Citation
Computer-assisted contraceptive choice

Material Information

Title:
Computer-assisted contraceptive choice a development and evaluation project
Creator:
McGill, William Lee
Publication Date:
Language:
English
Physical Description:
xiii, 78 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Contraceptives -- Interactive multimedia -- Research ( lcsh )
Contraceptives -- Interactive multimedia -- Evaluation ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 71-78).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by William Lee McGill.

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:
41470960 ( OCLC )
ocm41470960
Classification:
LD1190.L566 1998d .M45 ( lcc )

Downloads

This item has the following downloads:


Full Text
Computer-Assisted Contraceptive Choice:
A Development and Evaluation Project
by
William Lee McGill
B.A., University of Wisconsin-Milwaukee, 1976
M.S., Saint Louis University, 1980
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirement for the degree of
Doctor of Philosophy
Health and Behavioral Science
1998


UMI Number: 9921407
UMI Microform 9921407
Copyright 1999, by UMI Company. AH rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103


This thesis for the Doctor of Philosophy
degree by
William Lee McGill
has been approved by
Tamara Hoxworth
Date


DEDICATION
Dedicated to my mother for her constant and gentle assurances
throughout my all schooling and for her contributions to the audio
features of the computer program. Also to my wife for her support
and for the many extra tasks she assumed during the past four
years.


ACKNOWLEDGMENT
My thanks to the dissertation faculty for their support and
expertise in the preparation of this dissertation. Special thanks
to Clydette Stulp, my advisor, for her calming influence and
for helping to make the experience a positive one.


McGill, William Lee (Ph.D., Health and Behavioral Science)
Computer-assisted Contraceptive Choice: A Development and
Evaluation Project
Thesis directed by Assistant Professor Clydette Stulp
ABSTRACT
In the twenty years between 1970 and 1990, the human
population swelled by more than 1.6 billion. And although
unchecked population growth among humans poses the greatest threat
today to the survival of all species, the rate of unintended
pregnancy in the U.S. is among the highest of all developed
nations.
Explanations for this high rate of unintended pregnancy are
many, including dissatisfaction with the range of methods
available, misinformation concerning reproductive biology and
contraceptive methods, the effect of psychosocial factors and
contraceptive side-effects on contraceptive behavior, and the
foibles of risk estimation and decision-making. Given the
complexity of the contraceptive decision task, the risk of
unintended pregnancy is significant.
Among those using contraception (contraceptors), conception
risks vary with the method type, which range in effectiveness from
the near-perfect method of sterilization to fertility awareness
methods such as "rhythm." Focusing on intervention, the
continuum represented by the range of methods is best understood
in terms of the user vigilance required to avert an unintended
pregnancy. Indeed, nearly half of all unintended pregnancies
occur to those who are using contraception.
To improve contraception among contraceptors, one strategy
is to improve the initial choice of method by dispelling
misinformation, improving pregnancy risk estimation, and
facilitating rational decision-making. Thus, the aim of the
current study was to develop and pilot test an interactive
computer decision tool designed to assist with the choice of
contraceptive method.
Program development consisted of generating program
questions in the form of a questionnaire, constructing information
items, and of integrating questions, information items, and
multimedia into the computer program. Activities central to
program development included integration of theory and clinical
concerns. Resources included the recent relevant scientific
literature, a total of nine focus groups with family planning
professionals and patients, regular discussions with dissertation
faculty and family planning experts, and statistical analysis of
182 patient-completed Birth Control Choice Questionnaires.
Results of a pilot study suggested that the program was
user-friendly, improved family planning knowledge, and that it may
v


have influenced contraceptive choice. Based on the results,
continued development and evaluation of the program are warranted.
This abstract accurately represents the content of the candidate's
thesis. I recommend its publication. Q
Clydette Stulp_
vi


CONTENTS
CHAPTER
1. INTRODUCTION...........................................1
Purpose of the Project................................2
2. LITERATURE REVIEW.......................................3
The Problem of Unintended Pregnancy...................3
Determinants of Unintended Pregnancy..................4
Method Characteristics..........................4
Factors Affecting Method Choice and Adherence..6
User Characteristics............................7
Information Processing and Decision-Making...........10
Decision Task Variables........................11
Family Planning Assistance...........................12
Project Aim..........................................13
3. METHODS................................................15
Phase One............................................17
Materials and Procedure........................17
vii


CHAPTER
Site Selection..........................17
Development of the BCCQ.................17
BCCQ Draft One....................18
BCCQ Draft Two....................20
BCCQ Draft Three..................20
Patient Participants....................23
Informed Consent........................23
Population Demographics.................23
Participant Recruitment.................23
Sample Demographics.....................23
Materials and Procedure for Draft Three........24
Results of BCCQ Analysis..........24
Development of Information Items........30
Staff Focus Group Results.........32
Patient Focus Group Results.......33
Computerization.........................34
Develop the Program Algorithm.....35
viii


CHAPTER
Program Presentation...............38
Presentation Style.................39
Data Storage and Report Generation.......39
Program Testing..........................39
Phase Two Pilot Test.......................40
Materials and Procedures.......................40
Site Selection...........................40
Population Demographics..................41
Participant Recruitment..................41
4. RESULTS................................................43
How Well Did The Program Function....................43
What Affect did the Program have on Contraceptive
Knowledge?...........................................44
How did the Program affect Contraceptive Choice?....45
Analysis of Method Use.........................46
Respondent Profiles............................47
Presentation Parameters........................47
IX


CHAPTER
How did the User Feel about the Program?............48
Data Management Results.......................48
5. SUMMARY, LIMITATIONS, AND IMPLICATIONS................50
Phase One...........................................50
Summary.......................................50
BCCQ Development........................50
Program Form and Function...............52
Information Items.......................52
The Program as a Patient Education Tool.54
Limitations...................................54
Sampling Method.........................54
Data Collection.........................55
Potential Sources of Program Bias.......55
Cultural Bias.....................55
PP Organization Bias..............55
Developer Bias....................55
Self-presentation Bias............56
x


CHAPTER
Determinants of Program Development......56
Proposed Evaluation Project Study........57
Patient Sample and Sample Size....58
Outcome Measures..................58
Treatment Conditions..............58
Analysis..........................59
Summary.......................................59
Limitations...................................60
Sampling and Design Issues...............60
Testing Effects..........................61
Program Algorithm........................61
Implications and Recommendations..............62
Program Use in a Clinic Setting..........62
Education and Behavior Change............63
6. CONCLUSIONS............................................64
xi


APPENDIX
A. Example Consent Form...................................66
B. Patient Distribution by Age and Site..................67
C. Patient Distribution by Ethnicity and Site............67
D. Patient Distribution by Poverty Level and Site........68
E. Patient Distribution by Parity and Site...............68
F. Patient Distribution by Birth Control Method and Site.69
G. Knowledge Evaluation Questions and Answers............70
BIBLIOGRAPHY.....................................................71
xii


TABLES
Table
3.1 Overview of Methodology: Phase One and Two
3.2 BCCQ Scale Development Across Instrument Drafts
3.3 Birth Control Choice Questionnaire Draft Two:
Measure Ratings by Percentage of Staff
3.4 Results of Factor and Internal Consistency Reliability
Analyses
3.5 Planned Parenthood Staff Focus Group Discussion Topics
3.6 Planned Parenthood Patient Focus Group Discussion Topics
4.1 Pretest-Posttest Performance Comparisons
4.2 Program Attribute Performance Comparisons
4.3 Comparison of Pretest and Posttest Knowledge for
Reproductive Biology and Contraceptive Methods
5.1 Prospective Evaluation Project Treatment Conditions
xiii


CHAPTER 1
INTRODUCTION
In 1991, Lester R. Brown in a "Worldwatch Institute Report on
Progress Toward a Sustainable Society," wrote the following:
During the 20 years since the first Earth Day, in 1970, the
world lost nearly 200 million hectares [2.471 acres] of tree
cover, an area roughly the size of the United States east of
the Mississippi River. Deserts expanded by some 120 million
hectares, claiming more land than is currently planted to
crops in China. Thousands of plant and animal species with
which we shared the planet in 1970 no longer exist. Over
two decades, some 1.6 billion people were added to the
world's population-more than inhabited the planet in 1900 .
. This planetary degradation proceeded despite the
environmental protection efforts of national governments
over the past 20 years.
While its effects may seem distant in both time and place,
planetary degradation poses the most significant, foreseeable
threat to contemporary and future generations of all species
(Linden, 1992). Yet, the rate of unintended pregnancy in the U.S.
is among the highest of all developed nations (IOM, 1995).
Because developed nations are consumer nations, unintended
pregnancies represent an excess burden on planetary resources and
further accelerate the rate of global degradation.
Including both mistimed and unwanted pregnancies, the
lifetime risk of an unintended pregnancy is significant (Ross,
1989). In the U.S., the unintended pregnancy rate exceeds the
intended rate (IOM, 1995), and is among the highest of all
developed nations (IOM, 1995). However, such trends are enigmatic
given the availability of contraceptive methods that all but
eliminate the risk of accidental pregnancy (e.g., the one year
expected pregnancy rate for the Norplant implant is 0.09%,
Contraceptive Technology, 1994).
Contraceptive methods vary in several important ways that
can affect contraceptive adherence and discontinuation (e.g.,
side-effects, efficacy, convenience, and compatibility with the
user), thereby making method choice crucial. Indeed, even with
consistent use, a woman's long-term risk of having an unintended
pregnancy may depend most crucially upon her choice of method
(Ross, 1989).
Of all unintended pregnancies, about half occur among those
usng contraception (contraceptors). Contraceptors differ,
however, in their choice of and adherence to methods. In fact,
method selection is significantly associated with the likelihood
of adherence (IOM, 1995), suggesting that the rate of unintended
pregnancy might be reduced by improving the initial contraceptive
choice.
Unfortunately, as with other decision-making,
1


misunderstanding and lack of accurate information too often
influence contraceptive choice (Peipert & Guttman, 1993). Coupled
with the flaws inherent in human decision-making (Redelmeier et
al, 1993; Simon, 1959), the chance of a "best contraceptive
choice" is readily compromised and the odds of an unintended
pregnancy are increased.
One strategy for reducing the rate of unintended pregnancy
among contraceptors (as well as noncontraceptors) is to use
technology as an assistant. To that end, an interactive decision-
making and educational tool (computer program) might be developed
to assist with the decision process in hopes that better decisions
will result in better method choices (Beckman & Harvey, 1996;
Cramer, 1996).
Purpose of the Project
The objective of this project was to develop an interactive
computer decision tool to assist users with their choice of
contraceptive method. As a decision-aide, the program had the
goal of dispelling misinformation and myth, improving pregnancy
risk estimation for the various kinds of contraception, and
facilitating comparisons among methods. As a secondary objective,
a preliminary pilot test was also conducted.
2


CHAPTER 2
LITERATURE REVIEW
The Problem of Unintended Pregnancy
Probably the most widely used definition of intention today
derives from the National Survey of Family Growth (NSFG). Based
on a data collection effort spanning 40 years, the NSFG has
articulated several different definitions designed to capture the
nature of intention at conception. According to the IOM (1995),
the NSFG categories for pregnancy are as follows:
1. intended at conception: wanted at the time, or sooner,
irrespective of whether or not contraception was being used;
2. unintended at conception: if a pregnancy had not been
wanted at the time conception occurred, irrespective of
whether or not contraception was being used. Within the
unintended conception category, two subcategories emerge:
a. mistimed conceptions: those that were wanted by
the woman at some time but which occurred sooner
than wanted;
b. unwanted conceptions: those that occurred when the
women did not want to have any or anymore pregnancies
ever.
National-level statistics show that in 1987 alone, 3.1
million pregnancies were unintended. In fact, of the 21 million
reversible contraception users in 1987, about 1.5 million
contributed an unintended pregnancy (slightly less than half of
all unintended pregnancies; IOM, 1995). Unfortunately, over half
(53%) of the 3.04 million unintended pregnancies in 1994 happened
to contracepting women, a rate slightly higher than in 1987
(Henshaw, 1998; Mosher, 1998).
As reported by the IOM (1995), the consequences of
unintended pregnancy for the mother and society are manifold:
(1) approximately one-half of all unintended pregnancies end
in abortion. Abortion poses many potential medical and
psycho- social risks for the mother;
(2) because unintended pregnancy is more common among teens,
women older than 40, and unmarried women, these groups
suffer more of the medical and social consequences of
unintended pregnancy. For example, pregnant women above 40
are six times as likely as the general population to
experience maternal morbidity;
(3) sixty-seven percent of births occurred outside of
marriage in 1991 (up from 30% in 1970). Because many of
3


these are to single mothers, often teenaged, the fiscal and
social impacts to society are numerous. In fact, 54% of
mothers less than 16 and 44% of mothers aged 16-17 receive
AID to Families with Dependent Children within the first
five years of birth;
(4) prenatal health care is less likely in the case of
unintended pregnancy. Compared to women with intended
pregnancies, women with unwanted conceptions are 1.8 to 2.9
times more likely not to receive care until after the first
trimester.
To understand the disproportionate rate of unintended
pregnancy in the U.S., it is necessary to understand first that
about 75% of a woman's 30 plus reproductive years are spent trying
to avoid pregnancy (Forrest, 1993). Second, it is necessary to
understand that in the U.S., values regarding sex and sexuality
are often contradictory (IOM, 1995). For example, while youth and
sexuality are regularly emulated (consider television
advertising), premarital sex is often viewed with reprobation
(witness "Family Value" politics).
Coupled with the lifetime risk of pregnancy, such
contradictions are fodder for unintended pregnancy. As Haffner
(1994, in Best Intentions, 1995, 170) writes,
It has been suggested that one consequence of the residual
squeamishness in the United States about sex is that too
many individuals begin their sexual careers with a high
level of discomfort regarding sexual feelings and behavior,
and that this discomfort in turn impedes planning for both
sexual intercourse and for using contraception.
In effect, societal ambivalence about, and personal
discomfort with sex and sexuality create a social milieu in which
other factors, such as method and user characteristics become more
salient.
Determinants of Unintended Pregnancy
Method Characteristics
Underscoring the effort required for effective birth control
use, a series of important method distinctions can be drawn (IOM,
1995). Among the several methods of birth control available
today, distinctions can be made first between those that are
irreversible (e.g., sterilization) and those that are reversible
(e.g., methods other than sterilization). Among contraceptors,
nearly all accidental pregnancies can be attributed to the
reversible methods (IOM, 1995).
For the reversible methods, further distinctions can be
drawn between those that are user-independent (i.e., require only
that the user make occasional visits to her primary care provider:
e.g., Norplant), and user-dependent (i.e., require that the user
be responsible for "taking" the method as "prescribed": e.g.,
4


inserting the diaphragm before intercourse). Because they
substantially reduce the likelihood of user-failure, user-
independent methods tend to perform at levels much closer to their
true efficacy rates.
Among user-dependent methods, a still greater distinction
can be made between methods that are coitus-independent (i.e.,
require only that the user take some time prior to having sex:
e.g., the pill) and coitus-dependent (i.e., require application at
the time of having sex: e.g., condoms). Although coitus-
independent methods help preserve sexual spontaneity, they can be
forgotten. The reverse is true for the coitus-dependent methods:
spontaneity is often compromised but forgetfulness may be less of
an issue.
Ultimately, such distinctions among methods are important
because they relate to the user's ability to use contraception
well. For example, a coitus-independent method, the pill, may be
ill advised for the woman who has difficulty establishing a
routine.
Methods are often compared based on their perfect or typical
use rate. Typical use rates are typically calculated as the net
effect of actual method failure and user failure per couple year
on the rate of conception, while perfect use rates are based on
failure rates among those whose use is consistent and always
correct (Trussell et al, 1990).
This distinction has implications for user-dependent vs.
user-independent methods. For example, while the perfect use rate
for the user-dependent male condom is 98%, the one year
contraceptive typical use rate is 88%: a net effectiveness
reduction of 9% (Contraceptive Technology, 1994). By comparison,
the perfect use rate for the user-independent Norplant implant is
identical to its typical use rate (Contraceptive Technology,
1994). Such differences become especially significant with
continued use.
With perfect use, two out of 100 women using condoms will
get pregnant, compared to about two out of 2,500 who use Norplant
implants (Contraceptive Technology, 1994). While a 2% unintended
pregnancy rate may seem low, the difference between a rate of 2.0
and 0.0009 over five years of consecutive use translates into a
15% chance of pregnancy for the perfect condom user vs. a 0.4%
chance of pregnancy for the "perfect" Norplant user
(extrapolations based on formula by Ross, 1989). Note however
that a better estimate of Norplant five-year failure rate is about
2% based on the moderating effects of factors such as body weight
(Contraceptive Technology, 1994).
Unfortunately, perfect use cannot be expected among even the
most conscientious user. Thus, for the woman who depends upon
user-dependent methods, the risk of unintended pregnancy over time
can be significant (e.g., five year projections for pregnancy
based on typical use rates are 47% for the male condom and about
2% for Norplant).
Given the variability in typical use rates among the
different methods, it seems paradoxical that women seeking to
avoid pregnancy would use any but the most effective methods.
However, as Luker (1975) has noted, the decision to contracept
depends upon the perceived consequences of becoming pregnant vs.
those of contracepting. Stemming from issues such as method
5


convenience and side-effects, the decision is not always clear-cut
(IOM, 1995; Balassone, 1989).
Factors Affecting Method Choice and Adherence
Among the factors affecting method choice and adherence are
those of "convenience": Can others see or feel a Norplant
implant? How much will a Depo-Provera shot hurt? Will the pill
lead to weight gain or acne? Is a pelvic exam required to get a
diaphragm? How much will an IUD cost? While these are important
questions, perhaps the foremost issue is side-effects (Jones,
1995; Savonius et al, 1995; Balassone, 1989).
Numerous researchers have noted that both health concerns
about contraception in general, and specific methods in particular
can deter use (Cramer 1996; Jones, 1995; Zabin et al, 1991). For
example, in a study of 345 adolescent girls, Moore et al (1996)
found that health beliefs about the pill were predictive of
"intentions to use," and "consistency of use."
Different women have different health concerns. Some are
concerned with the more visible and predictable side-effects such
as acne, hair loss/growth, and weight gain. Cramer (1996), for
example, found that over 25% of the women who had stopped using
the pill did so because of "cosmetic" side-effects such as weight
gain. Others are concerned with the more rare but serious events
such as cardiovascular complications and breast cancer (Cramer,
1996). Unfortunately, all such concerns can have implications
for contraceptive choice and for contraceptive use (Luker, 1975;
Lindemann, 1974).
Unintended pregnancy can be attributed, in part, to
inadequate knowledge about specific methods (Dickens et al, 1975;
Mudd et al, 1978), particularly the distinction between adverse
effects that are minor and transient (usually lasting 3-6 months)
and those that require medical attention (Potter, 1991).
Concerns over contraceptive side-effects are often the
product of cultural folklore (Potter, 1991). Unfortunately,
folklore and misinformation can negatively affect contraceptive
decision-making (Jones, 1995; Peipert & Gutmann, 1993; Potter,
1991; Lowe & Radius, 1987).
In the U.S., concerns over cancer related to pill use are
widely held despite the fact that today's pill both poses
significantly reduced health risks for most women (Pratt &
Bachrach, 1987), and has salutary effects, which include
protection against certain kinds of cancer (e.g., ovarian cancer;
Peipert & Gutmann, 1993). Indeed, a 1993 Gallup poll showed that
over 50% had concerns about the pill and cancer, and Peipert and
Guttman (1993) found that nearly 50% of college women had health
concerns related to pill use and over 80% were unaware of the
health benefits of the pill.
Balassone (1989) has reported that those concerned with
health problems are more likely to discontinue pill use within the
first three months. Related studies have found that women who stop
using birth control tend to do so within the first year of use
(Potter, 1991), with pregnancy a likely outcome (Dickens et al,
1975; Mudd et al, 1978). Potter (1991) has suggested that the
trend to discontinue contraception within the first year is
6


