order
statistical tests were then performed (chi-square, t-tests) where needed to check for
statistically significant differences between groups. In addition, tests were done
between males and females to test for sex-differences among the independent variables.
- 55 -

These analyses provide an understanding of the sample in addition to addressing
Hypotheses 1-4.
In order to test Hypothesis 5, the interaction between and among the
independent variables, discriminant analyses were performed on both dependent
variables. Discriminant analysis is a statistical technique that allows one to study the
differences between two or more groups of objects with respect to several variables
simultaneously to statistically distinguish between two or more groups. Discriminant
analysis permits the determination of: 1) which, if any, of the independent variables are
useful in predicting the dependent variables; 2) how these predictive variables might be
combined into a mathematical equation to predict the most likely outcome; and 3) the
accuracy of the derived equation, or the goodness of the prediction model (Klecka,
1980). As the dependent variables are dichotomous, two-function discriminant
analyses were used to identify those independent variables that most discriminate
between departments that have low representation, and/or low representation ratios, and
those departments that have high representation, and/or high representation ratios.
Both direct and stepwise methods were used. Results of stepwise analyses are
presented here.
- 56 -

Descriptive Findings
As Tables 4.1 and 4.2 indicate, the respondents are top level managers who
represent departments ranging from very small (2 employees) to very large (5,500
employees) and who are at the top of their division or department. In fact, 73% of the
respondents are at the top of their division while 31% are at the top of their department
Fifty-seven percent of respondents are within two steps of the top of their department.
The majority of the respondents have budgetary responsibility (89%), managing
budgets that are for the majority (76%) over $1 million.
Respondents indicated that their organizations perform a number of functions as
displayed in Table 4.1. Rated on a scale of 1-5 where 5 indicates that the department
performed the function a great deal, respondent's departments are most likely to
perform administrative, policy and technical support functions and less likely to
perform regulatory, social services and law enforcement functions.
- 57 -

Table 4.1
Overall Organizational Characteristics
Current Position Information
Median Mean Standard
Number of People Supervise 42 192.43 550.63
Number of People Directlv Supervise 6 7.22 11.96
Number of Men Directly Supervise 3 5.80 13.59
Number of Women Directly Supervise 2 4.13 7.95
Number of Steps to Top of Division 0 .39
Number of Steps to Top of Department 1 1.17
Percent of Responses *
Manage Budget 89%
Size of Budget
Under $50K 3%
$50K-$150K 4%
$151K-$500K 10%
$501K-$1 million 7%
$1 million-$5 million 31%
Over $5 million 45%
Departmental Functions*
Average Rating
Technical Support 4.09
Policy Making 3.94
Administrative Support 3.56
Regulatory 3.33
Social Services 2.29
Law Enforcement 2.29
*Respondent perceptions of the degree to which their organization performs the
functions measured on a 1-5 scale where 1 = not at all and 5 = a great deal.
- 58 -

Table 4.2 illustrates that these respondents are, on average, middle-aged (mean
age = 47.54), career-level bureaucrats who have devoted the majority of their careers to
public service (mean time in public administration = 15.81; mean time in career =
21.94). The majority (81%) are married, live in a household with another adult (86%)
and have approximately two dependents living at home. The majority are male (77%)
and, furthermore, the majority primarily identify with an Anglo racial identity (91%).
Almost all of these executives are college-educated with the majority (75%)
holding an advanced degree. Twenty-five percent hold a doctorate or a professional
degree beyond the Masters. A small percentage are veterans (24%) and an even smaller
percentage are retired from the military.
The majority of respondents work at least 40 hours a week, with most working
between 40-50 hours (63%). Thirty-two percent work over 51 hours per week.
Overall, the respondents are very satisfied with their careers (mean satisfaction = 4.04,
1-5 scale) and see themselves as successful (mean success = 4.55,1-5 scale).
- 59 -