attributable to the presence of side-effects (before physiological
adjustments have occurred) and the absence of firmly established
'use regimens" or habits.
Beyond information issues related to contraceptive methods,
researchers have found that many teens and even adults lack
information about reproductive biology (Tanfer, 1994; Lowe &
Radius, 1987; Cvetkovitch & Grote, 1983; Forfeit & Forfeit, 1981),
and method types and use (IOM, 1995; Savonius, et al, 1995; Quint
et al, 1994; Grossman & Grossman, 1994). For example, in a study
of 200 women seeking an abortion, Savonius et al (1995) reported
that irregular use and breaks in contraception were common, and
that despite claims of sufficient contraceptive knowledge, "proper
use knowledge" was frequently lacking.
To summarize, several explanations have been offered for the
high rate of unintended pregnancy in the U.S. and suggest that,
for many, the decision to use a specific method of birth control
occurs amid societal ambivalence about sex-related matters, and
feelings of personal discomfort. At the same time, the range and
efficacy of medical methods available today have improved
substantially beginning with the introduction of the pill in 1960
(Contraceptive Technology, 1994). Yet, many women view their
options as a bittersweet choice between method effectiveness on
the one hand, and method side-effects on the other. In addition,
personal knowledge about the true nature of side-effects and their
health implications is often outdated, incorrect or misleading,
and is seldom balanced against the real and greater risks of
pregnancy and birthing (Jones, 1995). Finally, the salubrious
byproducts of birth control are little known or appreciated among
many women (Jones, 1995).
Given all of the above, the crucial choice of contraceptive
method is likely to suffer. Unfortunately, still other factors
seem to confound contraceptive decision-making.
User Characteristics
In addition to the above, method selection and use are often
influenced by a user's characteristics; e.g., marital status,
economic status, race/ethnicity, religiosity, age, guilt and
embarrassment, method knowledge, risk perception, lifestyle,
pregnancy motivations, and beliefs about the consequences of
becoming pregnant, among others. Among the demographic factors
more predictive of unintended pregnancy is marital status. While
the rate of unintended pregnancy among married women hovers around
40%, it is nearly twice as high among unmarried women (73% in
1990; IOM, 1995), and more than twice as high among unmarried
teens (86% in 1988; Alan Guttmacher Institute, 1994).
Poverty level is sinuously related to the risk of unintended
pregnancy. In 1988, the rate of unintended pregnancy was 75%
among women at poverty level, 64% among women 100-200% above
poverty level and 45% among women above 200% poverty. When
stratified by age however, the picture is more complicated. While
the risk of pregnancy is greater among teens (13-19) of low income
families than other teens, this relationship is reversed among
women in their 20s (about 65% of low income women are at risk
compared with about 75% of higher income women). By age 30
7


however, any differences among income groups disappear(IOM, 1995).
Although higher rates of unintended pregnancy among young
blacks compared to young whites are oft-noted (Kantner & Zelnik,
1972; IOM, 1995), differences tend to disappear with increasing
age (Tanfer & Horn, 1985; Zelnik, Kantner & Ford, 1981) Data
based on the 1988 NSFG suggest that non-Hispanic blacks (52%) are
at only slightly greater risk for unintended pregnancy than non-
Hispanic whites (49?).
Based on a survey of the sexual behaviors of 1,173 college
students, Baldwin et al (1992, 189) reported that, compared with
other ethnic groups, Asians were slightly more sexually
conservative and that ethnicity had little effect on contraceptive
behavior once background variables such as socioeconomic status
were controlled. All told, race/ethnicity appears to be a
moderate predictor of sexual behavior and unintended pregnancy
(IOM, 1995).
Given the above, it may be somewhat surprising that the IOM
(1995) reports that the most recent findings from the NSFG (1990)
show a rather large disparity in the rate of unintended pregnancy
among black women (62%) vs. white women (41%). However, since
contraceptive compliance is influenced by the type of method used,
varied method preferences among demographically and
socioeconomically distinct groups (Turner & Darity, 1980) might be
expected to affect the rate of unintended pregnancy among ethnic
groups.
Compared to other demographic factors, religion appears to
have less influence on contraceptive behavior and the likelihood
of unintended pregnancy (Baldwin, 1992; Grady et al, 1989; Baldwin
& Baldwin, 1988). In a multivariate analysis based on data
obtained from married women, Grady et al (1983) found that
contraceptive method, race, age, income, education, and parity
were all related to "risk of switching" among contraceptive
methods, but that religion was not. Reviewing data from a national
survey (Alan Guttmacher Institute, 1994-1995), Henshaw and Kost
(1996) found that among abortion patients, contraceptive use
varied only slightly among different religious groups.
Protestants (60%) were only marginally more likely to be using
contraception than Catholics (56%). Where differences are extant
however, D'Antonio (1994) suggests that they are often
attributable to the degree of "religiosity" (i.e., the extent to
which individuals give value to religion and adhere to the
proscriptions of their religion).
In general, contraceptive adherence improves with age
(Kantner & Zelnick, 1972; Beck & Davies, 1987, Baldwin, Whitley &
Baldwin, 1992). However, Beck and Davies (1987) caution that
"...age effects are moderated by other factors and deserve a more
detailed examination in order to be fully understood (p. 350)."
Several factors moderate the rate of unintended pregnancy
among different age groups. Among them are knowledge about and
attitudes toward sex and contraception as evidenced by (a) the
relatively high rates of unintended pregnancy among teens and
women over 40, and (b) the lower rates of unintended pregnancy and
abortion among married women (IOM, 1995).
For example, sterilization is often a preferred method of
birth control among those 40 and older. However, the condom, a
significantly less effective method, is the next-most preferred
8


method among older women (Thorneycroft, 1993), a preference that
may derive as much from exaggerated concerns about the health
risks of hormonal methods for older women as from beliefs in the
prophylactic benefits of the condom.
Such concerns exemplify how folklore can affect the rate of
unintended pregnancy. For example, citing Kaeser (1989), the IOM
(1995) reports that in the past, women 40 years and older were
considered to be at greater risk of adverse side effects from oral
contraceptives. Today, shifts to progesterone and lower dose
pills render such concerns obsolete for most women (Contraceptive
Technology, 1994). Nevertheless, among those 40-44, nearly one
fourth use no contraception (IOM, 1995), thus accounting for much
of the unintended pregnancy in this group.
While 18-29 year-old women tend to prefer oral
contraceptives (typical use rate 97%; IOM, 1995), women 17 and
younger seem to prefer the less-effective condom (typical use rate
88%), withdrawal, or no contraception at all (especially at sexual
debut). Thus, withdrawal and condom use best characterize the
"contraceptive" practices of adolescents, especially those in
their early teens, while older women tend to use sterilization or
condoms.
The trend toward more effective contraception use with
increasing age suggests that age is correlated with the
acknowledgment of sexuality (Tanfer & Horn, 1985; Fisher et al,
1983). In fact, the younger a woman is at first intercourse, the
less likely she is to use contraception and the more likely she is
to have an unintended pregnancy.
Common explanations for the age-related trend to use better
contraception (Tanfer & Horn, 1985) include:
(1) cognitive development facilitating (a) an orientation
toward the future (Steinlauf, 1979), (b) relief from "magical
thinking" (Cramer, 1996), and (c) a better grasp of behavioral
contingencies (Steinlauf, 1979);
(2) evolving attitudes toward premarital sex and sexuality
as a function of time and sexual experience (IOM, 1995).
Contraceptive use requires anticipation and planning, which
can offset the trend toward improved contraception with increasing
age (Milan & Kilmann, 1987). That is, because the more effective
methods of birth control require anticipation and preparation,
they may also mean more responsibility for the act, and
consequently, more guilt (Fisher et al, 1983). Thus, some have
suggested that emotional responses to sexual issues may play an
important role in contraceptive behavior.
Byrne and Byrne (Sexual Behavior Sequence, 1977) have
proposed that individual orientations toward sexuality exist on a
continuum ranging from erotophobia (a tendency to avoid sex) to
erotophilia (a tendency to approach sex). According to this
view, "response dispositions" (emotional responses associated with
sexual stimuli), emerge from the socialization process and serve
to guide sexual behavior. Specifically, Byrne & Byrne (1977)
propose that emotional responses become associated with specific
sexual cues during the socialization process and, over time,
determine the extent of erotophilia or erotophobia, a proposition
that seems well-supported (Milan and Kilmann (1987). Ironically,
inhibition may prevent contraception among erotophobes, but not
sexual intercourse (Byrne, 1983).
9


Sexual embarrassment may be also related to contraceptive
behavior (Grady et al, 1993; Keller & Sack, 1982). For example,
in one recent study of adult males aged 20 39 (Grady et al,
1993), nearly 30% felt that it was embarrassing to purchase
condoms and 20% found discarding them embarrassing.
Although the literature is somewhat equivocal, it is often
suggested that perceptions of infertility and poor assessment of
pregnancy risk can negatively effect contraceptive use. For
example, Rainey et al (1993) have suggested that those who believe
themselves less fertile are less inclined to use contraception.
For many women, estimates of personal risk of conception may
be established early in their sexual careers. In fact, a
significant proportion of women fail to use contraception at their
sexual debut with average delays between first sex and first birth
control use ranging from 6 to 12 months (Zelnik & Kantner, 1977;
Smith, 1984; Chilman, 1979).
Delays in contraceptive initiation can set the stage for the
oft-held perception of infertility (Rainey et al, 1993; Luker,
1975) or for the perception of invulnerability (Burger & Burns,
1988) .
Unfortunately, those who underestimate the risks of conception are
less likely to use more effective contraception, to use effective
contraception effectively (Burger & Burns, 1988; Tsui et al,
1991), or to have accurate knowledge of biology and contraception
(Whitley & Schofield, 1986).
Consistent with the above, Whitley and Hern (1991) have
found that women who use less effective birth control, tend to
overestimate the effectiveness of their method. However, they
have also found that risk estimates for the self and friends are
typically lower than for others, and that perceptions of
invulnerability to pregnancy are not negatively correlated with
effective contraception use. As Whitley and Hern (1991) suggest,
"people's risk estimates may be better for events, such as
pregnancy risk, over which they can take control (p. 108)."
Information Processing and Decision-making
As the above suggests, flaws in the decision process or risk
estimation can contribute to the chance of unintended pregnancy
among contraceptors. Summarizing the nature of human decision-
making, Hogarth (1987) has suggested that the limitations inherent
in information-processing have four consequences: (1) selective
perception, (2) sequential processing, (3) dependence on
heuristics, and (4) limited memory capacity. Importantly, with
each "consequence" comes bias.
Selective perception means exclusion of sometimes relevant
information and the misinterpretation of existing information.
Sequential processing introduces order effects, with information
introduced early (primacy) and late (recency) in a sequence having
higher "recall" salience than material that falls in-between.
Heuristics serve to recast complex and ambiguous decision tasks as
simpler ones, passively ignoring or actively dismissing
information along the way. Memory limits result in the
reconstruction of whole memories from the information bits that
have been successfully stored, thereby introducing the opportunity
10


to fill-in-the-blank with that which is personally "logical" fill.
Human information-processing proceeds according to
limitations inherent in the process (Hogarth, 1987; Simon, 1959).
At the same time, each decision task introduces a unique set of
cues that serves to further delimit or direct the decision
process. Thus, how a task is perceived or presented can have
significant impacts on the decision finally reached.
Decision task variables
Some of the more common task variables that can introduce
bias into the decision process include the following (Hogarth,
1987):
(1) the order in which information is presented (discussed
above);
(2) the extent to which information is presented as complete
and ordered (directs the observer away from omissions);
(3) the amount of information (with too much information the
observer cannot extract the most significant factors);
(4) the extent to which information is redundant (directs
the observer to the perception that only one conclusion is
possible);
(5) the simultaneous presentation of concrete quantitative
data (e.g., figures), along with qualitative information
(the observer tends to exclude one or the other data
source);
(6) the presentation of information in its negative (e.g.,
'not', 'no'; obscures meaning and increases the information-
processing demands);
(7) the presentation of outcome data in terms of gains or
losses, (establishes a framework for decision-making that
can alter the outcome);
(8) the presentation of data in succession or "all at once"
(more 'rational' choices follow from "all at once"
presentations).
Other sources of potential bias include the decision-makers
themselves. Decision-makers suffer a tendency (a) to ignore base-
rate data by focusing instead on "case" data, (b) to seek
information consistent with personal hypotheses by ignoring
information that is inconsistent, and (c) to rely on causal
frameworks or schemas by directing and biasing data
interpretation. As Hogarth (1987) suggests, the list of items
leading to judgmental bias seems interminable.
With respect to contraceptive decision-making, other factors
confound the decision process and can be added to Hogarth's
compendium:
(1) the limited number of available methods;
(2) the need to anticipate the consequences of a
contraceptive choice in terms of side-effects, lifestyle
compatibility, childbearing goals (e.g., timing) and method
efficacy;
(3) the number of lifetime "fertile" years;
(4) partner influences (a situational variable related to
11


contraceptive vigilance, Jaccard, 1996);
(5) the limited time contraceptive counselors have to
adequately cover and reinforce key issues (IOM, 1995).
The question then is "How can birth control decision-making be
improved?". One answer might be, education.
Family Planning Assistance
For many, information about sexual reproduction and birth
control is gleaned from numerous sources ranging from friends and
parents to school and the media (Peipert & Gutmann, 1993). In
fact, one study reported that only 3% of women identified their
physician as a source of information about the pill compared with
51% for the media (IOM, 1995). Relatively recently however, a new
medium, the computer, has become available as an educational tool
(Skinner & Allen, 1983).
Few would argue that the computer has become the predominant
technology in the contemporary U.S., affecting nearly all realms
of society, including health education. Relative to more
traditional educational methods however, computer-based
instruction has several obvious advantages such as multimedia
presentation and automatic data collection.
Additionally, evidence shows that computers can enhance
recall and recognition compared with text-only presentations
(Goldstein et al, 1994), improve knowledge (Healton & Messeri,
1993; Lapham et al, 1991), and elicit more truthful responses to
sensitive issues (Aral & Peterman, 1996; Lapham et al, 1991; Levin
et al, 1989) Further, patients seem to like using computers as
educational tools irrespective of age, education and socioeconomic
status (Kahn, 1993). Compared with other education modalities
such as pamphlets or videos, computer-based education has other
advantages. Patient education materials are often written at a
level exceeding the patient's reading level (Parker et al, 1996),
a situation that can be improved via the "talking computer". In
addition, computers can be programmed to respond to patients as
individuals by presenting information contingently based on
patient responses.
Beyond their value as educators, computers can enhance the
decision-making process by empowering patients. Under the more
encompassing rubric of "shared decision-making" or "demand
management", patient empowerment includes the provision of
relevant health information in an unbiased and objective fashion.
In practice, users are provided the support tools to obtain the
information that enables them to become more involved in the
medical care process (Vickery & Lynch, 1995, Anderson et al,
1991). Computers are among the tools often used. Computers offer
still other advantages.
Presented with a complex task, the decision-maker is likely
to rely on heuristic devices to simplify the process. However,
unlike its human counterpart, the computer as information
processor has nearly unbounded capability, and is subject neither
to errors of memory, logic, or probability estimation, nor to the
tendency to shirk "mental effort" when considering choice
alternatives. Given the many factors to be considered when
12


choosing a contraceptive, the computer would seem a promising
ally.
A computer program designed to assist women with
contraceptive decision-making should perform at least three
functions: (1) determine the extent and accuracy of the client's
birth control knowledge base and identify method preferences, (2)
correct misinformation relevant to reproductive biology, pregnancy
risk estimation and the various contraceptive methods, and (3)
provide additional information about the range and relative
efficacy of the various methods available. Consistent with the
behavior change literature (Fishbein et al, 1996), information
should be provided with respect to individual needs or risks
(e.g., medical, lifestyle, etc.), and consistent with the
cognitive psychology literature (Redelmeier et al, 1993),
information should be provided with attention to item framing
(i.e., items should be presented in such a way as to maximize
processes related to client logic and risk perception).
Here it is appropriate to draw a distinction between
decision aids and patient education programs. As proposed by
O'Connor and summarized by Llewellyn-Thomas (1995), decision aids
perform the following functions:
(1) provide information specific to the individual (as
opposed to less personalized and more generic information);
(2) focus on a choice among treatment alternatives where
alternatives are often value-sensitive and "carry high
stakes" (as opposed to choices among alternatives which are
characteristically more clear-cut for physicians);
(3) include exercises dependent upon probability estimation
and aimed at value clarification and treatment goals (as
opposed to choices among alternatives where dependence on
probability estimation is less an issue);
(4) provide information aimed at improving the choice of
treatment (as opposed to improving patient participation in
a predetermined treatment plan).
By its very nature, the process of choosing a birth control
method is highly personal, has "high stakes", is often value-
sensitive due to secular and religious proscriptions (Grady et al,
1989), and is contingent upon probability estimations with respect
to the risk of conceiving and the relative utility of various
contraceptive methods. At the same time, because it is highly
personal, the contraceptive choice itself is often left to the
patient. Consistent with O'Connor's model then, contraceptive
decision-making may benefit from the development of a computer
program that serves as a decision aide.
Project Aim
Because program development was expected to continue as a
post-doctoral project, the study aim was divided into two types,
immediate and long-term. Within this dichotomy, the goal was to
satisfy the immediate aim and to be guided by the long-term aim.
13


Further, the immediate aim of the project was divided into a
primary and secondary aim.
The primary aim was the development of a user-friendly
decision aide software program to assist users with their choice
of contraceptive method.
The secondary aim was to pilot test the program for its
effects on the following:
(1) improved knowledge of reproduction and contraceptive
methods;
(2) reported shifts to more efficacious contraceptive
methods;
(3) the quality of the user experience (i.e., was the
program
perceived as friendly, credible, etc.).
The long-term aim was to use technology as a tool for
enhancing the delivery of unbiased, consistent and personalized
contraceptive choice information. Thus, while untested, program
development proceeded with the long-term aims continuously in
mind.
14


CHAPTER 3
METHODS
The primary aim of this project was the development of a
computer program to assist women with their choice of birth
control. "Phase one" (PI) refers to all methodological issues
related to the primary aim including:
Step 1: Development and administration of the Birth Control
Choice Questionnaire (BCCQ); a self-administered
questionnaire that served as the source of questions for the
program;
Step 2: Development of information items designed to educate
patients about family planning (i.e., reproductive biology
and contraceptive methods);
Step 3: Computerization of question and information items;
defined as the integration of questions and information
items, including multimedia (color, audio, motion, pictures
of contraceptive methods and video clips) into the computer
program.
The secondary aims focused on program pilot testing relative
to its effects on (a) knowledge of reproduction and contraceptive
methods, (b) expressed choice of contraceptive method, and (c) the
user experience (i.e., was the program perceived as friendly,
credible, etc.). "Phase two" (P2) refers to all methodological
issues related to the secondary aims including:
(1) the development of pre- and post-program assessment
tools;
(2) computer administration of the assessment tools and
computer program;
(3) analysis of pre-post changes in outcome measures.
So that the reader might readily follow the methodology that
was used to develop and test the program, PI (development) and P2
(pilot testing) are presented separately with associated
subsections(Table 3.1 below). Further, within PI and P2, the
methodology is presented mostly in chronological order.
Two other items should be noted: (1) with the expectation of
future program versions, the decision was made to direct the
preponderance of time, effort and monetary resources to PI, the
development phase; and (2)the term "program" will be used
throughout the paper to refer to the interactive computer program.
15


NOTE TO USERS
Page(s) not included in the original manuscript
are unavailable from the author or university. The
manuscript was microfilmed as received.
16
This reproduction is the best copy available.
UMI


Phase One
Materials and Procedure
The phase one (PI) objective was the development of an
interactive program to help women with their choice of birth
control- The long-term objective was to use the interactive
program as a tool for enhancing the delivery of unbiased,
consistent and personalized contraceptive choice information.
Consistent with these objectives, the program was designed to
assure the following:
(1) the accurate, unbiased and objective presentation of
information,
(2) an interactive, informative, personalized, and pleasant
format,
(3) the inclusion of theory based and clinical (nontheory
based) questions relevant to contraceptive decision-making
and adherence (explained further below)-
To help assure that the final product reflected these
characteristics, review of the scientific literature and feedback
from family planning professionals and patients became central
activities. Thus, PI began with the recruitment of two family
planning clinics as a source of project participants and family
planning professionals.
Site Selection. Feedback from several family planning
professionals indicated that for contraceptive policy and
procedure, Planned Parenthood (PP) would serve well as a
representative of family planning clinics. Rocky Mountain Planned
Parenthood agreed to support the project and two area metropolitan
PP clinics (clinic A and clinic B) were successfully recruited.
Project development began with a visit to the two project
sites to understand the PP system in terms of patient flow and
client experience. This information provided a sense of the types
of questions appropriate for a family planning center.
The investigator oriented each clinic manager and her staff
to the project prior to PI startup. With each new data collection
phase, clinic managers received an advance written explanation of
project expectations and a follow up telephone call to trouble-
shoot anticipated problems. Each participating PP facility
received a sum of $300 to help compensate for project demands on
staff time.
P.eye.lOPmen.t of the Birth Control Choice Questionnaire. To
determine a set of program questions, three BCCQ drafts were
developed. BCCQ development was based on review of the scientific
literature, feedback from PP staff focus groups, and one-on-one
discussions with dissertation faculty and family planning
17


professionals.
Draft one consisted of 123 questions covering the following
domains: (a) knowledge, attitudes and beliefs about birth control
methods and reproductive biology, (b) history of contraceptive
use, physical health, pregnancy and sexually transmitted
infections, (c) contraceptive preference, and (d) patient
characteristics. After review by family planning professionals
and dissertation faculty, a second draft consisting of 105
questions emerged. Following critical item by item review of
draft two, a third draft consisting of 98 questions was generated.
Draft three was completed by 182 PP patients, and submitted to
Factor and Internal Reliability Consistency analyses. Questions
surviving the analysis and subsequent review process formed the
basis of the computer program.
Although appropriately varied with each BCCQ draft, several
elements of BCCQ development were shared:
(1) BCCQ development, especially question development,
depended, in part, upon feedback from four PP staff focus groups
(two per facility) consisting of an average of 5 staff members per
group or twenty total members that included a mix of behavioral
and clinical staff, and the clinic manager. Staff focus groups
were conducted during routine staff meetings conducted over a four
month period spanning the spring and summer;
(2) because "computer-administered questions" were the final
goal of development, the preferred modality for developing and
testing questions would have been the program itself (i.e., a
computer-administered questionnaire). However, the effort
involved in computerizing questions made it necessary to develop
items on paper before computerizing them;
(3) BCCQ items (questions and information) were designed to
be (a) brief and divested of technical jargon, (b) affectively
loaded for attitude items (e.g., response options ranging from
"very unhappy" to "very happy" were offered for the question "How
would you feel if you got pregnant now"), and (c) free of
universals such as "all" and "none" (DeHart & Birkimer, 1997).
Throughout development, efforts were made to assure that key focus
group issues appeared as BCCQ items, and that items were "user-
friendly."
BCCQ Draft One. Based in part on the literature review,
development of BCCQ draft one began with the following "domains":
(a) knowledge, attitudes and beliefs about birth control methods
and reproductive biology, (b)history of contraceptive use,
physical health, pregnancy, and sexually transmitted infections,
(c) contraceptive preference, and (d) patient characteristics.
Questions emerging from within these domains spanned the following
topic areas:
* (a) ambivalence (about having a baby)
* (b) baby anchor or desirability
(c) contraceptive history
(d) demographics
Me) lifestyle
(f) medical history and contraindications
Mg) perceived efficacy of birth control
18


*(h) motivations and risk perceptions
{i) pregnancy history
(j) sexually transmitted infection (STI) history
(k) sexual history
(l) method preference
Item development derived from both theory and nontheory
based concerns about birth control selection and use- Theory
based scales (e.g., motivations and risk perception; denoted above
by an asterisk) were presumed to be related by an underlying
construct reflecting specific aspects of the scientific
literature. Nontheory based scales (e.g., medical history) were
developed and organized based on face validity (i.e., a casual
review of how good an item or group of items appear, Litwin, 1991)
but were not expected to be related by an underlying theoretical
construct.
Table 3.2
BCCO Scale Development Across Instrument Drafts
Draft one scale Draft two scale Draft three scale FA/ICR
Screening items Dropped
Demographics Demographics Demographics NA
Sex history Sex history plus STI question Sex history/STI question NA
STI history Dropped Sexual activity NA
Pregnancy history Pregnancy history Pregnancy history NA
Contraceptive history Contraceptive history and method preference Contraceptive history and method preference NA
Medical history and contraindication Medical history and contraindication Medical history and contraindication NA
Method preference Moved to contraceptive history scale NA
Baby anchor Baby anchor (subsumed ambivalence) *Baby anchor Desire to delay and Chilabear ing behavior
Ambivalence Moved to baby anchor scale NA
Lifestyle Lifestyle Lifestyle Sexual frequency Sexual promiscui ty
19


Table 3.2 (Cont.)
Perceived efficacy of birth control Perceived efficacy of birth control Perceived efficacy of birth control Perceived efficacy of birth control
Side-effects and method convenience Side-effects NA
Beliefs about consequences of pregnancy Beliefs about consequences of pregnancy Beliefs about consequences of pregnane
Partner impact/support Partner support Partner support
Motivations and risk perceptions Perceived risk Perceived risk Perceived risk
True/false for reproductive biology and contraceptive methods True/false for reproductive biology and contraceptive methods NA
NA not applicable
Submitted to Factor and ICR Analyses
As table 3.2 shows, the original set of draft one topics evolved
with each draft of the BCCQ, yielding a final set of 14 scales.
BCCQ draft 1 contained 123 questions and was reviewed by two
PP staff focus groups and a nonPP family planning professional.
Feedback and revision resulted in a second draft of 105 questions.
BCCQ Draft Two. BCCQ draft two (105 questions) was
evaluated by two PP staff groups (one group per site)"by item" for
burden, clarity, and the appropriateness of the response options
and, following item evaluation, "by scale" for content validity.
To minimize staff burden, each clinic reviewed half of the twelve
total scales.
Draft two items were rated by staff focus groups using the
following criteria:
l=Data item is easily answered
2=Data item is answered with a little bit of thought
3=Data item is difficult or nearly impossible to answer
20


For example, the following question would be rated 1, 2, or 3 for
"burden": "If you had side-effects using birth control, how easy
would it be to tell us about it?"
Very easy
Easy
Neither (not easy or hard)
Hard
Very hard
Item clarity (understandability) was defined as "how clear
the item is relative to the stated rationale of the scale." Thus,
for each of the twelve scales a rationale was provided. For
example, the rationale for the "lifestyle" scale was as follows:
The next few questions are about your day-to-day life. Your
answers to these questions are important since the kind of
birth control that is best for you depends partly on how you
live day-to-day. For example, for some women, it may be
difficult to remember birth control if they are very busy.
Thus, within the lifestyle scale and with the above rationale in
mind, each item was reviewed for its understandability or clarity.
Clinicians were asked to assess items relative to the rationale.
Item clarity consisted of four qualitatively distinct levels as
below:
l=Meaning of item is clear;
2=Meaning of item is slightly unclear;
3=Meaning of item is somewhat unclear;
4=Meaning of item is very unclear.
Response options were judged by their appropriateness for
the question. That is, a simple yes/no dichotomy may be most
appropriate for some questions (e.g., "Have you ever had a breast
lump?") while a polychotomous response set ("not at all", "a
little", "somewhat", "very much") may be more appropriate for
other questions (e.g., "I would like to have a baby in the
future") .
Response options were evaluated as follows:
l=Appropriate;
2=Less than appropriate;
3=Inappropriate.
Using the above definitions, each staff reviewer posted her
item ratings to a score sheet. To analyze the results of the item
ratings, frequency data were tabulated for each measure.
To evaluate overall item performance (i.e., across the three
measures for a given item), rating labels were relabeled
"maximum", "intermediate" and "minimum." For example, burden was
recoded as follows:
(1) Item is easily answered=Maximum rating;
(2) Item is answered with a little bit of thought=Intermediate
21


rating;
(3) Item is difficult or nearly impossible to answer=Minimum
rating.
Because the Item Clarity measure consisted of four ratings;
both the second and third or "middle ratings" were recalculated as
"Intermediate." Table 3.3 provides a summary of the staff item
measurement evaluations.
Table 3.3
Birth Control Choice Questionnaire Draft ..Two.;
Measure Ratines bv Percentage of Staff
Maximum rating Intermediate rating Minimum rating
Clarity 47 53 0
Burden 47 53 0
Response options 40 59 1
Note: N=105 total items. Rows do not total to 105 due to missing
responses.
As Table 3.3 shows, for all three measures, slightly less
than half of all items received the "maximum rating" and just over
half of all items receive an "intermediate rating". Only one item
was given the minimum possible rating due to one rater's
evaluation of "appropriateness of response options" as
"inappropriate."
To measure validity, PP staff reviewers also provided
content validity scale ratings for a total of 12 scales.
Accordingly, a written rationale explaining the intent of the
proposed scale was temporarily added to the BCCQ.
Content validity, operationally defined as, "Does the scale
measure what it is intended to measure based on the stated
rationale?" was rated on a four point scale including the
following:
l=The scale appears valid; all necessary items are included
and all of the items contribute, as is, to the scale.
2=The scale appears mostly valid; some necessary items may
be missing or some of the items may not contribute, as is,
to the scale.
3=The scale appears only somewhat valid; a number of
necessary items may be missing or a number of the items may
not contribute, as is, to the scale.
4=The scale does not appear valid; most of the necessary
items may be missing or most of the items may not
contribute, as is, to the scale.
Of the 12 scales rated, 8 (67%) were considered "valid" or
"mostly valid" by all raters. The remaining four scales (23%)
were rated by at least one evaluator as either somewhat or not
22


valid. All scales with suboptimal item (e.g., clarity) or content
validity scale ratings (operationalized as less than the maximum
rating by all raters on any attribute) were carefully reviewed and
appropriately modified or deleted.
BCCQ Draft Three. Modifications to BCCQ draft two resulted
in BCCQ draft three (98 items). BCCQ draft three was subsequently
completed by 182 PP patients.
Patient Participants. One hundred eight-two participants
completed the final version of the BCCQ and approximately 20
participated in the patient focus groups (described later). All
participants were sampled from the same patient population
(described below).
Informed Consent. For all phases of patient data
collection, prospective participants were informed that their
cooperation was voluntary, confidential, and that their refusal to
participate would in no way affect the quality of their care. In
all cases, the study objective, participation requirements, and
contents of the consent form were reviewed with the prospective
patients. That the questions were sensitive, dealing with issues
related to sex and contraceptive use was also noted. All
participants were informed that they could refuse to answer any
questions and could stop the study at any time. Those consenting
to participate were asked to read the consent form, sign one copy
and retain another for their records (Appendix A).
Population Demographics. Over calendar year 1996, the two
sites saw a total of 5,072 patients (clinic A, 3022; clinic B,
2050). Comparisons of patient demographics between the two
facilities found no significant differences for age, parity and
contraceptive method used. By Chi Square analysis, a
significantly greater proportion of Clinic A clients were found to
be Hispanic (p<.001), and of lower income (p<.001)than Clinic B.
Neither facility serves a significant proportion of Black
Americans (<2%) or other reported ethnic groups (<4%) Because
the two clinics showed few significant demographic differences,
data were combined and treated as a single sample throughout the
remainder of the project (Appendices B through F).
Participant Recruitment. Eligible participants were all
women presenting to Clinic A and B who were seeking contraception,
over 17, English-speaking, not pregnant, and judged by staff to be
cognitively and physically able to participate. It should be
noted that about 8% of the total patients did not receive
contraception at their initial visit, in most cases because the
reason for visit was other than to obtain birth control (e.g.,
pregnancy testing, follow-up to a previous visit). Such patients
were ineligible for the project.
To minimize the likelihood of "seasonal" or "major event"
effects on contraceptive behavior (e.g., Magic Johnson announcing
that he had HIV and the ensuing increase in requested HIV testing
nationwide; CDC, 1993), data collection occurred concurrently at
the two PP facilities.
Sample Demographics. Using a script provided by the
principal investigator, patients were recruited by clinic staff at
23


the start of the clinic visit and completed the BCCQ immediately
after their visit. Participants received $15 for their time.
A total of 182 patients completed the BCCQ (92 for clinic A
and 90 for clinic B). Based on Chi Square analysis, no significant
differences between the proportion of patients enrolled and the
clinic population were found for race/ethnicity or for methods
used.
Other demographic comparisons could not be made. Age data
were collected using noncomparable categories. Parity (number of
children) and percent of poverty data were not collected using the
BCCQ. Unfortunately, rate and other data for those who refused or
who were ineligible were not obtained due to the burden of patient
recruitment on already busy clinics.
Materials and Procedure for Draft Three
The third and final version of the BCCQ totaled 98 items,
including 74 questions and 24 true/false items, and was used to
collect data from 182 PP patients (92 for clinic A, 90 for clinic
B) over the same period of about eight weeks in fall.
With program length constraints in mind, draft three
analysis emphasized item elimination as opposed to revision.
Although some items were modified, all items retained were moved
directly into the program without further testing.
Because the BCCQ was designed as a self-administered
questionnaire, skip patterns were not used. Instead, appropriate
options were made available within the questions themselves. For
example, a series of items asked "Over the last 6 months, how
often have you had sex WITHOUT birth control because..." and
finished with various options such as "you had problems using your
birth control correctly," or "your partner did not want to use
birth control." For these types of questions, the response
option, "I have not had sex in the last 6 months" eliminated the
need for a skip pattern.
To assure readability, the Regional Quality Assurance
Manager of Rocky Mountain Planned Parenthood critically reviewed
draft three. Revisions were made as necessary.
Results for BCCQ Analysis. Data management and analysis
activities were conducted using two software packages: EPI INFO
6.04b, and SPSSx PC v4.0. EPI INFO was used to create data entry
screens and SPSSx was used for data analysis.
BCCQ frequency data for 182 Planned Parenthood (PP) patients
were used to help identify possible problem items; i.e., items
with missing data or with evidence of floor or ceiling effects.
Stratified analyses were run on a few select variables to gain a
better understanding of the relationship among variables.
Because item variability seemed sufficient, no items were
seriously modified or eliminated based solely on frequency or
stratified data unless supported by other information (e.g.,
clinical input). While draft three nontheory based scales (e.g.,
demographic and history scales) were limited to review and
modification based on frequency data and feedback from family
planning professionals, theory based scales were additionally
evaluated.
24


Although derived from the literature and clinical expertise,
it was decided that the eight theory based scales (Table 3.2
above) should be verified using Factor Analysis. Thus, a Factor
Analysis (FA), using oblique rotation, was conducted. Because
oblique rotation preserves the correlations among scales, it can
be assumed to provide a more realistic representation of the
constructs underlying the scales or factors (Norusis, 1990).
Although a few trichotomous items were included, the
preponderance of items targeted for FA were 5 point Likert-type
items. Factors were selected using Cattell's scree method and
eigen value cutoffs of 1.0 (Norusis, 1990), and items were
included whose scale loadings were above 0.4.
Missing values were of two types: nonresponse values
(respondent did not answer) and skip values (default values
assigned to questions that are irrelevant to a given respondent).
Mean substitution was used for nonresponse values. Nonresponse
values were rare (usually less than 2% of the cases).
Skip values were sparingly used, affected an average of
eight respondents, and were recoded to a conservative valid value
based on the question. For example, the question "Over the last
six months, how often have you had sex WITHOUT birth control
because you could not afford it?" offered six options ranging from
"never" to "always" and including a skip response "I have not had
sex in the last six months." For analysis, the skip response was
recoded to "never."
Using draft three of the BCCQ, eight theory based scales
were submitted to FA:
(a) adherence (e.g., "How often have you had sex WITHOUT
birth control because...?")
(b) desirability of childbearing (e.g., "How would you
feel if you had a baby now?")
(c) barriers or method convenience (e.g., "Would you use a
method if it meant having a shot?")
(d) beliefs about the consequences of pregnancy (e.g., "If
you had a baby now, it would be hard to stay in
school.")
(e) lifestyle (e.g., "How often do you sleep at your own
home?")
(f) partner support (e.g., "How often do you and your sex
partner talk about birth control?")
(g) perceived efficacy of birth control ("For the method
you use now, how likely are you to get pregnant in the
next 12 months?")
(h) perceived risk of pregnancy (e.g., "If you had sex
just once without birth control, how likely would you
be to get pregnant?")
Based on the FA, a total of 14 factors emerged (Table 3.4).
25


NOTE TO USERS
Page(s) not included in the original manuscript
are unavailable from the author or university. The
manuscript was microfilmed as received.
26-28
This reproduction is the best copy available.
UMI