Table 4.2
Overall Individual Demographics
Median Mean Standard
Number of Years in Current Position 4 5.15 4.61
Number of Years with the State 17 16.32 9.02
Number of Years in Public Admin 16 15.81 7.85
Number of Years in Career 21 21.94 7.19
Age 47 47.54 7.05
Number of Dependents Living at Home 2 1.90 1.04
Satisfaction (1 = not at all satisfied,
5 = extremely satisfied) 4 4.04 .867
Perception of Success
(1 = not successful, 5 = successful) 5 4.55 .577
Percent of Responses
Male 77%
Female 23%
Primary Racial Identity
African-American 3%
Anglo 91%
Asian 1%
Hispanic 5%
Native American <1%
Marital Status
Never married 6%
Married 81%
Divorced/separated 11%
Widowed 2%
Living Situation
Living alone 12%
Living with other adults and
dependents 86%
Sole adult with dependents 2%
- 60 -

Table 4.2
Overall Individual Demographics (continued)
Percent of Responses
High School Graduate 2%
Some College or College Grad 23%
Up to M.A./M.S. 50%
Up to Ph.D./professional degree 23%
Post-Doctoral 2%
Military Service
Veteran 24%
Retired from the Military 2%
Time Spent Working During the Average Week
32 hours or less 1%
33-40 hours 4%
41-50 hours 63%
51-60 hours 32%
- 61 -

There are no clear dominant sex-role identities as shown in Table 4.3. Overall,
as indicated by the total score (Bemscore), the group falls within the androgynous
range (mean Bemscore = -.527) although respondents are almost equally split between
the four sex-role categories with a slightly higher proportion in the masculine category.
Table 4.3 Overall Sex-Role Identity
Average Masculine Score
Average Feminine Score
Average Sex-Role Identity Score
Unweighted Weighted
Mean Meehan Median
4.50 5.05 5.10
4.47 4.50 4.50
-.527 -.500
Percent of Respondents
Sex Role Categories
Masculine 28%
Feminine 25%
Androgynous 25%
Undifferentiated 22%
- 62 -

Average scores on the DSI subscales fall within the range of the averages of the
normative samples as determined by the instrument's creators (Rowe and Mason,
1987) and shown in Table 4.4. Although there is considerable variance in the
distribution of dominant or very dominant styles, there are more dominant/very
dominant analytical decision styles in this sample as compared to Rowe and Mason's
(1987) normative data (41% in current sample; 31% in normative data).
Table 4.4 Overall Decision Style
Average Scores on Style Scales
Respondents______Range Normative Mean
71.010 20-160 74 '
91.683 20-160 89
82.709 20-160 79
58.845 20-160 54
Preferences for Styles
Percent of Respondents
Directive Analytical Conceptual Behavioral
Least Preferred 39% 29% 26% 37%
Back-up 40% 30% 40% 33%
Dominant 14% 14% 13% 14%
Very Dominant 7% 27% 21% 16%
Combined Dominant and Verv Dominant 21% 41% 34% 30%
- 63 -

Respondents were also asked to respond to a series of questions related to their
perception of gender and race-related employment issues that currently affect state
government Respondents tended to agree with statements indicating that employees
should reflect population parameters in regard to race and sex. In addition, respondents
agreed that women and minorities should be given preference over white men in hiring,
as can be seen in Table 4.5.
Respondents were less likely to agree that white males have been discriminated
against or that they have been personally discriminated against. In addition, the
majority of respondents (76%) indicated that they were not often left out of informal
decision-making situations.

- 64 -

Table 4.5
Perceptions _of Gender/Race Issues in State
Agreement with Policy Statement
Average Rating*
Employees in state government should reflect the ethnic/gender makeup of the population. 3.56
Minorities should receive preference where applicants are of equal ability and minorities are underrepresented. 3.54
Employees in state government should reflect the gender makeup of the population. 3.46
Females should receive preferences where female and male applicants are of equal ability and females are underrepresented. 3.36
White males in this state are sometimes discriminated against in hiring and promotion. 2.95
I personally believe that I have been discriminated against in either hiring or promotion. 1.74
* Rated on a 1-5 scale where 1 = strongly disagree and 5 = strongly agree.
How often do you believe you are left out of informal discussion with colleagues when
important decisions are made?
Percent of Responses
Often 8%
Sometimes 16%
Occasionally 23%
Rarely 46%
Not at All 7%
- 65 -