Reliability analyses were run on each scale to determine its
internal consistency (Table 3.4). To avoid violating the
assumptions of the model, items worded or scored negatively
relative to the scale were recoded (e.g., 1=5, 2=4, etc.). (This
is not an issue for FA, because negative items are simply
interpreted negatively.)
Items were considered candidates for elimination based on
review of the "corrected item-to-total correlation" and "alpha if
item deleted" values. Based on the default "casewise deletion"
for ICR analysis, data for 12 respondents were eliminated; leaving
a sample of 170.
In an iterative process, factors with poor internal
consistency coefficients (Cronbach alphas less than .7; (Norusis,
1990) were reviewed using the combination of frequency data,
Factor and ICR analyses, and previous expert advice concerning
item importance. In some cases, scales were reconstituted (items
shifted from one scale to another) and Factor and Reliability
analyses rerun. Because Cronbach's alpha tends to increase with
the total number of items, alphas less than .7 were often
tolerated when the number of scale items was small (two or three
items/scale; Table 3.4).
The combination of Factor and ICR analyses served a twofold
purpose: (1) item reduction and (2) scale verification. Item
reduction was important if the length of the program was to be
limited to 20 minutes. Scale verification served to determine how
the items for each scale contributed to that scale. Both
processes were guided in part by the previously obtained feedback
of focus groups and individual reviewers.
Within each scale, item elimination was judiciously
effected, taking into consideration FA item loadings, reliability
coefficients (Cronbach's alpha), item variability, and the
pragmatic value of the question. As a general rule, items scoring
well on the analyses (i.e., good validity, reliability, and
variability) were retained, while items scoring poorly were
dropped (about 10%). Items retained were sometimes modified.
Other items were retained for their significant pragmatic
value. For example, the item "Would you use a birth control
method if it meant having to touch yourself?" has pragmatic value
for determining the appropriateness of barrier methods and was
included in the program despite the fact that it contributed
little to the "adherence" scale. In the case of borderline
items, item variability was also reviewed using frequency data.
For example, 90% of all respondents answered that they "never" or
"rarely" have sex without birth control because they "forget to
bring it". Still, this item was retained due to the high risk of
unintended pregnancy among the remaining 10%.
As shown in table 3.4, the original set of 8 factors was
expanded to 14. Of the 14, 11 were finally retained for program
use. With time constraints in mind, items within factors were
dropped if they performed less well statistically and/or were
determined to be of lesser practical value for determining
contraceptive choice.
The three scales that were dropped from the program included
the following, "partner support", "beliefs about the consequences
of becoming pregnant", and "perceived risk". While statistically
sound (based on ICR and Factor analyses), the "partner support"
29


scale was determined to be beyond the program's immediate scope.
Although the "beliefs" scale also performed well statistically, it
was felt that more in-depth questioning would be needed to
generate relevant feedback and would be better developed in a
later version. The "perceived risk" scale did not perform well
statistically and was converted into a set of information items.
Although not constructed as a scale, a set of 24 "True/False"
items was converted to factbacks and factoids.
Development of Information Items. Analysis of BCCQ data
yielded a final question set for computerization and set the stage
for the development of the information items. PI program
information development focused on three information types: (a)
factbacks (contingent information), (b) factoids (noncontingent
information), and (c) recommendations.
As distinguished from "feedback," "factbacks" are computer
responses made contingent upon answers to questions. For example,
information concerning the risk of unintended pregnancy is
provided contingent upon the marital status of the woman (e.g., a
"never-married" woman immediately receives the factback that
"Almost 90% of all pregnancies to never-married women are an
accident").
Factoids provide information deemed important for all
respondents and occur noncontingently. For example, when
transitioning between the sexual history and lifestyle questions,
the factoid 8 of 10 women who dont use birth control get
pregnant each year appears on screen.
Based on responses to questions asked, birth control
recommendations and justifications are provided for up to three
methods. Importantly, these "three best methods" are presented
only as options.
Because justification statements are a product of both
individual responses to questions and the particular method
suggested, a variable number of justifiers occurs per method. For
example, if the program selects the pill as a good birth control
option, the following might appear on screen:
The following information is based on your answers to the
earlier questions. The pill might be a good method for you
because...
You want to have a baby in the future. You are a person of
habit. You will tell the clinic if you have side-effects.
Privacy about birth control is not of great concern to you.
You have used the pill in the past and might like to use it
now.
Please talk with your doctor or nurse about any questions or
concerns you might have.
A somewhat different set of justifiers might appear for another
pill user (depending upon her responses) and another set would
appear if the program selects Depo-Provera as an option.
Information items were developed based on feedback from PP
staff, individual consultations with family planning
professionals, and information derived from the scientific
30


literature and PP materials.
To help develop the information items, The "Planned Parenthood
Fact Book", authored by the Regional Quality Assurance manager for
the Rocky Mountain Planned Parenthood, was directly imported into
the program and served as the primary resource for feedback items.
The primary data source for the PP Fact Book is the well-respected
and widely used text, Contraceptive Technology (1994).
Feedback on the overall orientation and content of the
program, including the factbacks, factoids, recommendations, and
program algorithm was provided via three staff focus groups (two
at clinic A, one at clinic B) and four patient focus groups (two
at each clinic) held over a period of three winter months).
Each patient focus group was conducted by the same two
experienced female focus group leaders, both of whom had work
experience in family planning settings. With the exception of the
two staff focus groups dealing with item ratings that relied on
hard copy materials, all focus groups were taped and for purposes
of program development, summarized in written form. The focus
group facilitators were paid $50 each per focus group. Patients
were paid $20-$30 for participation (with the higher amounts being
paid when recruitment was slow). Staff focus groups were
conducted by the investigators 3.5 and 3.6 describe the purpose of
each focus group.
Table 3.5
Planned Parenthood Staff Focus Group Discussion Topics
Group # Content
1 (L) Utility of the program for respondent and clinician
2 (A) Suggestions for scales and questions
3 (L) Review of 1st half of questions 1 (burden/clarity/content validity) |
4 (A) Review of 2nd half of questions (burden/clarity/content validity)
5 (L) Review of factoids and factbacks (validity/accuracy)
6 (A) Review of recommendations and overall approach (including order of items and scales, factbacks, factoids, recommendations)
L=Clinic B , A=Clinic A
31


Table 3.6
Planned Parenthood Patient Focus Group. Discussion Topics
Group # Content
1 (A) Discussion about what women need to know about birth control and about reproductive biology
2 (L) Confirmation of focus group 1
Discussion of what information the computer program miaht nrovide. and how
L=Clinic B, AR=Clinic A
All focus groups were taped. A faulty recording, however,
resulted in the loss of all data from the third and final patient
focus group. All other focus group tapes were reviewed and
summarized in written form by the investigator and used to direct
program development.
Focus groups targeted the program's scope (i.e., what should
be included/excluded), utility (i.e., how can the program benefit
the patient or clinician), and format (e.g., should the program
allow users to browse method information at leisure).
The process of building program information items began by
asking clinicians about their interaction with patients. Perhaps
the most telling question asked was, "What do you want patients to
know?" Responses to this question focused on patient tolerance
for method side-effects and lead to a greater emphasis on
perceived method convenience (i.e., method characteristics that
influence preference and consistent use).
The process was later refined by asking staff to respond to
the results of the BCCQ frequency analysis. Specifically, for
responses signaling a potential risk of unintended pregnancy,
staff were asked what information should be provided (e.g., "What
would you tell a patient who thinks that, "While using her current
method of birth control, she is very likely to be pregnant in the
next 12 months") .
With a draft set of information items in hand, staff focus
groups focused on the accuracy, completeness and utility of
information items.
Staff Focus Group Results. Three key issues emerged from
the clinician focus groups: (1) confidentiality and anonymity;
(2) report generation; and (3) clinic use. With respect to
confidentiality and anonymity, staff considered whether or not
same patient information should be tracked across visits using an
assigned ID# or personal identifiers such as name. Emphasizing
that the value of the program for the patient may be enhanced with
anonymity, most staff recommended that no personal identifiers be
collected.
Concerning clinic use, staff suggested that either (a) the
program be available in the waiting room and used on an ad lib
basis, or (b) the program be completed by all patients as part of
the routine visit. Response was mixed. Arguments for ad lib use
held that the program should be used as an adjunct to care.
32


Arguments for routine use held that the program should replace
routine education functions and free the clinician for more 'in-
depth' patient interaction. Nevertheless, a preference for the ad
lib strategy was evident.
With respect to report generation, staff were split on
whether or not clinicians should receive reports, but nearly all
agreed that patients should receive reports. Because these issues
have direct implications for the utility and future use of the
program, but few implications for program development, they will
be discussed further in the final chapter of the manuscript.
Patient Focus Group Results. Patient focus groups were
conducted from a "fresh" vantage point, asking patients what
information they would like (e.g., "What kind of information would
you like about birth control?"), and how they perceived computer
feedback (e.g., "How would you feel about getting birth control
information from a computer?"). Patient focus groups were not
used to evaluate individual questions or information items.
With respect to the scope and utility of the program,
patient participants were asked some of the same kinds of
questions directed to staff focus groups. Patients were asked how
they felt about using a computer to help them with contraceptive
choice, including how they might respond to computer questions. A
few patients seemed distrustful of, or otherwise disinclined,
toward the use of a computer for assistance with contraceptive
choice. However, the large majority indicated that they liked the
idea, and many said they would be able to be more honest with the
computer. Although longitudinal data would be valuable for
program improvement, the opportunity to capture respondents'
perspectives about "tracking patient information across visits"
was inadvertently missed.
PP patients were also asked: (1) who should receive reports
(patient and/or clinician), and (2) what information should be on
the reports. Few respondents expressed concern about report
sharing with clinicians. Indeed, the majority of the participants
readily shared personal information about sex and birth control
with the male investigator. These "data" suggest that family
planning patients may be comfortable sharing personal information
about sexual behavior. Because demographic and other information
on focus group participants were not obtained however, the extent
to which past experience with contraception or with family
planning clinics may have influenced this perspective is unclear.
When asked about clinic use, patient groups focused on the
"who" and "how". Concerning "who", respondents emphasized the
potential utility of the program for women newly seeking
contraception and for couples. With respect to "how", respondents
felt that the program should be available on an "ad lib" basis and
should provide the option for users to browse information about
various methods.
Based on the analysis of both quantitative and qualitative
data, a final question set was determined and information items
and multimedia features were added to the program. Once
computerized, program revisions to information items and
multimedia presentation were rare. As a last step, content and
algorithm specifications(in the form of print screens and written
documentation) were reviewed by a paid family planning clinician
33


and revisions made.
Computerization. PI computerization (item transfer to the
program) consisted of several steps including (l)the
computerization of questions and information items, and (2) the
computer-administration of (a) the program pretest, (b) the
program, and (c) the program posttest.
Computerization depended upon the following software and
hardware.
Software included the following:
Macromedia Authorware 3.5 for Windows 95, the
principal software used to develop the program and to
track on user responses;
Data Translation Broadway 2.0 for Windows 95, software
used to download video-cam audio-visual clips to the
computer for importing into Authorware;
PhotoStudio 2.0, software used to download digital
pictures of birth control methods;
Windows 95, the operating system.
Hardware included the following:
Compaq IBM-compatible personal computer (PC), 133
megahertz PC with audio and MPEG capability, a 2
gigabyte hard drive and 48 megabytes of RAM;
Ricoh digital camera, used to take birth control
method photographs.
Several program features or approaches were standardized in
the computerization process (e.g., all question items include both
the initial and a confirmatory question). These features affect
both the presentation of questions and information items and
include the following:
(1) During computerization, particular attention was paid to
ensuring a positive patient experience (e.g., was the program
friendly, easy to use, credible, etc.), and to enhancing
information salience via the judicious use of multimedia and the
application of principles derived from the decision-making
literature to item construction (e.g., better decisions tend to be
made when all choices are presented together rather than
separately).
(2) To streamline the program, many BCCQ polychotomous items
were rewritten as dichotomies or trichotomies. For example, the
Likert-type item, "How likely are you to use birth control every
time you have sex?" (not at all likely, somewhat unlikely,
somewhat likely, very likely, extremely likely), was rewritten to
read, "Do you think you will use birth control every time you have
sex?" (yes, no, maybe).
(3) All program questions have two parts, the initial
question and a confirmatory question. The confirmatory question
provides an opportunity for the respondent to correct mistakes or
to reconsider a response. Only confirmed responses are posted as
data to the response data files.
(4) Information was provided in an attention-getting way
with sensitivity to "information overload." Information was
sometimes restricted to a high risk subset of users (e.g., women
indicating a positive orientation toward childbearing receive
information cautioning about the delicate balance of "child
34


desirability" and "contraceptive vigilance").
(5) Because the literature suggests that knowledge of
reproductive biology and conceptive risk is frequently lacking,
all users received the same information concerning reproductive
biology and conception risk and all users were classified per
their contraceptive history. This strategy also saves the program
time that might be otherwise spent assessing the user's knowledge
base (a 20 minute program was the goal).
(6) All users received a relatively brief overview of the
advantages and disadvantages of methods prior to being asked about
method preference. As a time-saving strategy, in-depth method
knowledge is provided only for the "best three method options."
Development of the Program Algorithm. "Algorithm" is used
here to describe the process by which the program determines the
best three method options. In general, the "best three" are
determined by incrementing or decrementing a counter for each
medical method based on the respondent's answers to questions
posed by the computer.
Consider a woman sitting at the computer as it begins to ask
questions. Initially, all method counters are set to zero. With
each question the woman answers, counters are incremented and
decremented. For the IUD for example, a counter called "IUDMAYBE"
is incremented each time the woman's response to an "IUD-related
question" affirms it as an appropriate method. The counter is
decremented when the converse is true. The final value for
IUDMAYBE serves as the numerator. A parallel counter, called
IUDDEN, counts the number of times IUD relevant questions are
asked and serves as a denominator.
The denominator provides an adjustment for differences in
the total number of questions asked for each method. Total number
is variable with the method (ranges from 9 for tubal ligation to
17 for the IUD), and with the user circumstance.
Best options are determined by calculating the following
percentage: CV/Q 100, where CV is the counter value for that
method and Q is the total number of questions asked for that
method. Using the IUD example provided earlier, the formula would
appear as follows: IUDMAYBE/IUDDEN 100.
Continuing the IUD example, A variable called NOTIUD is also
initialized to zero and set to "one" only if it is contraindicated
based on medical or psychosocial factors. If NOTIUD is set to one
at any time in the program, the IUD is excluded as a birth control
option for that patient.
Four elements drive the program algorithm: (1) the primary
questions, (2) the secondary questions (3) "contraceptive user
status", and (4) method experience and preference.
Primary questions are those asked of all respondents or
users, and help assure a "common" patient experience. Via the
primary questions, all respondents experience a significant
portion of items (e.g., questions and feedback) in the exact same
way and mostly in the same order. Because different respondents
provide different responses to the same questions however, the
information received can vary among respondents (e.g., information
about risk of unintended pregnancy for teen women vs. women 40 and
over differs). Derived from the BCCQ, the program has 42
primary items in addition to questions about method use and
35


preference to which all participants must respond.
Secondary items derive from the same topic areas as the
primary screening questions, but are asked contingent upon
individual responses to questions, and serve to "personalize" the
program. The majority of the secondary questions are presented
as a function of contraceptive user status because of its presumed
influence on respondent knowledge, attitudes, and information
needs. Although variable, most respondents encounter about 20
secondary questions.
Contraceptive user status derives from a respondent's
experience with contraception and is based on medical methods only
(i.e., those available exclusively through a medical
professional). The presumption is that those inexperienced with
medical methods have different education needs than those with
experience.
The following contraceptive user typology was used:
(a) "nonusers" patient has never used a medical method;
(b) "past users" patient has used medical methods before
but is not currently;
(c) "current users" patient is using medical methods now.
Note that "now" was intentionally not defined for the user, in
part because it would be a difficult and time-consuming term to
operationalize.
Compared with current or past users, nonusers are presumed
less informed about medical methods. For the nonuser, additional
information concerning methods is provided, much of it focusing on
pragmatic issues (e.g., how much will a Depo-Provera shot hurt?).
In general, users may be at risk of intermittent use if they
have difficulty using a method well (e.g., fitting it into their
lifestyle, side-effects), or are uncomfortable using it. Thus,
current and past users who are interested in continuing use with a
familiar method are presented a series of questions assessing
wellness of use (WOU) (e.g., "How often have you had sex without
your method because you ran out").
Within the program algorithm, method experience, method
preference and WOU evaluation exert additional influence. While
this influence is complex, the general approach is as follows:
(a) If a user indicates a preference for a method, it is
incremented.
(b) Wellness of use interacts with preference to affect
method counters.
(i) If the WOU assessment is favorable, the method is
incremented. Further, if proper use is reported and no
other options are preferred by the user, a message
appears on screen acknowledging "proper use of the
preferred method." No other options are presented and
the program is, in effect, finished.
{ii) If the WOU assessment is unfavorable, up to 3
"best options" are presented ("unfavorable use" was
defined with the help of a clinician). In this case,
the preferred method may or may not be presented among
the "best options," depending on the combined effect
of the usual algorithm and the WOU evaluation on that
method's counter.
(c) If the user indicates no experience with any preferred
methods, the preferred methods are incremented and treated
36


in the algorithm as usual.
(d) If the user indicates current or past experience with a
method but no preference for that method, it is made
ineligible; the assumption being that the method is not of
interest to the user.
With respect to the algorithm and program-respondent
interactions, all methods are treated, including "none" and
"abstinence." Because the program is to assist respondents with
the choice of a medical method; however, only medical methods are
presented as best options.
By default, all users receive information on the male
condom. This approach serves to treat sexually transmitted
infection (STI) issues without having to spend significant program
time determining STI risk.
The following is an example of how the best three methods
are chosen. Assume that the program asks 20 questions pertaining
to Norplant and that 10 responses are "Norplant favorable". A
Norplant score of 50 would be obtained (10/20*100). If 50 were
the highest score obtained for any method, Norplant would appear
as the first method recommended.
Other factors help determine the top three options however.
In the case of tied scores, a hierarchy determines which method is
presented. The hierarchy is based on following:
(1) a continuum of user-dependency (i.e., irreversible to
coitus-dependent methods);
(2) the overall efficacy of the method (i.e., tribal ligation
is more effective than Norplant implants);
(3) the requisite rate of method replacement (tubal ligation
is permanent, Depo-Provera must be replaced every three
months).
While seemingly a complex process, the three criteria tend
to be related and, with the exception of the relative rankings of
Norplant implants and the IUD, agreement among family planning
professionals about relative rankings was easily obtained. The
hierarchy is as follows (descending order): tubal ligation,
Norplant implants, the IUD, the Depo-Provera shot, the pill, the
cap/diaphragm (per PP policy and because of their similarities,
the cap and diaphragm are treated as equivalent).
Based on the hierarchy, if Norplant and the diaphragm both
scored 50, Norplant would appear as option 1 and the diaphragm as
option 2. Further, because only the top three options are
presented, if Norplant and the diaphragm were both tied for third,
the diaphragm would rank fourth and would not appear as an option.
Because it is in an alpha test stage, the user is informed
neither of the hierarchical presentation of options nor the
absolute scores obtained. In other words, the user is unaware if
the top ranked method scores 50 and the next ranked method scores
a distant 10. Only methods with values above zero are offered as
options (zero and negative percentages are ignored).
Depending upon how method valuation (e.g., preference,
perceived convenience) interacts with contraceptive user status
and side-effects, different questions and information items occur.
For example, a user who has experience with a preferred method
37


will experience a different set of questions/information than a
user who has no experience with a preferred method.
Program Presentation. Questions and information items were
constructed to maximize their salience and influence on the
decision-making process. Cognitive factors such as primacy and
recency effects, serial vs. sequential presentation, and framing
effects, as well as style considerations such as program pacing,
credibility and friendliness were all factors in the program
development process. While many of these factors were considered
during PI item development, it was not until computerization that
they could be implemented (e.g., presentation pace).
With "cognitive considerations" in mind, the program
routinely states contraceptive use statistics positively; i.e., in
terms of their effectiveness in preventing pregnancy (e.g., 87%
effective) rather than their ineffectiveness (e.g., 13% chance of
getting pregnant). This approach is based on the risky shift
phenomenon in which decision makers tend to make choices that are
risk averse when statistics are stated in terms of gains and risk-
seeking when the same statistics are framed as losses (Hogarth,
1987). By extension, decision-makers should be risk averse and
select more effective methods (be risk averse) when effectiveness
data are presented positively or as gains.
Decisions are often made based on "heuristics," relatively
automatic mental strategies that serve to simplify the decision
process. While often adaptive, heuristics can oversimplify the
decision process (Redelmeier et al, 1993). The program was
designed in part to help the user compensate for this tendency. A
few examples are below.
With respect to contraceptive decision-making, PP staff
noted that patients often request a birth control method because a
relative or friend uses that method. In such cases, the decision
to use a given method derives from the greater salience of
information obtained from a friend or family member (i.e.,
availability), and from the tendency to shortcut thorough review
of all options (i.e., "satisficing"; a term coined by Simon,
1959) PP clinicians observed that such "decisions" frequently
close from consideration alternative and ostensibly better
methods.
The program overcomes tendencies to rely on "availability"
and to "satisfice" by providing a complete overview of all methods
(the computer presents comparative information aloud without the
option to abort), emphasizing individual differences and as space
permits, comparing alternative methods on a single screen.
Simultaneous comparisons may facilitate better appreciation of the
pros and cons of various methods (Hogarth, 1987).
Although the odds remain the same for independent events,
the Gamblers fallacy refers to the belief that "the odds of a
chance event increase if the event hasn't occurred recently." The
Gambler's fallacy or "representativeness heuristic" is at the root
of the misconception that the risk of pregnancy is greater at the
tenth than the first coital event. To help correct misconceptions
based on the Gambler's fallacy, the program emphasizes the
relatively constant risk of conception in the absence of good
birth control through creative use of multimedia.
Contrary to what might be expected by the representativeness
38


heuristic (Redelmeier et al, 1993), sex without conception often
leads to the self-perception of infertility (Rainey et al, 1993).
Thus, the program issues a "Warning" about construing ones own
infertility from the luck of "sex without conception." Where
possible, risk information is presented using graphics or analogy
to enhance its understandability and salience.
Redelmeier et al (1993) report that risk is often
dichotomized as either dangerous or safe regardless of exposure.
In effect, cumulative risks are often ignored. However, assuming
constant fertility, the average couple using a method of 95%
effectiveness has a conception risk of about 70% over a 10 year
period (Ross, 1989).
The computer explains cumulative risk by drawing an analogy
between the cumulative risk of death by auto accident (depending
upon the quality of car driven) and the cumulative risk of
conception (depending upon the quality of contraceptive used).
Presentation Style. Program design considerations included
sensitivity to the psychological and emotional factors that might
affect user interaction with the program, (i.e., sexual
embarrassment, sexual guilt, erotophobia, etc.). Thus, the
program attempts to normalize sexual activity by speaking
unabashedly but reassuringly (especially for the sake of the
sexually inexperienced), about sex and birth control(e.g., even if
you have sex only once in awhile, it is important to use a "good
method of birth control every time).
Most of the program is read aloud. To create a friendly and
supportive environment, verbal reinforcements (e.g., "good work",
"correct") are used where appropriate, and gentle, assuring vocal
tones are used throughout the program.
To facilitate identification of contraceptive methods,
photographs of each method were taken using a digital camera and
downloaded into the program. Other attributes included in the
program include color, motion and audio-visual clips, and are
aimed at making the program engaging, fun, relaxing, user-friendly
and credible (discussed further in phase two below).
Data Storage and Report Generation. For each question, user
responses were stored to a data file and "time to answer a
question" data were stored to a tracking file. The program has
the capacity to generate reports based on stored data. While the
report capabilities of the program have been tested, this feature
was not available at the time of data collection.
Program Testing. Program testing included several
strategies. Throughout development, the program was tested
repeatedly by the developer and occasionally by volunteers to be
sure that the algorithm and accompanying features worked properly.
After questions and information were added to the program, a
family planning clinician reviewed "print screens" of the program
and a detailed explanation of the program algorithm. Finally,
clinic management and training staff at Clinic C (an area family
planning clinic not associated with PP) completed the program and
provided additional feedback. Requisite changes to clinical
information were made, and all known program bugs were resolved
before initiating the pilot test phase.
39


Phase Two Pilot Testing
Materials and Procedure
Following computerization, a preliminary and limited pilot test of
the completed program was conducted. Constrained by time and
finite monetary resources, design limitations included the
following: a small convenience sample of 26, no comparison
groups, and limited outcome data (e.g., actual contraceptive
choice, method adherence, and unintended pregnancy rate data were
not collected) The pilot test phase did include the development
of pre/post assessment tools targeting (a) improvements in family
planning knowledge, (b) changes in expressed contraceptive choice,
(c) patient experience with the program (e.g., satisfaction, user-
friendliness) .
The pre-assessment tool contained a total of 14 knowledge
questions (5 reproductive biology and 9 medical methods) and one
question concerning expressed method choice. Although only a
proxy for actual method choice, the choice question, "What method
would you choose if you had to pick a method right now?", was
designed to capture the effect of the program on method choice.
Knowledge evaluation items were paraphrased derivatives of
information provided in the program.
Containing a total of 26 questions (plus one method choice
question), the post-assessment tool included all of the same 14
questions asked in the pre-assessment (Appendix G) plus 12
questions concerning patient experience with program attributes
(Table 4.2 in Chapter 4). Questions evaluating program attributes
and patient experience with the program were based on other
studies involving computers as educators (Lapham et al, 1991;
Skinner & Allen, 1983).
Because answers to the pretest questions were provided in
the program, posttest scores were expected to provide a measure of
knowledge gains due in part to the program (and in part due to the
pretest experience). To reduce the influence of "testing
effects" on posttest performance, patients were informed only that
there would be family planning questions before and after the
program. It was not indicated that these would be the same
questions.
As with the program itself, assessment tools were
computerized using Authorware software. Following development and
testing of the program, the data collection phase of P2 was
initiated.
Site Selection. Indicating that the duration of the
project exceeded expectations, PP discontinued their participation
before phase two could begin. With the help of the Colorado
Department of Public Health and Environment, the support of Clinic
C, an area family planning clinic not associated with PP, was
enlisted.
40


Population Demographics. Over calendar year 1997, Clinic C
reportedly saw a total of 2091 patients. Unfortunately, the
available data must be interpreted with caution. For several
parameters, ethnicity, parity, and method, the raw numbers do not
sum to 2091 and are incorrect by as much as 31%. Where the data
do appear more stable (age and poverty), demographic differences
between clinic C and clinics A/B are apparent.
For the sake of interpretation, age and poverty level were
recoded as younger age (18-24) vs. older age (25 and older) and
poverty (at or below poverty level) vs not poverty (above poverty
level), respectively. By Chi Square analysis, significant
differences between Clinic A/B and Clinic C patients were found
for age (p<.000) and poverty level (p<.000). Compared with
Clinics A/B, Clinic C serves a greater proportion of low income
(51% vs 15%) and older clients (64% vs 54%), perhaps because
Clinic C serves a large population of college students.
Participant Recruitment. Eligible participants were Clinic
C clients over 17, English-speaking, not pregnant, and judged by
staff to be cognitively and physically able to participate.
Project participants were first screened for project eligibility.
Per a written script, those judged eligible were asked if they
would be willing to participate in a project concerning
contraceptive choice.
A total of 26 Clinic C patients agreed to complete a 30
minute computer program along with a 5 minute pretest and a 10
minute post-test (45 minutes total).
Patient recruitment strategies for P2 depended upon several
approaches, effectively creating a convenience sample. Most
participants were actively recruited by telephone from the pool of
Clinic C's established patients, some were recruited from the
patient waiting room, and using a snowball strategy, a few were
referred by friends of project participants (numbers on the
relative proportions are not available). Referred patients were
Clinic C clients.
Patient recruitment information was not obtained due to the
burden of patient recruitment on an already busy clinic. Thus,
enroller and refuser rates were not obtained. Given the small
sample size and the unreliability of the population data, analyses
for representative sampling were not conducted.
Once agreeing to participate, respondents were directed to a
private room where they signed a consent form, and completed the
program and the computer-administered pretest and posttest. As
described in PI, patients were fully informed of their rights as
participants and consent forms were signed.
Of the 26 project participants, 5 (19%) had a medical
condition that correctly caused the program to abort with a
gently-worded recommendation to consult with a clinician. This is
expected because certain medical conditions require direct
interaction with a clinician. For purposes of data collection,
the program was allowed to proceed in two cases where the medical
condition was reportedly already resolved. Although the exact
41


number is unknown, the majority of participants were not seeking
birth control at the time of project participation and were
encouraged to see an on-site clinician when concerns arose.
On average, data collection took less than 45 minutes per
respondent (estimated times): (a) 5 minutes for the computer-
administered pre-assessment, (b) 20-30 minutes for the computer
contraception program, and (c) 10 minutes for the post-assessment.
Data in the form of patient responses to questions were
automatically stored to text files on the computer. Patients
received $20 for their participation.
To start the program, the respondent received instructions
from the computer to "press any key on the keyboard," after which
the program maintained all interactions with the user. For the
benefit of novice computer users, the program began with a
tutorial in which the user was required to point and click a mouse
on various objects. Experienced users could simply click a "move
on" button.
42


CHAPTER 4
PILOT STUDY RESULTS
Phase one program development constituted the primary focus
of the project and is described thoroughly in the Methods section.
The secondary, and significantly smaller phase two (P2) pilot test
provided an opportunity to alpha test the program and to obtain
preliminary data about its effects on family planning knowledge
and contraceptive preference in a clinic setting. The results of
phase two analysis are below.
Pilot testing focused on (1) improvement in family planning
knowledge (i.e., knowledge of reproduction and contraceptive
methods), (2) changes in stated contraceptive choice, and (3) the
overall user experience (i.e., was the program perceived as
friendly, credible, etc.). Restated as questions, P2 focused on
the following: (a) "How well did the program function?" (b) "What
affect did the program have on contraceptive knowledge?" (c) "How
did the program affect contraceptive choice?" and (d) "How did the
user feel about the program?".
How Well Did the Program Function?
Several parameters were used to judge the adequacy of
program functioning: (1) the mechanics (i.e., was the program
error free?), (2) data management (i.e., did the program post
response and tracking data to files correctly?), and (3) the
algorithm(i.e., did the program respond appropriately to
individual users?).
Overall, the program performed well in the alpha test phase.
Although participants were encouraged to report any problems
encountered, the few programming errors detected were minor and
corrected with the first four participants. In no case did the
program "crash."
Respondents were too alike to fully test all program facets.
For example, all of the respondents had experience with medical
methods. Thus, program segments dedicated to those without
medical method experience went untested. Nevertheless,
descriptive statistics based on tracking data and user responses
to questions (discussed later) revealed no major problems.
Behind-the-scenes program functions, such as "determining
contraceptive user status" and "posting data to files", occurred
with very few problems.
Despite the overall good program performance, some
programming oversights/errors were identified. For example,
current users were asked a series of questions to determine the
"wellness of use" (WOU). The result of the WOU evaluation is a
simple dichotomy "OK" or "Not OK". Although a key analytic
variable, WOU was mistakenly not posted to a data file and had to
be recreated.
43


What Affect Did the Program Have on Contraceptive Knowledge?
The pilot test phase included 26 Clinic C participants. Of
these, 23 (89%) completed the full pilot test and 3 (11%) were
excluded for medical reasons; 1 because of a previously detected
breast lump and 2 because of unexplained bleeding in the past 4
months.
Family planning knowledge was determined by comparing mean
performance at pretest vs. posttest using the same 14 questions; 5
focused on reproductive biology and 9 on contraceptive methods (at
least one question per medical method).
Power analysis showed that a sample size of 30 would be
necessary to detect a 20% difference between means for a Paired T-
test, beta=.90, p<.05. Although the actual sample size was
smaller, N=23, significant differences were found.
Based on the mean comparison, contraceptive knowledge was
38% greater at posttest (mean number correct=ll.9) compared with
pretest (mean number correct=8.6), a mean improvement of 3.3. By
two-tailed Paired T Test, a significant difference was found, t
value=7.552, p<.000; [CI]=2.40s3.30s4.21, 95%.
Descriptive data showed that no respondent answered all
questions correctly at posttest. Further, the two respondents
scoring highest at pretest (12 of 14 correct responses) actually
answered one less question correctly at the posttest. These data
suggested that the program did not serve everyone equally well as
an educational tool, and triggered a more in-depth data analysis.
A total of 322 pre/post response combinations occurred (23
respondents x 14 items). Table 4.1 below shows the number and
percentage of responses by response combination category.
Table 4.1
Pretest-Posttest Performance Comparisons
Pretest right-posttest wrong 4.3%
Pretest wrong-posttest right 27.9%
Pretest right-posttest right 57.1%
Pretest wrong-posttest wrong 10.6%
Overall, 274 (85%) of the 322 item response combinations
were correct at the posttest, with nearly one-third reflecting
learning (#2 above). Of the 322, 48 (15%) were incorrect at
posttest, indicating either "nonlearning" (10.6%, #4 above) or
"unlearning" (4.3%, #1 above).
To understand whether particular respondents might account
for a majority of the non- or unlearning, data were reviewed "by
respondent" (Table 4.2).
44


"By respondent" analysis showed that unlearning occurred in a
total of 12 (52%) cases, with 10 of the 12 experiencing unlearning
for only one item, and the remainder experiencing unlearning for
two items. Data also showed that nonlearning occurred in a total
of 21 (91%) cases. Of the 21 cases, 11 (52%) experienced
nonlearning for only 1 item, 9 (43%) experienced nonlearning for 2
items, and one (5%) experienced nonlearning for 3 items. In
effect, unlearning and nonlearning were fairly ubiquitous events
and not respondent-specific. Review of item data proved more
revealing.
Masked by the mean data, an item-by-item review of pre/post
data revealed that unlearning occurred for 5(36%) of the 14
knowledge test items. Specifically, one or more respondents with a
correct response at pretest responded incorrectly at posttest on
the following items:
(1) "There are about 250,000 sperm in a man's ejaculate" 2
respondents;
(2) "If 10 women did not use birth control for a year, how
many would get pregnant" 1 respondent ;
(3) "Aside from 'having your tubes tied', the pill is the
best method against pregnancy" 7 respondents;
(4) "A woman who uses the IUD, is protected against
pregnancy for at least one year" 3 respondents;
(5) "Having the tubes tied" is a safe and permanent method
against pregnancy" 1 respondent.
As these data show, of 14 pre/post responses, item #3
accounted for exactly half of the unlearning experienced, and item
#4 accounted for another fifth. Together, these 2 items accounted
for over 70% of all unlearning, suggesting the need for revision
to the two questions and/or to the relevant program content areas.
How Did the Program Affect Contraceptive Choice?
To evaluate the program effect on contraceptive choice,
respondents were asked at pretest and posttest to identify the
method they would choose right now (expressed choice). Changes in
expressed choice at posttest compared to pretest are referred to
here as "method shifting".
In general, a positive program effect would be a method
shift from less effective to more effective methods. However,
some respondents are current users or recent past users (less than
one year) of medical methods, while others are not. Still others
prefer to continue using their current or past-used method, while
others do not. Current or past medical method users, who prefer
to continue using that method, are evaluated for WOU. These
factors complicate the analysis of method shifting.
Data were analyzed using several different denominators:
(1) all respondents (n=23); (2) medical method users only (n=15);
(3) nonusers only (n=8); and (4) those who are judged by the
program as "in need" of a different method (i.e., candidates for
method shifting, n=14). Respondents are considered candidates for
methods "method shifting" who (1) are confirmed by WOU evaluation
45


as good users of their current or recently used method, but
indicate a possible interest in other methods, or (2) "fail" the
WOU assessment. Respondents are not considered candidates for
"method shifting" who (1) are confirmed by WOU evaluation as good
users of their current or recently used method, and (2) would
prefer to use that method. In effect, their only method choice is
"program confirmed".
Analysis of Method Use
Overall, analysis of method shifting proved complex, and
again, the small sample size proved a limiting factor. While the
overall rate of method shifting (N=23) was 13%, rates as low as 8%
were found for the medical users (i.e., those who ever used a
medical method) and as high as 25% were found for the nonusers
(i.e., those who had never used a medical method).
Because women using their partner's vasectomy are not
evaluated for WOU, only 13 women were evaluated for WOU. Of the
13, 11 (85%) were confirmed for WOU, 2 (15%) were not. Method
shifting was found for only one woman in the medical methods
group, and she had passed the WOU.
Obviously, it is not entirely reasonable to evaluate program
performance based on such a small sample. Fortunately, additional
insight into the complexity of user circumstance vis-a-vis program
function can be gleaned from individual patient profiles. Two
respondent types are of particular interest for the program, (1)
users who fail the WOU evaluation, and (2) nonusers of medical
methods.
Respondent Profiles
One respondent was 25, married, had no children and had
never had an unintended pregnancy. She wanted to become pregnant
in the next three years but was unsure if she wanted pregnancy
within the next 12 months. She was a current pill user but
indicated that she "sometimes" had sex without her method because
she ran out of or could not afford it. She also indicated that
she sometimes worried she might be pregnant. Based on the above
responses, a negative score on the WOU resulted, and other methods
were suggested as "good or better" than the current method. This
woman did not change her method choice.
Another respondent was 28 years old, single, did not live
with her sex partner, and wanted to become pregnant within the
next 3 yearsbut not within the next 12 months. She would be
unhappy if she became pregnant now. She had had only one partner
in the past 12 months and expected to have sex in the next two
weeks. She was currently using the condom, suppository and
withdrawal. She sometimes had sex without birth control because
she ran out of it, and she estimated that she was "somewhat"
likely to get pregnant in the next 12 months. She had never had a
child or become accidentally pregnant. This woman indicated that
she would change her method choice to the pill.
Patient profiling revealed the need for future program
review and modification. For example, a woman who was monogamous,
46


previously used the pill and currently uses only her partner's
vasectomy was inappropriately assessed for WOU for her past-used
method. Although vasectomy is disallowed as a medical method
(because it is not female-controlled), WOU assessment of a past-
used method was probably not warranted in this situation.
Although only medical methods should be assessed for WOU,
patient profiling revealed that condoms, a nonmedical method, were
evaluated. This problem probably stems from an earlier program
design approach that was incompletely corrected. No other
nonmedical methods were WOU assessed.
Presentation Parameters
Post-program pilot testing included the number and percent
of patients responding to each attribute below.
Table 4.2
Program Attribute Presentation Ratings
Program feature Response percentages*
Program Pacing Too fast 0 About right 87 Too slow 13
Amount of information presented Too little 4 About right 96 Too much 0
Actual vs. expected amount learned Less than expected 9 Amount expected 48 More than expected 44
Felt program was accurate Yes 87 Somewhat 13
Program induced tension No 96 Yes 4
Program was confusing No 100
Program was easy Yes 100
Program was friendly Yes 96 Somewhat 4
User experienced with computers Yes 78 Somewhat 17 No 4
More honest with... Clinician 13 Computer 26 Same 61
Like talking computer Yes 78 Somewhat 13 No 9
Recommend to a friend Yes 96 No 4
*Cells may not total to 100% due to rounding error.
Note: N=23.
47