Comparing Women and Men Managers
One of the key goals of this study was to compare women and men managers
on the series of variables measured. The following section addresses the similarities
and differences between women and men managers, focusing specifically on the
differences. Simple, first-order statistical tests were performed to test for statistically
significant differences between men and women. For continuous variables, pooled
variance, two-tailed t-tests were used to test; for categorical variables, chi-square tests
of independence were used. Statistically significant differences are indicated by the **
in the tables. Any difference with a alpha probability of less than 5% (p<.05) was
accepted as statistically significant
Although it appears that men, on average, tend to be in charge of larger
departments or divisions than women, this difference is not statistically significant
Men and women, excluding the outliers, tend to manage departments that are roughly
the same size.
In regard to their position, roughly the same percentage of men and women are
at the top of their division (male = 71%; female = 80%) as indicated in Table 4.6.
However, fewer men are at the top of their department than women (28% of men at
top, 41% of women at top (mean male = 1-27 steps; mean female = -89 steps t = 1.68
p = .02). This is not reflective of the total population of department heads in the State
of Colorado but, more likely, a reflection of sampling bias in that more women heads
of departments responded to the survey than did men.
Men and women rated the functions of their department the same except for the
technical support and social services function. Men perceived that their organizations
performed a greater degree of technical functions than did women (mean male = 4.19;
mean female = 3.751 = -2.91 p = .004). On the other hand, women perceived that
their departments performed more social services functions than did men (mean male =
2.16; mean female = 2.72 t = 2.60 p = .01).
- 66 -

Table 4.6
Women and Men Organizational Characteristics
Current Position Information
Number of People Supervise
Number of People Directly Supervise
Number of Men Directly Supervise
Number of Women Directly Supervise
Number of Steps to Top of Division
Number of Steps to Top of Department* **
Male Mean (Median) J Female lean (Median)
215(49) 115(30)
7(5) 7(7)
6(3) 4(3)
4(2) 4(4)
.45 .18
1.27 .89
Percent of Responses
Male Female
Manaee Budget 91% 84%
Size of Budget
Under $50K 4% 2%
$50K-$150K 4% 5%
$151K-$500K 9% 13%
$501K-$1 million 7% 9%
$1 million-$5 million 31% 30%
Over $5 million 45% 41%
Departmental Functions*
Average Rating
Males Females
Technical Support** 4.19 3.75
Policy Making 3.90 4.09
Administrative Support 3.66 3.29
Regulatory 3.36 3.21
Social Services** 2.16 2.72
Law Enforcement 2.37 2.04
*Respondent perceptions of the degree to which their organization performs the
functions measured on a 1-5 scale where 1 = not at all and 5 = a great deal.
- 67 -

Women respondents are younger (mean male = 48.36; mean female = 44.86 t = -
3.74 p < .001), have been in their current position for fewer years
(mean maie = 5.68; mean female = 3.39 t = -3.74 p < .001), have been with the state for
fewer years (mean male = 17.75; mean female = 11.52 t = -5.33 p < .001), and have been
in their career for fewer years (mean male = 22.84; mean female = 18.93 t = -4.12
p < .001).
Women and men respondents also differ on other key demographic variables.
Fewer women respondents are married (men = 88%; women = 59%,
chi square = 31.26 p < .001), more women live alone (men = 8%, women = 24%; chi
square = 23.44 p =.0001), and more women hold higher degrees (p < .05) as is seen in
Table 4.7. Also, as would be expected, fewer women are veterans than are men.
- 68 -

Women and Men Individual Demographics
Male Mean (Median-) Female Mean Median-)
Number of Years in Current Position** Number of Years with the State** Number of Years in Public Admin.** Number of Years in Career** Age** Number of Dependents Living at Home 5.68(4) 17.75(18) 17.06(17) 22.84(22) 48.36(48) 1.92(2) 3.39(2) 11.52(11) 11.63(12) 18.93(17) 44.86(43) 1.96(2)
Satisfaction (l=not at all satisfied, 5 = extremely satisfied) 4.05(4) 4.01(4)
Perception of Success (l=not successful, 5=successful) 4.57(5) 4.49(5)
Percent o f Responses
Male Female
Primary Racial Identity
African-American 2% 5%
Anglo 92% 87%
Asian 1% 1%
Hispanic 5% 7%
Native American <1%
Marital Status**
Never married 4% 16%
Married 88% 59%
Divorced/separated/widowed 8% 25%
Livine Situation**
Living alone 8% 24%
Living with other adults and
dependents 89% 75%
Sole adult with dependents 3% 1%
**p < .05
- 69 -