How Did the User Feel about the Program?
Presentation style ratings indicated a favorable experience
for most program users. Although over 20% were, at best, only
"somewhat" experienced with computers, 100% felt that the program
was both easy to use and not confusing. Over 95% indicated that
they would recommend the program to a friend.
The literature suggests that respondents tend to be less
forthcoming in face-to-face interviews. However, twice as many
users reported that they felt they could be more honest with the
computer (26%) than with a person (13%), but almost two-thirds
(60%) stated that they could be equally honest with both. The
diference between the literature and the findings here may be
attributable to the fact that respondents in the current study
were projecting their expected not their actual behavior.
Clinic C patient and staff evaluations of the complete
program, including the relative order of scales, and of items
within scales as well as respondent experience with the program,
were informally obtained by the investigator immediately following
program completion. Because the opportunity for post-program
debriefing was not expected, and because the debriefing was brief
(lasting about 5 minutes in most cases), few of the same questions
were asked across participants.
Results of the "interviews" suggested that the program was
very well received and that the multimedia format enhanced the
presentation. As with the PP patient focus groups, a number of
respondents felt that women who were naive about medical
contraception would benefit most from the program. Others felt
that the program should be completed by all patients, noting that
the program prompted "thought" about alternative methods. Noting
her religious orientation, one woman acknowledged that she found
the program disconcertingly "cavalier" about the sterilization
methods. One of the "medically ineligible" women reported that
the program seemed to end abruptly and perhaps somewhat alarmingly
upon determining her medical status.
Data Management Results
Three types of data were automatically collected by the
program: (1) a computer-generated unique identifier (full
date+time; e.g., 080119980901) to allow file linking of same-
patient data, (2) user responses to program questions, and (3)
tracking data or "time spent in the interaction," including both
"time spent answering the initial question" and "time spent
confirming the answer." Although a unique identifier for each
patient was simultaneously posted to all data and tracking files,
personal identifiers were not collected by the program. Because a
unique identifier is generated anew with each visit, tracking
patients across visits is not possible.
Although user responses to questions were of only moderate
value in program testing and evaluation, program frequency data
48


did allow for review of unexpected or out-of-range data and was
useful to the method shift analysis discussed earlier. (The
program does not conduct real-time edit or range checks, although
for nearly all questions the only available response options are
those presented by the program.)
With respect to tracking, each question consisted of two
elements: (1) the initial question( e.g, "How old are you"),
and (2) the confirmatory question (e.g., "So you are X [insert
user age] years old?"). The number of minutes spent within each
element was tracked automatically.
Tracking data were collected to help identify outliers. To
that end, tracking data (i.e., time spent answering a question)
were analyzed relative to their respective grand means for both
the primary (mean 5.01, median 5.15) and confirmatory questions
(mean 1,88, median 1,80).
Relative to the grand mean, data were identified as outliers
that fell in the lower and upper quartiles for the primary
questions (3.03 and 6.43, respectively), and for the secondary
questions (1.50 and 2.32, respectively). While a more
conservative cutoff might have been used, such as the lower and
upper 10%, the quartiles permit a wider range of potential
problems to be identified, an approach that seems warranted at
this early stage of development. Outlier data are understood as
indicators of response burden and will serve to direct future
program modifications. Of course, response burden can be due to a
weakness in item construction, respondent processing demands, or
both.
Based on the above data, the mean total time spent per
interaction was 6.89 seconds. However, "the sum of time spent in
all interactions" seriously underestimates the "total amount of
program time", which includes the significant time spent with
information items and, to a lesser extent, with secondary
questions (presented to patient subsets). Due to a programming
fault however, the total time to complete the program could not be
calculated exactly. Total estimated time was 20-30 minutes
depending upon the user. Because pretest/posttest data are not
expected to be retained with the program, relevant tracking data
were considered unimportant and not collected.
Tracking data were also used to determine whether the value
of confirmed responses could be justified in terms of the time
spent confirming. Given an average response time of less than two
seconds to confirm, and participants' feedback concerning the
value of confirming responses, the process seems worthwhile.
49


CHAPTER 5
SUMMARY, LIMITATIONS AND IMPLICATIONS
This chapter is organized into three main sections: Summary,
Limitations and Implications. The Summary and Limitations
sections are divided into a program development (PI) and pilot
test (P2) phase.
Although significantly more resources, including time,
effort and project monies (85%) were dedicated to development than
pilot testing, development proceeded with full knowledge that this
was only the first (version 1.0) of several program versions.
Indeed, post-dissertation expectations are for further refinements
to version 1.0 before release. The Implications section is
concerned with how the program might be utilized in a clinic
setting. Other implications that might be drawn based on P2 data
must be considered "suggestive" only.
Phase One
Summary
This project was concerned with the development and pilot
testing of a computer program to assist women with their
contraceptive decision-making. Program development included a
three step process: (1) the development and administration of the
self-administered Birth Control Choice Questionnaire (BCCQ); (2)
the development of family planning information items; and (3) the
computerization of BCCQ and information items. Computerization
included the introduction of multi-media into the program,
including color, audio, birth control method photographs, and
video clips.
BCCQ Development. Development of questions and information
items proceeded from a review of the recent scientific literature,
feedback from PP staff and patient focus groups, and discussions
with individual family planning professionals and dissertation
faculty. Based on an iterative development-evaluation process
that included a series of focus groups and Factor and Internal
Consistency Reliability analyses, three drafts of the BCCQ were
developed.
Draft one (123 questions) was developed following an initial
review of the literature and several discussions with family
planning professionals. Continued literature review and additional
feedback from individual family planning professionals and
dissertation faculty resulted in draft two (105 questions).
Staff evaluation of BCCQ draft two suggested good content
validity (i.e., a maximum rating by all reviewers) for 8 (67%) of
12 scales. The four scales receiving less than a maximum rating
50


were reviewed and modified.
Draft three item revision, directed by Internal Consistency
Reliability (ICR) and Factor analyses (FA) as well as focus group
input, led to the discovery of some additional factors and the
reinterpretation of several others. All but three factors
identified were retained for later use as program questions.
Table 3.2 provides a summary of the scales as developed through
the various stages of the BCCQ.
The combination of staff focus group and statistical
analyses of the BCCQ resulted in significant item and scale
modification (the reconstitution of items contributing to scales).
Of the 12 original BCCQ draft one scales, 3 were dropped, 9 were
retained but with modifications, and another 6 were added;
yielding a final set of 15. A set of "True-False" items received
mixed staff review and was ultimately converted to information
items.
Feedback from the patient focus groups confirmed that
program development was "on the right track." For example,
questions such as "What features are you looking for when choosing
a method of birth control," yielded patient responses such as
"convenience," "ease of use," and "lifestyle fit," thus confirming
data obtained from the staff focus groups and the literature. At
the same time, the patient focus groups confirmed the potential
value of the program. For example, when asked "Where do you get
your information about birth control", one participant stated that
when she originally obtained Depo-Provera, "The first person I
talked to didn't mention the bleeding." Bleeding was only
confirmed as a Depo-Provera side-effect at her second visit three
months later. Because the program standardizes the information
presented, this user would have been informed of Depo-Provera
side-effects such as bleeding, at her initial visit.
Program development presented a number of challenges. For
example, the seemingly simple process of shooting good quality
digital photographs of contraceptive methods was surprisingly
difficult; a process made worse by software idiosyncracies, as
photographs were passed from one software package to another. Two
personal communication audio-visual clips (in which a person
appears on screen and speaks to the patient) required numerous
"takes" and still needs improvement.
Due to the subjective nature of clinical decision-making,
same segments of the program algorithm and associated screens were
sometimes revised several times (e.g., differing perspectives
about the appropriate age cut-off for pill use among smokers).
The program includes several key elements, primary vs.
secondary questions, contraceptive user status, method preference,
wellness-of-use (WOU), and presentation of the three best birth
control options. Each of these elements serves to personalize the
program, and all performed well.
Integrating the key elements is a complex program algorithm.
While it is extremely unlikely that the current algorithm, as is,
would completely satisfy all those involved in its development,
feedback suggests that the program is compatible with the
perspective of most and therefore usable by most. In fact, the
majority of family planning professionals who used or observed the
final program gave positive reviews, and several indicated
interest in future use or in the development of population-
51


specific modules (e.g., a version specifically for teen males).
Program Form and Function. Several key elements determined
the nature of the program: the questions, the program algorithm,
the information items, and the "three best methods". Each element
has implications for the type of program that has finally emerged.
As a purely educational tool, the program might have asked
no questions, but simply taken the user through a series of
educational screens. The questions allowing user input made it an
interactive tool however, enabling the program to function as more
than an electronic page-turner.
Ignoring the data suggesting that information needs to
address specific risk behaviors if it is to effect behavior change
(Fishbein et al, 1994), the program might have presented the same
questions and information to all respondents. By employing a
rather sophisticated algorithm, however, the program presents
questions and information contingently, focusing on issues such as
personal risk and WOU for the preferred current or past-used
method.
By developing a program that helps patients identify
personal risk, as per the HIV prevention counseling model (Centers
for Disease Control and Prevention, 1989), the program is more
likely to effect behavior change (Kamb et al, 1998). Where the
program cannot satisfy specific components of the CDC prevention
counseling model (e.g., assertiveness training), the program may
enable care providers to do so by answering other patient
questions and concerns upfront. Further distinctions between
purely educational and counseling models are of interest to this
discussion.
Llewellyn-Thomas (1995) distinguishes between patient
education tools and decision aids. By design, the program
incorporates many of the elements of a decision aide including:
(1) the provision of information specific to the individual; (2)
the facilitation of patient choice among treatment alternatives;
and (3) the provision of patient information aimed at improving
the treatment choice (as opposed to improving patient
participation in a predetermined treatment plan). This is
significant because it sets the stage for patient involvement in
treatment selection (Llewellyn-Thomas, 1995), or in this case,
contraceptive selection.
Information Items. As a function of its education and
behavior change components, the program can be viewed as a shared
decision-making tool. "Shared decision-making" or "demand
management" can be defined as "the use of decisions and self-
management support systems to enable and encourage its consumers
to make appropriate use of medical care" (Vickery & Lynch, 1995,
p. 552). With the goal of "appropriate use," demand management
focuses on "patient empowerment". Patient empowerment means
involvement of the patient as a responsible member with the
health-care team (Anderson et al, 1991). One mechanism for
patient empowerment includes the provision of relevant health
information in an unbiased and objective fashion; an objective of
the program throughout all development phases.
As it empowers, information helps the patient arrive at a
rational health or medical decision based on the synthesis of
52


benefits and risks (Vickery & Lynch, 1995). However, it is well
known that the synthesis of benefits and risks depends upon what,
and how, information is presented. Llewellyn-Thomas (1995) offers
these and other questions: "How detailed should the information
be?", "What types of benefits should be described," "How should
probabilities be described in numerical, verbal, or graphic
form? (p. 105)
For these and related questions, the answers depend on the
combination of sound judgment and the scientific literature. For
example, the literature suggests that the benefits of hormonal
contraceptives are little known (Potter, 1991). Thus, for the
current project, a decision to highlight contraceptive benefits
was made. Further, because it is so widely used, the benefits of
the pill were noted along with a recommendation to talk to the
provider about the salutary nature of other hormonal methods.
The answers also depend on constructing items that
compensate for the heuristic tendencies of decision-makers. Thus,
program items were written in simple language with complementary
multimedia, and with attention to the cognitive factors affecting
decision-making such as primacy/recency effects (i.e., memory is
better for items presented first and last in a list), framing
effects (e.g., decisions tend to be more risk averse when "framed"
in terms of gains as opposed to losses), and risk dichotomization
(decision-makers have a tendency to dichotomize risk as "all or
none"). For example, with primacy effects in mind, one of the two
"personal communication" audio-visual clips was purposely selected
and designed to appear at the start of the program. By
taking heuristic tendencies into account, the program helps
compensate for human tendencies that bias information.
Nevertheless, the program is directive. Not only is it oriented
toward medical methods, but it uses patient input to determine a
list of best method options. (This approach seems consistent with
current interests in "Informed Choice", whereby the physician
assesses the patient, reviews treatment options in terms of risks
and benefits, and the patient then selects from among the options
provided based on personal preference [Vickery & Lynch, 1995].)
Program content derives from empirical evidence and clinical
expertise and serves to orient the program toward the medical
methods, particularly the user-independent methods. In fact,
aside from the male condom, no information is provided on
nonmedical methods. Based on this orientation, the program
concludes by presenting the user with her three best medical
method options.
The "three best" approach facilitates good decision-making
by using patient responses to program questions to reduce the set
of birth control options from over a dozen to a more manageable
number. Within this approach a hierarchy further influences which
methods will be offered as options by resolving ties in the
direction of more efficacious methods.
Unfortunately, the more efficacious methods are those
typically seen as being associated with side-effects. However,
given the myths and misinformation about birth control that
characterize the knowledge base of many, education has the
potential to yield a more accurate and lower estimate of true
potential harm, and facilitate the choice of better methods
(Balassone, 1989).
53


The Program as a Patient Education Tool. Education is a
necessary but not sufficient condition for behavior change
(Healton & Messeri, 1993; Allgeier, 1983) Yet, patient education
and involvement in medical decision-making can improve health
outcomes, including adherence (Kahn, 1993; Kaplan et al, 1989),
and help moderate liability concerns for care providers (Kahn,
1993).
In practice however, systematic patient education is too
often lacking (Kahn, 1993). In a recent survey of nearly 1000
college students, Sawyer and Pinciaro (1998) found that knowledge
about Norplant and Depo-Provera was relatively poor compared with
other hormonal methods, and that improved knowledge about these
methods resulted in a greater likelihood of their later use.
Other studies have found that more reliable contraceptors are more
educated about contraception (Sander et al, 1992).
These studies have implications for the education value of
the current program, especially where patients are "nonusers" or
are established users who are inadequately informed about family
planning. Among established users, inadequate knowledge can be
due to several factors: inadequate patient education protocols
and materials, provider bias, or inexperience with specific
methods (Westhoff et al, 1993). Unfortunately, the program may
prove less beneficial in the case of care provider bias. It is
noted that provider bias may be due to the same factors (e.g.,
misinformation) that affect patients.
limitations
Program development depended first upon generating a
question set that would drive the program algorithm and would
determine, in part, the kinds of information to be provided. The
program question set was the outcome of an iterative development-
evaluation loop that resulted in three drafts of the BCCQ, with
the final draft being completed by 182 family planning clinic
patients. This process was central to the development of
questions and information items.
Items and scales were rated by PP staff for burden, clarity,
and the appropriateness of response options as well as the content
validity of the scales. Although anticipated, time and staff
constraints precluded the opportunity to apply ratings to the
final question set. The evaluation process should be repeated as
the next program version is developed.
This same kind of difficulty applies to draft three. Once
analyzed, BCCQ data were incorporated directly into the program
without further testing. As a future goal, other drafts will be
developed and administered to larger patient samples and the data
analyzed using the same techniques. This method will help yield a
more stable question set.
Sampling Method. PI participants were recruited from among
all eligible patients presenting to the clinic during the
54


recruitment period. Data were not collected on those refusing
participation. Further, reliability and validity testing were
extremely limited. Thus, project limitations make any effort to
generalize the results of the project premature.
Data Collection. Data collection required that all patient
responses and tracking data be posted to data files. All told, 1
of 50 (2%) question response data elements and 2 of 100 (2.OS)
tracking data elements are known to have been lost due to
programming errors (i.e., not posted to files).
The current project used monetary patient incentives
throughout all phases of data collection. Not surprisingly,
evidence exists for the positive effect of patient incentives on
participation rates (Stevens-Simon et al, 1997). Although the
utility of patient incentives for participation was not assessed
here, anecdotal evidence suggests that incentives may have been
helpful. On several occasions, recruitment rates improved only
after incentive amounts were increased (from $25 to $35 for
several of the focus groups).
Although incented data might seem suspect, some evidence
suggests that patient incentives can yield more valid data for
sensitive issues (Aral & Peterman, 1996). While interesting,
"whether", or "how", patient incentives affected data quality is
beyond the scope of this project. Other data quality issues are
germane however.
Potential Sources of Program Bias
Cultural Bias. Of 5,072 patients seen at two family
planning clinics, the majority were above poverty level (90%)
white (88%), had no children (74%), and used the pill (58%).
Available data suggested that those completing the BCCQ differed
in no significant way from the larger population. Although not
quantified, both staff and patient focus group participants
appeared demographically similar to the larger patient population.
Thus, program development proceeded with little input from ethnic,
poor, or high parity women and to that extent, the program may be
culturally biased.
PP Organization Bias. Focus group feedback depended
exclusively on staff from two PP affiliated clinics. Thus,
wherever the PP family planning perspective differs from that of
other family planning clinics, the program may be biased.
However, it is likely that such biases, where they exist, are
nominal for several reasons: (1) many biases about the suitability
of particular birth control methods were identified early
(although the true source of these biases is unknown), and offset
using the literature and the input of family planning experts with
nonPP affiliations; and (2) no PP-specific "biases" were observed
within the program by any nonPP professionals completing it. In
fact, several remarked that the program seemed to be "on target."
Developer Bias. The program was developed for a female
audience by a male. Aware of possible gender-based programmer
biases and audience preferences, compensatory efforts were made.
55


For example, evidence suggests that women prefer programs that
function as educational tools as opposed to games (Huff & Cooper,
1987) and are more sensitive to, and more accurate in their
interpretation of, facial cues than are men (Hall, 1978). Thus,
the program was designed with an educational style, purposefully
avoiding gamelike segments, and including two segments in which a
person speaks directly to the user.
One consequence of demand management and its tools is for
care provider liability. As patients become increasingly involved
in the decision process and as informed consent becomes more
"informed", care providers become potentially less liable for
outcomes (Kahn, 1993). This depends of course on assuring that
the content of demand management education materials is accurate,
unbiased, reliable, and valid.
To the extent that such liability concerns influenced
program development however, the program may be overly cautious
about the medical methods. For example, pre/post comparisons of
the item "Aside from 'having your tubes tied', the pill is the
best method against pregnancy", showed that 13 patients answered
this question correctly vs. 7 at pretest. In the absence of
clarification from respondents, these results seem puzzling.
However, the pill does represent the "status quo" in
reversible contraception, being easily the most widely prescribed
and used reversible medical method in the U.S.(IOM, 1995). In a
pill-popping society (Sinclair, 1995), other methods may be more
likely than "the pill' to be regarded by patients and even family
planning professionals with varying degrees of concern or
suspicion. To the extent that the developer unwittingly adopted
the status quo or responded to liability concerns, even
unconsciously, the presentation of user-independent methods might
have suffered. This issue needs to be considered in future
versions of the program.
Self-presentation Bias. Evidence exists suggesting that
self-presentation bias is less likely with computer surveying as
opposed to face-to-face (Aral & Peterman, 1996; Czaja, 1987).
Self-presentation bias will be discussed more later in this paper.
In addition to concerns about bias, other factors affected
program development, such as the clinic setting. That is, factors
related to how the program might optimize care from the
perspective of both patient and clinic staff.
Determinants of Program Development
While it is conceivable that the program could be used in
nonclinic environments (e.g., schools), development specifically
targeted the clinic setting. Other uses would require minor
program adaptations. For example, references to "clinic staff"
would probably need to be removed or reworded.
A program lasting no more than 20 minutes was recommended by
nearly all family planning staff. This meant that the time spent
asking questions had to be carefully considered.
As a time-saving strategy, question-answer program segments
designed to present information contingently (based on an
56


evaluation of user knowledge) were simply rewritten as
noncontingent information segments. This strategy increased the
average length of the program (because the information segment was
always presented), but decreased its maximum length (because no
one received both question and information). Unfortunately, this
strategy also resulted in fewer opportunities to personalize the
program.
Despite the above efforts, the program still needs trimming.
Current length is 20 to 30 minutes for most patients. Thus,
future program revision will have two main aims: (1) quality
improvement and (2) length reduction.
While evidence suggests that multi-media presentations can
benefit the education process and ultimately enhance learning, the
demands of multimedia on computer hardware are quite real. The
current program is over 400 megabytes with about half due to the
audio features. While the hardware demands of program development
are greater than those for program use, computers meeting only
minimum specifications will be sluggish; particularly with multi
media-saturated segments.
Finally, because the program was developed using Title X
funds (federal monies available for family planning activites,
excluding abortions) for use in Title X clinics, method costs
could not be itemized in the program. These federal restrictions
to presenting cost information will probably necessitate the
development of a program add-on containing cost information.
A Proposed Evaluation Project Study Design
This project has been successful in developing version 1 of
a computer program to assist women with their choice of
contraceptive method. Further development however will depend
upon a more sophisticated evaluation of program performance and
its effects on learning and contraceptive decision-making. One
expected evaluation will involve JSI Research and Training
Institute and should span 11 states; 6 in the Western states and 5
in the Midwest. A study design for the 11 state evaluation is
proposed below.
Two clinics will be recruited from each state for a total of
22 participating clinics. Because socioeconomic status is an
important factor related to contraceptive use, all clinics will be
classified prior to recruitment as either above poverty or below
poverty. In each state, one clinic will be randomly selected from
among those clinics above poverty and one from among the clinics
below poverty. Thus, 11 participating clinics will be classified
as above poverty, and 11 will be classified as below poverty. In
addition to the above criterion, clinics eligible for
participation will be those who have an IBM-compatible personal
computer and an area available for program administration. The
above data will be gathered via an internal survey.
Each clinic will collect data on 200 patients for a total of
4400 patients across 22 clinics. As table 5.1 below shows,
patients within each clinic will be randomly assigned to one of
four conditions:
57


Table 5.1
Prospective Evaluation Project Treatment Conditions
Pretest Program Clinic Posttest
Condition 1 (Enhanced) X X X X
Condition 2 (Substitution) X X
Condition 3 (Standard) X X X
Condition 4 (Hawthorne) X X
Patient Sample and Sample Size. Patients will be all those
presenting to the family planning clinic who are seeking a method
of birth control. All eligibility criteria related to the
dissertation project apply to the proposed evaluation.
This design will allow sufficient power to stratify analyses
by poverty level (above/below poverty). Power analysis showed
that a 20% difference would be detected with 225 patients per
condition, beta=.80, p<.05.
Outcome Measures. Four outcome measures will be collected:
(1) changes in pretest/posttest family planning knowledge;
(2) rate of method shifting among nonusers and medical
users;
(3) rate of method shifting among those at/above poverty vs.
those below;
(4) user experience ratings; e.g., presentation style.
Treatment Conditions. As table 5.1 above shows, the
proposed evaluation will have four treatment conditions:
(1) the enhanced condition will receive program assistance
followed by the usual clinic assistance;
(2) the substitution condition will receive program
assistance with minimal staff assistance. Minimal meaning
that (a) clinicians will cover only information additional
to that provided by the program, and (b) patients will be
encouraged to depend upon their interaction with the program
for their contraceptive decision making. Contraceptive
methods will be provided as usual. In essence, program
assistance is substituted for clinical assistance in the
contraceptive education and decision process;
(3) the standard condition will receive the usual clinic
assistance without any program assistance;
(4) the Hawthorne condition will receive pretest/posttest
interactions but no other intervention.
The above design will permit the following comparisons.
Condition 1 will provide a measure of maximum learning (program
and clinic). Condition 2 will provide a measure of program
58


learning only. Condition 3 will provide a measure of clinic
learning only. Condition 4 will provide a measure of the effects
of the pretest on posttest performance and of test-retest
reliability.
Analysis. Analyses will include univariate (frequencies and
Chi square) and multivariate statistics. Multivariate analyses
will use a repeated measures ANOVA (Analysis of Variance)
technique.
Using the above design, each clinic will randomly assign 25
participants to each of the four conditions. Data will be
combined across clinics to yield a total of 1,100 patients per
condition.
The proposed evaluation would be useful for determining the
effectiveness of the program and for directing future program
modifications. Outcome measures would provide the opportunity to
establish the reliability and validity of the tool and to
determine its effects on learning and contraceptive choice.
Evaluation of program fit in a clinic setting would be useful for
determining program-clinic fit (e.g., whether 10 or 20 minute
program versions would be more suitable).
Summary
The program pilot test effort was preliminary and limited. A
convenience sample of 26 patients tested the program. For
patients who refused participation or were ineligible, no rate or
demographic data were collected. Because of limited funds, no
comparison groups were used. Outcome data were collected at the
immediate posttest only. Thus, the results must be interpreted
with considerable caution.
Overall, results of program testing and evaluation were
positive. Not only did the program not crash, but it performed
with few programming bugs. Of the 26 family planning patients
alpha-testing the program, 5 were correctly identified as
ineligible because of preexisting medical conditions. When asked,
all respondents affirmed that the methods suggested as "best
options" seemed sensible (i.e., were consistent with responses
given).
The pilot test found statistically significant gains in
family planning knowledge and suggested that the program may
facilitate the choice of more effective methods. Respondents
indicated a positive experience with the program. Anecdotally,
the overwhelming majority of family planning professionals who
viewed the full program provided positive feedback.
As found in other computer-assisted health education
projects (Goldstein et al, 1994; Healton & Messeri, 1993; Kahn,
1993; Consoli et al, 1987), program pilot test data suggested that
the program improved respondent knowledge. Statistically
significant gains in family planning knowledge (reproductive
biology and contraceptive methods) by paired T-test were found for
23 patients completing the program, mean improvement of 38% from
pretest to posttest.
Data also suggested a possible program influence on
59


expressed contraceptive choice. Three of 23 (13%) respondents
indicated method shifting (i.e., a pretest-posttest shift in the
method they would choose "right now"). Although the small sample
size again calls these results into question.
Posttest evaluation showed that patient experience with the
program was quite positive. Of 11 program performance
attributes, 9 (82%) were rated positively by at least 87% of all
patients, results consistent with those of other studies (Kahn,
1993). Two attributes received less favorable ratings: (1)
affinity for the "talking" computer (78% rated positively), and
(2) user honesty with the computer (26%) vs. clinician (13%) .
That "talking" was rated lower than other attributes is not
completely unexpected.
A talking computer is expected to most benefit those whose
cognitive development or cultural background makes written text
difficult (Romer et al, 1997). Because all project respondents
had some college and 48% had at least a college degree, some may
have found the talking superfluous or distracting. (Note that the
audio is optional. Turning the volume off does not affect program
performance.) Nevertheless, future program versions should be
evaluated for reading level using, for example, the SMOG formula
(Contraceptive Technology, 1994).
Although contrary findings exist (Millstein & Irwin Jr.,
1983; Millstein, 1987), considerable evidence suggests that
computers can significantly reduce social desirability bias (Romer
et al, 1997; Aral & Peterman, 1996; Levine et al, 1989). Thus, it
is somewhat surprising that most users (61%) reported that they
could be equally honest with both clinician and computer. Only
26% felt they could be more honest with the computer.
As explanation, evidence suggests that volunteers for sex
surveys are more likely to be liberal in their attitudes toward
sex (Catania et al, 1990). Because of previous experience with
family planning clinics and with the processes involved in
obtaining birth control, project participants may have been
especially comfortable with face-to-face discussion of sex and
family planning issues (Herold & Goodwin, 1981). If so, those
newly seeking contraception or who are otherwise uncomfortable
about obtaining birth control may find computer interaction easier
than face-to-face interaction, and hence may benefit more from its
use. Alternatively, respondents were only stating what they
think they would do, actual behavior may differ substantially.
Future research will help determine the effect of the program on
response bias.
^imitation?
Sampling and Design Issues
The pilot test data must be interpreted with caution for two
main reasons: (1) the sampling method resulted in a relatively
small convenience sample, and (2) the study used a "one-group
pretest-posttest design", a quasi-experimental design in which no
comparison group is used (Cook & Campbell, 1979).
A convenience sample of 26 participants was recruited via
60


telephone, routine visit, and occasionally by patient referral.
As a condition of participation, all participants were established
clients of the family planning clinic. Serious threats to external
validity exist for P2 because of the recruitment method and the
absence of information about those who refused.
Lacking a comparison group, the pretest/posttest comparison
is fraught with difficulties for causal inference; including
history, statistical regression, maturation, attrition, and
testing effects (Cook and Campbell, 1979). However, many of
these factors are due to data-biasing events that might occur with
lengthy pretest-posttest intervals; (e.g., history, maturation,
attrition) and do not apply to the current project. Testing
effects are a likely issue however.
Testing Effects
Testing effects refer to the influence of pretesting on
posttest performance independent of the effects of the
intervention itself. Because no time intervened between
pretesting, program administration, and posttesting, testing
effects may be quite likely. Absent a comparison group and
information about test-retest reliability however, the testing
effects are impossible to quantify. Still, because the patient
instructions did not specify that the same questions would be
asked at pretest and posttest, the testing effects may have been
somewhat attenuated.
Although data suggest some program influence on
contraceptive choice (13% of the 23 respondents indicated a
pretest-posttest shift in "expressed choice" the professed
method of choice "right now"), two other issues seem relevant
here: (1) expressed choice vs. actual choice, (2) expressed choice
vs preferred choice.
Expressed choice is only a proxy for actual choice. Program
effects on actual method selection and use were not tested in this
preliminary pilot test. Thus, even if causality exists, the
actual effect of the program on method shifting is probably
overestimated. (It is noted that prechoice must always be a
proxy.)
More favorably, three of the four women who indicated a
method shift and passed the WOU evaluation, also indicated an
interest in a more effective method. In effect, these women had
no ostensible need to change methods because they used their
current method well, yet they reported a potential preference for
a more efficacious method. Possibly, these women indicated an
interest in other methods because program education improved
knowledge about the more efficacious medical methods about which
users are less likely to be fully informed (The Contraception
Report, 1998; Sayer & Pinciaro, 1998). Coupled with care provider
support, it seems that these women might have selected a more
effective method.
Program Algorithm
The program algorithm is complex, thereby making it
61


difficult to fully evaluate its accuracy and clinical validity.
At the same time, the small sample size, coupled with its relative
homogeneity, precluded thorough review of all program branches.
On the other hand, feedback from both patients and family
planning care providers completing the program indicated that it
functioned as expected, and that the contraceptive options offered
"made sense". Although criticism was consistently solicited by
the investigator, nearly everyone indicated that program feedback
was sensible and consistent with responses to questions. Thus,
while minor faults may exist, reviewer comments suggest that the
program is robust with respect to such faults.
Implications and Recommendations
Program Use in a Clinic Setting
Staff suggested that either (1) the program be available in
the waiting room and used on an ad lib basis, or (2) the program
be completed by all patients as part of the routine visit.
Although response was mixed, a preference for the "waiting room"
was evident. Preference for the waiting room approach may
reflect a lack of experience and confidence with the program.
Specifically, there may be a direct positive relationship between
experience with the program and dependence on it for routine care.
As a related issue, staff expressed different views as to
how the program might effect the patient visit. Some felt that
they would depend upon the program to cover the basics and would
use the visit to help the patient with more complex issues.
Others felt that they would feel obliged to review the content of
the program, although this was not necessarily viewed negatively.
Others held opinions that fell in between.
The question was posed to staff as to whether or not patient
information should be tracked across visits. To do so, patients
would need to be identified, at minimum, by a permanently assigned
ID# or by personal identifiers such as name. Most PP staff felt
that the program should function to help the patient as opposed to
staff. Expecting that some patients might be more honest with the
program than with clinic staff, the overwhelming recommendation
was that no personal identifiers be collected. A smaller majority
felt the patient should remain completely anonymous, and that
responses and data should remain confidential (unavailable to
staff).
Because the program has the potential to generate reports
for both patient and clinician, staff were asked: (1) who should
receive reports? (patient and/or clinician) and (2) what
information should be included?. Staff were split on whether or
not clinicians should receive reports (harkening back to the issue
of anonymity and confidentiality), but nearly all agreed that
patients should receive reports. The suggestion was often made
that patients should have the option to share the reports with the
clinician. If generated by the program however, patient and
clinician reports would likely contain different kinds of
information.
62


Education and Behavior Change
The program was designed both to educate and to effect
behavior change. To that end, data concerning expressed pre/post
choice of method were collected. Expressed choice is a proxy for
actual choice and can be understood as a measure of behavioral
intention.
The term "behavioral intention" immediately conjures up
sophisticated theories such as those generated by Albert Bandura,
(Social Cognitive Learning, 1983), Marshall Becker (Health Belief
Model, 1974), and Martin Fishbein, (Theory of Reasoned Action,
1975). At this early stage of development, the program bares
little resemblance to the product one might expect from such
theories.
At a 1991 workshop entitled "Factors Influencing Behavior
and Behavior Change", the above-named theorists and others derived
eight core variables believed to account for most of the variance
in behavior. Their list included the following: (1) intention (2)
environmental constraints, (3) ability, (4) anticipated outcomes,
(5) norms, (6) self-standards, (7) emotion, and (8) self-efficacy.
According to the theorists, the first three constructs are
necessary and sufficient for producing behavior and that the
latter five constructs function as mediating variables. Of these
eight factors, the program was most concerned with #4, educating
users about the advantages and disadvantages of performing the
behavior (e.g., choosing an effective contraceptive method), and
#8, assessing self-efficacy (e.g., the tendency to form habits as
relates to pill use). Future versions should take into
consideration others of these eight factors and further refine the
two areas already included.
63


CHAPTER 6
CONCLUSIONS
Each year, 137 million people are born, 53 million die and
the world's population grows by 84 million (Reid, 1998).
Geometric increases in global population growth translate into
"population doubling" every 35-50 years.
At first, the picture in the U.S. appears somewhat less
troublesome. The U.S. constitutes only 5% of the world's
population and its population doubles only once every 100 years
(Linden, 1992). However, the U.S. is a consumer, using 25% of the
world's energy, producing 22% of all carbon dioxide, and
accounting for 25% of the global gross national product (Elmer-
Dewitt, 1992). Yet, at an International summit (the United
Nations Conference on Environment and Development) held during the
Bush administration, U.S. delegates insisted that "the American
lifestyle is not up for negotiation" (Elmer-Dewitt, 1992, p. 58).
If so, the only reasonable alternative to the continued U.S.
onslaught on planetary resources may hinge upon reducing its
population growth rate. Given that the U.S. rate of unintended
pregnancy exceeds the rate of intended pregnancy, accidental
pregnancy would seem a reasonable starting point.
Since 1960, medical science has developed a number of
alternative methods of birth control, some of which all but
eliminate the risk of unintended pregnancy (Contraceptive
Technology, 1994). Among the methods available, the user-
independent methods perform especially well, eliminating many of
the human factors that drive the disparities between perfect and
typical use rates and thereby account for much of the unintended
pregnancy among those contracepting.
Compared with other methods, such as the barriers or
spermicides for example, the user-independent methods perform well
under the most adverse conditions. They protect against pregnancy
regardless of lifestyle, drug and alcohol habits or partner
objections. Moreover, they require only occasional (e.g., once a
year) interaction with the family planning care provider for
sustained protection. Yet, each year the rate of unintended
pregnancy in the U.S. remains high.
Luker (1975) and others have suggested that contraceptive
use can be attributed in part to how the benefits and risks of
conceiving are weighed against the benefits and risks of using
contraception. If so, accurate and complete information about the
available methods is crucial if a rational decision is to be
reached with respect to method choice and consistent use.
Evidence suggests that family planning education remains
wanting. Many who are sexually active function without adequate
knowledge or with misinformation about both conception and
contraception (Tanfer, 1994).
Missing and misinformation continue as obstacles to
effective contracepting for a host of reasons. However, even with
64


good information, decision-making is not a simple process.
Depending upon the task, decision-makers are variably inclined to
depend on heuristic devices for their decisions. Further,
depending upon how the task is presented or understood, decision-
makers are variably influenced by attributes of the decision task.
Given the significant need for improved contracepting, improved
information and improved decision-making, the computer presents
itself as a viable, if only partial, solution.
Recent forays of the computer into the health arena, have
demonstrated that it may have a place as educator and decision-
aide. Using the computer as educator, and coupled with the
support of the care provider, the patient may be enabled to make
better, more informed health decisions.
Returning to concerns about the high rate of unintended
pregnancy in the U.S., the question might be asked whether a
computer could not be used to help women with their choice of
contraception, with the expectation that improved choices would
mean more widespread use of the more efficacious methods.
The primary purpose of the present project was to develop a
computer program to assist women with their choice of
contraceptive method. Secondary aims focused on evaluation test
of the program.
The project was successful in that a computer program
suitable to use in a clinic setting was developed and tested with
positive results. Feedback from family planning professionals and
patients alike suggest that the program is not only user-friendly
but helpful as an educational tool. Pilot test data suggest that
the program improved contraceptive knowledge and may influence
contraceptive decision-making in the desired direction.
Considerable time and effort were dedicated to program
development by the programmer, dissertation faculty, over 20
family planning professionals and over 200 of their patients.
However, whether the program ever proves valuable to patients in
their contraceptive decision-making will undoubtedly depend upon
its acceptance by patients and, perhaps more critically, by clinic
staff. Program effectiveness in a clinic setting may depend
upon at least three items: (1) routine use of the program in the
patient visit, (2) unbiased clinic staff support for program-
generated method options, and (3) staff support of the patient as
equal partner in the decision process. Lacking any of these three
elements, program quality will probably be of little consequence.
To help ensure program utility in a clinic environment,
clinic "ownership" may be crucial. Indeed, while further
evaluation and development are necessary and should help improve
the program, these processes must be conducted with the support
and input of its potential users, both patients and clinicians.
With version 1 of the program completed, further evaluation
and development are certain. Given the interest expressed among
family planning professionals, it is evident that technology is
viewed by many as an ally. Hopefully, as version two is
developed, so too will the relationship between the tool and its
users.
65