iJ. Women and Men Individual Pemo2raphics
Percent o Responses
Male Female
High School Graduate 3% 0%
Some College or College Grad 26% 14%
Up to M.A./M.S. 48% 56%
Up to Ph.D./professional degree 21% 29%
Post-Doctoral 2% 1%
Total Post B.A**. 71% 86%
Militarv Service**
Veteran 30% 4%
Retired from the Military 2% 0%
Time Spent Working During the Average Week:
32 hours or less <1% 4%
33-40 hours 5% 1%
41-50 hours 64% 59%
51-60 hours 31% 35%
**p < .05
- 70 -

Men and women also differ with respect to sex-role identity (see Table 4.8)
with women's Bemscore reflecting a somewhat higher degree of masculinity (although
still in the androgynous range) than do men's (mean male = --4638; mean female = -
.7386 t = -2.06 p = .04). In addition, more women (34%) have a masculine sex-role
identity than do men (22%). Conversely, more men (30%) have a feminine sex-role
identity than do women (20%).
Table 4,8 Women and Men Sex-Role Identity
Total Resoonses
Male Female
Average Masculine Score 4.96 5.12
Average Feminine Score 4.50 4.38
Average Sex-Role Identity Score** -.4638 -.7386
Percent of Respondents
Male Female
Sex Role Categories
Masculine** 22% 34%
Feminine 30% 20%
Androgynous 25% 26%
Undifferentiated 23% 20%
**p <.05
Women and men are the same in regard to decision style with one exception.
More women fall into the least preferred category for the Behavioral style than do men
(men = 33%, women = 51%; chi square = 9.37 p = .02) as can be seen in Table 4.9.
In addition, women are more likely to have a directive dominant or very dominant
decision style (31%) than are men (19%).
- 71 -

Table 4.9
Women and Men Decision Style
Average Scores on Style Scales
Total Responses
Male Female
Directive 70.48 72.86
Analytical 91.28 93.04
Conceptual 83.08 81.48
Behavioral 55.57 52.42
Preferences for Styles
Percent of Female Respondents
Directive Analytical ConceDtual Behavioral
Least Preferred 37% 27% 28% '51%
Back-up 32% 28% 43% 22%
Dominant 22% 15% 11% 16%
Very Dominant 9% 30% 18% 11%
Combined Dominant and' Verv Dominant 31% 45% 29% 27%
Percent of Male Respondents
Directive Analytical Conceptual Behavioral
Least Preferred 39% 29% 25% 33%
Back-up 42% 31% 39% 36%
Dominant 12% 13% 14% 13%
Very Dominant 7% 26% 22% 18%
Combined Dominant and Verv Dominant 19% 39% m> 31%
- 72 -

In regard to race and gender related employment issues, women and men rated most
factors similarly as can be seen in Table 4.10. However, men more than women are likely
to agree that White men are sometimes discriminated against (mean male = 3.14; mean
female = 2.31 t = -5.22 p < .001) and women more than men are likely to agree that they
have personally been discriminated against at some time (mean male =1.61;
mean female = 2.19 t = 3.56 p < .001).
- 73 -

Table 4.10
Women and Men Perceptions of Gender/Race
Issues in State Government
Agreement with Policy Statement
Average Rating*
Male Female
Employees in state government should reflect the ethnic/gender makeup of the population. 3.52 3.67
Minorities should receive preference where applicants are of equal ability and minorities are underrepresented. 3.48 3.77
Employees in state government should reflect the gender makeup of die population. 3.36 3.67
Females should receive preferences where female and male applicants are of equal ability and females are underrepresented. 3.29 3.57
White males in this state are sometimes discriminated against in hiring and promotion.** 3.14 2.31
I personally believe that I have been discriminated against in either hiring or promotion.** 1.61 2.19
* Rated on a 1-5 scale where 1 = strongly disagree and 5 = strongly agree.
How often do you believe you are left out of informal discussion with colleagues when
important decisions are made?
Percent o f Responses
Male Female
Often 9% 4%
Sometimes 15% 17%
Occasionally 23% 23%
Rarely 46% 49%
Not at All 7% 7%
**p < .05
- 74 -

Comparing Organizations
In classifying departments as either low or high for the representation dependent
variable or as tilted or balanced/skewed (B/S) for the representation ratio dependent
variable, individual responses were aggregated up to the department level. Department
level responses were then aggregated up to the dependent variable level. As was
shown in Table 3.2, every department is classified into the two groups of both
dependent variables.
Before moving on to a discussion of the tests of the hypotheses, it is interesting
to look at the dependent variables in terms of organizational characteristics. Table 4.11
lists a select set of variables by respondents from low representation departments and
those from high representation departments. Table 4.12 lists the same for the
dependent variable measuring the balance of women across levels.
As can be seen on both tables, there are few differences between types of
respondent groups on most of the selected variables except on several of the
organization function variables.
In total, there are 167 respondents from low representation departments and 141
respondents from high representation departments; 193 respondents are from tilted
departments and 115 from balanced/skewed departments.
- 75 -

Table 4.11
Select Organizational Characteristics bv
Representation at Tod
Current Position Information
Low MeanfMedianl High Mean (Median)
Number of People Supervise 249(62) 136(30)
Number of People Directlv Supervise 8(5) 6(6)
Number of Steps to Top of Division .69 .28
Number of Steps to Top of Department 1.36 .99
Departmental Functions*
Average Rating
Low High
Technical Support 4.18 3.99
Policy Making 3.86 4.04
Administrative Support 3.66 3.50
Regulatory 3.40 3.25
Social Services** 1.91 2.78
Law Enforcement*** 2.64 1.86
^Respondent perceptions of the degree to which their organization performs the
functions measured on a 1-5 scale where 1 = not at all and 5 = a great deal.
** t=-5.01 p=.000
*** t= 4.31 p=.000
- 76 -

Table 4.12
Select Organizational Characteristics bv
BeftresfiRlatifiD- -Rati a
Current Position Information
Mean (Median!
Number of People Supervise** 263(74)
Number of People Directly Supervise 8(6)
Number of Steps to Top of Division . 61
Number of Steps to Top of Department 1.27
Departmental Functions*
Average Rating
Tilted B/S
Technical Support 4.10 4.17
Policy Making*** 3.83 4.14
Administrative Support 3.72 3.35
Regulatory**** 3.07 3.78
Social Services***** 2.66 1.67
Law Enforcement 2.20 2.48
*Respondent perceptions of the degree to which their organization performs the
functions measured on a 1-5 scale where 1 = not at all and 5 = a great deal.
** t=2.96 p=.003
*** t=-2.31 p=.022
**** t=-3.99 p=.000
***** t=5.52 p=.000
- 77 -

Testing the Hypotheses
The results are presented in three sections. The first section addresses
Hypotheses 1 and 4, both of which are related to dominant styles in organizations. The
second section addresses hypotheses 2 and 3, both of which are related to the
relationships between the DSI and the BSRI. The final section addresses hypothesis 5
which tests the ability of the variables measured to discriminate between the groups and
to predict the outcomes of the dependent variables.
Hypothesis 1 states that there will be clear dominant styles in organizations as
measured by the DSI and the BSRI. Furthermore, this hypothesis states that the styles
are primarily masculine and analytical/directive. Hypothesis 4 states that there is a
relationship between decision style/sex role identity and the numbers of women at top
levels of management as measured by the dependent variables.
As already shown in Table 4.3 and Table 4.4, there are no clear, dominant sex-
role styles or decision styles in the sample. Rather, respondents appear to be arrayed
across the style choices, almost equally distributed in regard to sex-role style and only
somewhat more representative of an analytical decision style. Furthermore, when
respondents were aggregated into organization groups, there were no clearly dominant
styles within the groups (see Tables 4.13-4.16).
Tables 4.13 and 4.14 indicate that respondents representing these department
classifications fall almost equally into the four sex-role identity categories. There is no
clear dominant sex-role identity. In addition, there are no statistically significant
differences between respondents representing low and/or tilted organizations and
respondents representing high and/or balanced/skewed organizations in regard to sex-
role identity, indicating a lack of relationship between sex-role identity and the
dependent variables.
- 78 -

Table 4.13
Sex-Role Identity bv Representation at Tod
Average Masculine Score
Average Feminine Score
Average Sex-Role Identity Score
Total Responses
Low High
5.01 4.97
4.47 4.47
-.549 -.494
Percent o f Respondents
Low High
Sex Role Categories
Masculine 24% 25%
Feminine 27% 29%
Androgynous 27% 23%
Undifferentiated 22% 23%
Table 4.14 Sex-Role Identity bv Representation Ratio
Average Masculine Score
Average Feminine Score
Average Sex-Role Identity Score
Total Responses
Tilted B/S
5.01 4.97
4.48 4.56
-.529 -.515
Percent of Respondents
Sex Role Categories Tilted B/S
Masculine 23% 27%
Feminine 26% 30%
Androgynous 29% 20%
Undifferentiated 22% 23%
- 79 -

Tables 4.15 and 4.16 indicate the distribution of decision style by dependent
variable group. Except for the slightly higher proportion of analytical styles and the
smaller proportion of directive styles, the respondents are distributed almost equally
across the categories. Also, there are no statistically significant differences between
respondents representing low and/or tilted organizations and respondents representing
high and/or balanced/skewed organizations in regard to preferred decision style. These
results indicate a lack of relationship between decision style and the dependent
- 80 -

Table 4.15
Decision Style bv Representation at Tod
Preferences for Styles
Percent of Respondents in Low Rep Departments
Directive Analytical Conceptual Behavioral
Least Preferred 35% 31% 27% 34%
Back-up 46% 27% 42% 36%
Dominant 14% 13% 13% 13%
Very Dominant 5% 29% 18% 17%
Combined Dominant and Verv Dominant 19% 32% 31% 30%
Percent of Respondents in High Rep Departments
Directive Analytical Conceptual Behavioral
Least Preferred 42% 25% 25% 40%
Back-up 34% 34% 38% 28%
Dominant 14% 16% 13% 16%
Very Dominant 9% 25% 24% 16%
Combined Dominant and Verv Dominant 23% 41% 37% 32%
- 81 -

Table 4.16
Decision Style bv Representation Ratio
Preferences for Styles
Percent of Respondents in Tilted Departments
Directive Analytical Conceptual Behavioral
Least Preferred 39% 26% 25% 37%
Back-up 42% 32% 42% 32%
Dominant 15% 14% 12% 13%
Very Dominant 4% 28% 21% 18%
Combined Dominant and Verv Dominant 19% dm 31% 30%
Percent of Respondents in High Rep Departments
Directive Analytical Conceptual Behavioral
Least Preferred $ OO co 33% 28% 38%
Back-up 00 CO 28% 36% 32%
Dominant 13% 14% 15% 15%
Very Dominant 11% 25% 21% 15%
Combined Dominant and Verv Dominant 22% 36% 30%
- 82 -

Using average scores on the BSRI, departments were classified by the
dominant sex-role identity within the department. Table 4.17 lists the dominant sex-
role identity and percent dominant/very dominant responses on the DSI by department
There is a great deal of variability across departments in regard to sex-role identity and
decision style. In addition, dominant sex-role identity and decision-style appear to be
unrelated to the number of women in top management or to the gender balance of top
- 83 -

% Dominant and Very Dominant
Department BSRI Directive Analytical Conceptual Behavioral Deptrep Deptrat
Military Affairs Andro 50% 50% Low Tilted
Highways Andro 24% 33% 24% 48% Low Tilted
Pnhlii* Qafpty Andro 11% 22% 47% 33% Low Tilted
Institutions Andro 17% 31% 45% 22% High Tilted
Labor and Employ Femin 8% 50% 58% 8% Low Tilted
Revenue Femin 16% 53% 32% 31% Low Tilted
Health Femin 10% 37% 47% 26% High B/S
Corrections Femin 16% 55% 28% 27% Low Tilted
Education Femin 23% 39% 44% 56% High B/S
Office of Gov Masc 40% 33% 34% . 8% High B/S
Administration Masc 44% 75% 12% 6% Low Tilted
Local Affairs Masc 60% 60% Low Tilted
Regulatory Affairs Masc 50% 30% 10% 50% High B/S
Legislature Masc 22% 88% 33% 11% High B/S
State Masc 50% 100% High B/S
Natural Resources Undiff 20% 32% 32% 30% Low B/S
Social Services Undiff 20% 39% 30% 42% High Tilted
Personnel Undiff 29% 58% 14% 43% High Tilted
Law Undiff 50% 50% High B/S
Agriculture Undiff 25% 25% Low Tilted
Treasury Undiff 100% 100% High B/S
Table 4.17 Sex-Role Identity and Decision Style By Department

Hypotheses 2 and 3 addressed the direct relationships between decision style
and sex-role identity. Specifically these hypotheses stated that analytic and directive
styles are positively related to the BSRI masculine score and that behavioral styles are
positively related to the BSRI feminine score. Table 4.18 lists the results of this
Both the directive and analytical styles are negatively related to the BSRI
feminine score (r directive = -18; r analytical = --13), positively related to the masculine
score (r directive = -30; r analytical = -10) and negatively related to the total score
(r directive = -.33; r analytical = -.16). Overall, the magnitude of these correlations are
relatively small but does indicate a moderate relationship between directive styles and a
masculine sex-role identity. The conceptual style is not related at all to the BSRI. The
behavioral style is positively related to the feminine scale (r behavioral = -34),
negatively related to the masculine scale (r behavioral = --34) and positively related to
the total score (r behavioral = -47).
Table 4,18 Relationship between the BSRI and the DSI
Pearson Correlation Coefficients
Feminine Masculine Bemscore
-.18** .30** .33**
-.13** .10** -.16**
.04 -.10** -.03
.34** -.34* .47**
- 85 -

Hypothesis 5 states that sex-role identity, decision style and some demographic
variables discriminate between organizations with high and low representations as well
as between organizations with tilted or balanced/skewed representation ratios.
Two separate discriminant analyses were performed on both dependent
variables. Prior to running the analyses, all potential independent variables were
analyzed for multicollinearity. After discarding some of the potential independent
variables because of multicollinearity, the following variables were included in all of the
discriminant analyses:
BSRI Feminine and Masculine Scores
DSI Scores
Size of Organization (number of people supervise)
Steps to Top of Division
Years in Position
Perceptions of Success
Satisfaction with Career
For the second set of analyses, the organization function variables were
included in an attempt to maximize the variance accounted for by the analysis.
Both direct and stepwise analyses were performed. The analysis parameters
were set to minimize the amount of residual variance using standard set-points for
variable exclusion and inclusion in the stepwise models. Stepwise discriminant
analysis results are presented here.
When the organization functions are excluded, the analyses which result have
canonical correlation coefficients of .34 to .35, explain 12% of the total variance in the
dependent variables, and correctly classify 62 to 63% of the cases (see Table 4.19).
The variables that discriminate the most between the dependent variable groups are
demographics, although conceptual and analytical styles do discriminate, to some
degree, between low and high representation organizations.
- 86 -

When organization function is included in the analyses (see Table 4.20), the
ability to discriminate increases significantly (residual variance = .69 to .71; percent
correctly classified = 73 to 71%). However, the effect of decision style drops even
lower. In fact, the organization function variables appear to be the most discriminating
variables except for age and education. The reader is reminded that organization
function is a measure of the respondent's perceptions of the degree to which their
organization performs specific functions rated on a 1-5, Likert-type scale.
- 87 -

Table 4.19
Results of Discriminant Analyses
(excluding organization function)
Dependent Variable = Proportion of Women at Top
Canonical Correlation .34
Residual Variance .88
Percent Correctly Classified 63%
Discriminant Function
Standardized Discriminant
Function Coefficients
Variable (absolute value)
Education .80
Age .50
People Supervise .49
Analytical Style .41
Conceptual Style .23
Perceptions of Success .21
Dependent Variable = Representation Ratio
Canonical Correlation .35
Residual V ariance . 8 8
Percent Correctly Classified 62%
Discriminant Function
Standardized Discriminant
Function Coefficients
Variable (absolute value!
Age .71
Education .67
Years in Position .45
# Steps to Top .33
People Supervise .25
Behavioral Style .23
- 88 -

Table 4.2Q
Results of Discriminant Analyses
(including organization function)
Dependent Variable = Proportion of Women at Top
Canonical Correlation . 5 5
Residual Variance .69
Percent Correctly Classified 73%
Discriminant Function
Standardized Discriminant
Function Coefficients
Variable (absolute valuel
Social Services Function .69
Education .57
Age ,49
Regulatory Function .41
Administration Function .29
Feminine Score .23
Law Enforcement Function .23
People Supervise .22
Analytical Style . 18
Years in Position . 18
Policy Function .15
Perceptions of Success . 13
Dependent Variable = Representation Ratio
Canonical Correlation .53
Residual Variance .71
Percent Correctly Classified 74%
Discriminant Function
Standardized Discriminant
Function Coefficients
Variable (absolute value!
Social Services Function .71
Law Enforcement Function . 5 8
Age .56
Education .41
Years in Position .34
Steps to Top .20
People Supervise . 18
Technical Function . 13
- 89 -

Summary of Hypotheses Testing
The findings of the hypotheses testing are as follows:
Hypothesis 1 stated that there would be clear dominant styles in
organizations as measured by the similarities of top management in regard to
sex-role identification and decision style. Furthermore, the hypothesis stated
that these styles would be primarily masculine and analytical/directive. The
findings indicate there are no clear, dominant sex-role styles or decision styles
in this population. Rather, respondents appear to be arrayed across the style
choices, almost equally distributed in regard to sex-role style and only
somewhat more representative of an analytical decision style. Furthermore,
when respondents were aggregated into dependent variable groups, there were
no clearly dominant styles within the groups. Respondents fall almost equally
into the four sex-role identity categories, regardless of their departmental
classification. Except for the slightly higher proportion of analytical styles and
the smaller proportion of directive styles (as compared to national averages), the
respondents are distributed almost equally across the decision-style categories.
In addition, there is a great deal of variability across individual departments in
regard to sex-role identity and decision style.
Hypothesis 2 stated that analytic and directive decision styles would be
positively related to a masculine sex-role identity. Hypothesis 3 stated that
behavioral and conceptual decision styles would be positively related to a
feminine or more androgynous sex-role identity. Analytical styles are only
nominally related to sex-role identities. Directive styles are moderately
positively related to masculine sex-role identity and, alternatively, negatively
related to the feminine scale. The conceptual style is not at all related to the
BSRI. The behavioral style is positively related to the feminine scale and
negatively related to the masculine scale.
- 90 -

Hypothesis 4 stated that masculine-analytic/directive organizations, as
defined by the majority of top management, would have fewer women in top
management both as a function of total numbers and in comparison to women at
other levels of management. Androgynous/feminine-behavioral/conceptual
organizations would have more women in top management and would be more
gender balanced. The findings show a great deal of variability in sex-role
identification and decision style across departments. Given this variance,
departments could not be classified into the categories listed in this hypothesis.
Whether or not a relationship exists between the dominant sex-role identity and
representation at the top cannot be measured here because of the significant
degree of variance in departments of style. Indeed, there is no relationship
because there is no one, dominant decision or sex-role style within or across the
organizations in this sample.
Hypothesis 5 stated that decision style and sex-role identity would
discriminate between organizations that have more women in top management
and/or are more gender balanced and organizations that have more men in top
management and/or are less gender balanced. The findings indicate that sex-
role identity and decision style have minimal discriminating ability in both first-
order and interaction relationships. The variables that appear to discriminate the
most between the dependent variable groups are demographics. Although
conceptual and analytical decision styles do discriminate to some degree
between low and high representation organizations, their ability to predict the
representation of women is minimal.
When organization function is included in the analyses, the ability to
discriminate increases significantly. In fact, the organization function variables
are the most discriminating variables after age and education.
- 91 -