Appendix A
Example Consent Form
Planned Parenthood has agreed to participate in a research
project about women's choice of a birth control method. Through
this research we hope to improve family planning care.
As part of this commitment to improved family planning care,
we are part of a project about how a computer might assist with
the choice of a birth control method. This project hopes to
increase our understanding of whether a computer can be of help to
a woman in her choice of birth control.
We are asking you to participate in the project. If you
agree to do so, you will be asked to be in a focus group. The
focus group will involve five other women like yourself and you
will be asked to review topics and questions which might be part
of a computer program to help women choose their birth control
method. Some of the questions deal with sensitive issues
concerning sex and birth control. Though you will not be asked
about your sex or birth control behavior, you will be asked what
you think of the questions.
The focus group will be audiotaped. However, the
information you provide will be anonymous and confidential. You
will not be identified by name in the focus group or on the
audiotape. Your clinician may know that you have participated in
this project but will not know what you say in the focus group
unless you decide to tell her. Once you have completed the focus
group, you should not discuss the comments of other focus group
members in any way which would identify that person to other focus
group members or to anyone else.
There are no known risks for participating in this project.
Your participation is voluntary. You may refuse to participate at
any time and such a decision will not effect the usual health care
you receive from your clinician. If you withdraw from the project
and do not complete the focus group however, no payment will be
made to you.
If you should have any questions at any time, you may ask a
PP staff member or you may contact the Principal Investigator,
Bill McGill at 436-7167. Further, if you have any questions about
your rights as a research subject, you may contact the University
of Colorado-Denver Office of Academic and Student Affairs at 556-
2550.
Please complete the bottom of this page if you are willing to help
with this project. I have read the above information and have
been given the chance to ask questions, which have been answered
to my satisfaction. There will be no additional costs associated
with this project. Therefore, I agree to participate in the
project about the use of a computer program for helping with birth
control choice and have been given a copy of this form.
Respondent's signature Date
66


Appendix B
Patient Distribution bv Acre and Site
1 Percent of patients (N=5072) N=2209
AGE Clinic A Clinic B Mean Clinic A/B Clinic C
14 and under 0.7 0.2 <1 0.1
fl 15-17 8.0 8.9 8.4 6.7
1 18-19 10.1 13.3 11.4 7.7
| 20-24 31.5 28.3 30.2 25.9
| 25-29 25.1 25.6 24.8 26.7
30-34 12.6 11.7 12.2 13.5
35-39 6.9 7.0 7.0 8.1
40 and older 5.0 4.9 5.2 10.3
Appendix C
and Site
Percent of patients (N=5072) N=1691
ETHNICITY Clinic A Clinic B Mean Clinic A/B Clinic C
Anglo 85.1 92.3 88.0 78.8
Afro American 0.6 0.7 <1 1.1
Hispanic 10.1 3.9 7.6 6.0
Native Amer/Indian 0.3 0.2 <1 <1
Asian/PI 2.3 2.1 2.2 2.8
Other 1.5 0.7 1.2 10.2
67


Appendix D
Patient Distribution bv Poverty Level and Site
Percent of patients (N=5072)
INCOME Clinic A Clinic B Mean Clinic A/B Clinic C N=2209
100% of Poverty or below 17.3 0 10.3 51.2*
101-150% of H Poverty 19.2 3.8 13.0 17.6
| 150-200% of Poverty 13.5 7.7 11.1 8.8
Above 200% of Poverty 23.1 80.8 46.4 22.4
Unknown 26.9 7.7 19.2 0
May be due to a significant college population.
Appendix E
Patient Distribution bv Parity and Site
Percent of patients
| PARITY Clinic A Clinic B Mean Clinic A/B Clinic C*
II None 73.4 75.3 73.5
One 12.2 8.4 13.3
Two 7.2 12.1 7.7
Three 5.8 2.1 3.8
| Four 0.7 1.6 <1
9 Five or more 0 0 0
| Unknown 0.7 0.5 1.2
Available data not accurate
68


Appendix F
Patient Distribution bv Birth Control Method and Site
Percent of patients
CONTRACEPTIVE METHOD Clinic A Clinic B Mean Clinic A/B Clinic C+
Cervical Cap 0.1 0.1 <1
Depo-Provera 17.3 14.2 16.0
Diaphragm 1.8 2.1 1.8
Emergency BC 0.3 0 <1
Foam/Condom/Sponge 6.0 7.2 6.5
IUD 0.7 0.5 <1
Natural Family Planning 0.2 0.0 <1
Norplant 1.0 0.3 <1
Orals 55.6 60.9 57.7
Surgical Sterilization 0.7 0.9 <1
None 15.2 12.9 14.3
Unknown 1.2 0.8 1.1
69


Appendix G
Knowledge Evaluation Questions.and Answers
ITEMS ResDonse ODtions (correct answer)
B1: There are about 250 thousand sperm in a man's ejaculate ("come") True/False/Don't know
B2: In the U.S., an accidental pregnancy happens about once every... 24 hours/Hour/10 minutes/10 seconds
B3: Once in a woman, sperm live for about 5 days. True/False/Don't know
B4: If 10 women did not use birth control for a year, how many would get pregnant 1/3/5/S
B5: Taking the pill is riskier than giving birth True/False/Don't know
Cl: The pill can be good for a woman's health True/False/Don't know
C2: Aside from "having your tubes tied", the pill is the best method against pregnancy True/False/Don't know
C3: If a woman doesn't want others to know she is using birth control, the Depo-Provera shot is a good choice True/False/Don't know
C4: A woman who uses the IUD is safe against pregnancy for at least one year True/False/Don't know
C5: Norplant implants keep a woman from getting pregnant for 5 years True/False/Don't know
C6: Barrier methods of birth control such as the diaphragm, cap and condom are usually free of side-effects (problems) True/False/Don't know
C7: "Having the tubes tied", is a safe and permanent method against pregnancy True/False/Don't know
C8: Emergency birth control (the "morning-after pill") is legal in the U.S. True/False/Don't know
C9: Birth control side-effects (problems) are usually temporary and not serious True/False/Don't know
70


BIBLIOGRAPHY
Alan Guttmacher Institute. (1994). Uneven and Unequal;
Insurance Coverage and Reproductive Health Services. New York:
The Alan Guttmacher Institute.
Allgeier, A.R. Informational barriers to contraception. In
D. Byrne and W.A. Fisher (Eds.)Hillsdale, NJ: Lawrence Erlbaum.
Anderson, R.M., Funnell, M.M., Barr, P.A., Dedrick, R.F., &
Davis, W.K. (1991). Learning to empower patients. Results of
professional education program for diabetes educators. Diabetes
Care, 14(7), 584-590.
Aral, S.O., & Peterman, T.A. (1996). Measuring outcomes of
behavioral interventions for STD/HIV prevention. International
Journal of STD & AIDS, 7 (Suppl. 2), 30-38.
Balassone M.L. (1989). Risk of contraceptive
discontinuation among adolescents. Journal of Adolescent Health
Care, 10, 527-533.
Baldwin, J.D., Whitely, S., & Baldwin, J.I. (1992). The
effect of ethnic group on sexual activities related to
contraception and STDs. The Journal of Sex Research, 29(2), 189-
205.
Bandura, A. (1986) Social Foundations of Thought and
Action: A Social Cognitive Theory. Englewood Cliffs, NJ:
Prentice-Hall..
Beck, J.G., & Davies, D.K. (1987). Teen Contraception: A
review of perspectives on compliance. Archives of Sexual
Behavior. 4(16), 337-369.
Becker, M.H. (1974). The Health Belief Model and personal
health behavior. Health Education Monographs,2 324-508.
Beckman, L.J., & Harvey, S.M. (1996). Factors effecting the
consistent use of barrier methods of contraception. Obstetrics &
Gynecology, 88(3) (Supplement), 65S-71S.
Brown, L.R. (1991). State of the World: A Worldwatch
Institute Report on Progress Toward a Sustainable Society.
Burger, J.M., & Burns, L. (1988). The illusion of unique
invulnerability and the use of effective contraception.
Personalitv_and Social Psychology Bulletin, 14, 264-270.
Byrne, D. (1983). Sex without contraception. In
Adolescents, Sex and Contraception. Byrne, D. & Fisher, WA,
71


editors. Hillsdale, NJ: Lawrence Earlbaum Associates.
Byrne, D., & Byrne, L.A. (1977). Exploring Human Sexuality.
New York: Harper and Row.
Catania, J.A., Gibson, D.R., Chitwood, D.D., & Coates, T.J.
(1990). Methodological problems in AIDS behavioral research:
Influences on measurement error and participation bias in studies
of sexual behavior. Psychological Bulletin. 108(3), 339-362.
Chilman, C.S.. (1979). Teenage pregnancy: A research
review. Social Work. 24. 492-498.
The Contraception Report, Modern IUDs: Part 1, (1998),
September, Vol 9 (4) 4-8.
Cook, T.D., & Campbell, D.T. (1979). Ouasi-
Experimenta-tion.;.. Design & analysis issues for field settings.
Boston: Houghton Mifflin Company.
Consoli, S.M., Said, M.B., Jean, J., Menard, J., Plouin, P-
F., Chatellier, G. (1995). Benefits of a computer-assisted
education program for hypertensive patients compared with standard
education tools. Patient Education and Counseling, 26, 343-347.
Cramer, J.A. (1996). Compliance with contraceptives and
other treatments. Obstetrics and Gynecology. 88. 4S-12S.
Cvetkovich, G., & Grote, B. (1983). Adolescent development
and teenage fertility. In Byrne D., Fisher, W.A., (Eds.),
Adolescents, Sex, and Contraception. Hillsdale, NJ: Lawrence
Earlbaum Associates.
D'Antionio, W.V. (1994). Human sexuality, contraception and
abortion: Policies, attitudes and practices among major American
religious groups. Paper prepared for the Committee on Unintended
Pregnancy, Institute of Medicine. Washington DC: National Academy
Press.
Dickens, H.O., Mudd, E.H., & Huggins, G.R. (1975).
Teenagers, contraception and pregnancy. Journal of Marriage and
Family Counseling. 1. 175-181.
Elmer-Dewitt, P. (1992). Summit to save the earth: Rich vs.
poor. Time, 139 (22), 42-58.
EPI INFO, v. 6.04b, Centers for Disease Control and
Prevention, Atlanta, GA.
Fishbein, M., & Ajzen, I. (1980). Understanding attitudes
and predicting social behavior. Engelwood Cliffs, NJ: Prentice-
Hall.
Fishbein, M., Bandura, A., Triandis, H.C., Kanfer, F.H.,
Becker, M.H., & Middlestadt, S.E. (1991). Factors influencing
behavior and behavior change: Final Report Theorists Workshop.
72


Washington, D.C.
Fishbein, M., Middlestadt, S.E., & Hitchcock, P.J. (1994).
Using Information to Change Sexually Transmitted Diease-Related
Action: An Analysis Based on the Theory of Reasoned Action.
Preventing AIDS: Theories and Methods of Behavioral Interventions.
Diclemente and Peterson, Eds. Plenum Press, New York.
Fisher, W.A., Byrne, D., & White, L.A. (1983). Emotional
barriers to contraception. Adolescents, Sex, and Contraception.
NJ: Lawrence Erlbaum Associates, 207-239.
Forfeit, J.R., & Forfeit, K.G. (1981). Risk-taking and
contraceptive behavior among unmarried college students. PopuI
Environ., 4, 174-188.
Forfeit, K.G., & Forfeit, J.R. (1978). Correlates of
contraceptive behavior among unmarried U.S. college students.
Stud Fam Plann. 31, 6-40.
Forrest J.D., (1993). Timing of reproductive stages.
American Journal of Obstetric Gynecology, 82, 110.
Goldstein, M.K., Clarke, A.E., Michelson, D., Garber, A.M.,
Bergen, M.R., & Lenert, L.A. (1994). Developing and testing a
multimedia presentation of a health-state description. Medical
Decision Making, 14, 336-344.
Grady, W.R., Hayward, M.D.. Billy J.O.G., & Florey, F.A.
(1989). Contraceptive switching among currently married women in
the United States. Journal of Biosocial Sci SuppI.. 11. 117-132.
Grady, W.R., Hirsch, M.B., Keen, N., & Vaughan, B. (1983).
Contraceptive failure and continuation among married women in the
United States. Studies in Family Planning. 14, 9.
Greep, R.O. (1998). Whither the global population problem.
Biochemical Pharmacology, 55(4). 383-386.
Grossman, R.A., & Grossman, B.D. (1994). How freouentlv is
emergency contraception perscribed?. 26. 270-271.
Haffner D. (1994). Sexuality Issues and Contraceptive Use.
Paper prepared for the Committee on Unintended Pregnancy,
Institute of Medicine. Washington DC: National Academy Press.
Hall, J.A. (1978). Nonverbal sex differences:
Communication accuracy and expressive style. Baltimore and
London: The Johns Hopkins University Press.
Hatcher, R.A., Trussell, J., Stewart, F., Stewart, G.K.,
Kowal, D., Guest, F., Cates, W.. & Policar, M.S. (1994).
Contraceptive Technology (16th ed.). New York: Irvington
Publishers, Inc.
Healton, C.G., & Messeri, P. (1993). The effect of video
73


interventions on improving knowledge and treatment compliance in
the sexually transmitted disease clinic setting: Lesson for HIV
health education. Sexually Transmitted Diseases. 19, 326.
Henshaw, S.K. (1998). Unintended pregnancy in the United
States. Family Planning Perspectives. 30(1), 24-29 & 46.
Henshaw, S.K., & Kost, K. (1996). Abortion patients in
1994-1995: Characterististcs and contraceptive use. Family
Planning Perspectives. 4(28), 140-147.
Herold, E.S., (1981). Contraceptive embarrassment and
contraceptive behavior among young single women. Journal of Youth
and Adolescence,10, 233-242.
Hogarth, R. (1987). Judgment and Choice; The Psychology of
Decision. New York: John Wiley & Sons, Ltd.
Huff, C., & Cooper, J. (1987). Sex bias in educational
software: The effect of designers' stereotypes on the software
they design. Journal of Applied Social Psychology, 17(6), 519-
532.
Institute of Medicine. (1995). The Best Intentions;
Unintended Pregnancy and the Well-Being of Children and Families.
Washington, DC: National Academy Press.
Jaccard, J. (1996). Adolescent contraceptive behavior: The
impact of the provider and the structure of clinic-based programs.
Obstetrics & Gynecology. 88(3) (Supplement), 57S-63S.
Jones, E.E., (1995). Androgenic effects of oral
contraceptives: Implications for patient compliance. American
Journal of Medicine, 98, (Suppl. 1A), 116S-9S.
Kaeser, L. (1989). Reconsidering the age limits of pill
use. Family Planning Perspectives. 21. 273-274.
Kahn, G. (1993). Computer-based patient education: A
progress report. M.D. Computing. 10(2). 93-98.
Kamb, (1998). Efficacy of risk-reduction counseling to
prevent human immunodeficiency virus and sexually transmitted
diseases: A randomized controlled trial. JAMA. 280. 1161-1167..
Kantner, J.F., & Zelnik, M. (1972). Sexual experience of
young unmarried women in the United States. Family Planning
Perspectives, 4, 9-18.
Kaplan, S.H., Greenfield, S., & Ware, J.E. (1989).
Assessing the effects of physician-patient interactions on the
outcomes of chronic disease. Medical Care, 27(3) (Supplement),
S110-S127.
Keller, S., & Sack, A. (1982). Sex guilt and the use of
contraception among unmarried women. Contraception. 25. 387-393.
74


Lapham, S.C., Kring, M.K., & Skipper, B. (1991). Prenatal
behavioral risk screening by computer in a health maintenance
organization-based prenatal care clinic. American Journal of
Obstetrics and Gynecology, 165(3), 506-514.
Levine, S., Ancill, R.J., & Roberts, A.P. (1989).
Assessment of suicide risk by computer-delivered self-rating
questionnaire: preliminary findings. Acta Psvchiatr Scand, 80,
216-220.
Linden, E. (1992). Population: The uninvited guest. Time.
139 (22), 54.
Litwin, How to measure survey reliability and validity.
Llewellyn-Thomas, H.A. (1995). Patients' health-care
decision making: A framework for descriptive and experimental
investigations. Medical Decision Making, 15(2), 101-106.
Lowe, C.S., & Radius, S.M. (1987). Young adults'
contraceptive practices: An investigation of influences.
Adolescence. 22. 291-304.
Luker, K. (1975). Taking Chances: Abortion and the
decision not to contracept. Berkeley, CA: University of
California Press.
Milan, R.J., & Kilmann, P.R. (1987). Interpersonal factors
in premarital conception. Journal of Sex Research.23. 289-321.
Millstein, S.G. (1987). Acceptaility and reliability of
sensitive information collected via computer interview.
Educational and Psychological Measurement. 47, 523-533.
Mosher, W.D. (1998). Design and operation of the 1995
National Survey of Family Growth. Family Planning Perspectives.
30(1), 43-46.
Mudd, E.H., Dickens, H.O., Garcia, C.R., Rickels, K.,
Freeman, E., Huggins, G.R., & Login, J.J. (1978). Adolescent
health services and contraceptive use. American Journal of
Orthopsychiatry. 48, 495-504.
Norusis, M.J. (1990). SPSS base system user's guide.
Chicago, IL: SPSS, Inc.
Parker, R.M., Williams, M.V., Baker, D.W., & Nurss, J.R.
(1996) Literacy and contraception: Exploring the link.
Obstetrics & Gynecology. 88(3) (Supplement), 72S-77S.
Perloff, L.S. (1983). Perceptions of vulnerability to
victimization. Journal of Social Issues, 39(2), 41-61.
Piepert, J.F., & Gutmann, J. (1993). Oral contraceptive
risk assesment: A survey of 247 educated women. Obstetrics and
Gvnecoloav. 82. 112-117.
75


Potter, L.S. (1991). Oral Contraceptive Compliance and Its
Role in the Effectiveness of the Method. In J.A. Cramer & B.
Spiker (Eds.), Patient Compliance in Medical Practice and Clinical
Trials (pp.195-205). New York: Raven Press.
Pratt, W.F., & Bachrach, C.A. (1987). What do women use when
they stop using the pill? Family Planning Perspectives. 19(6),
257-266.
Prochaska, J.O., & DiClemente, C.O. Toward a comprehensive
model of change. In W. R. Miller & N. Heather (Eds.), Treating
Addictive Behaviors Processes of Change (pp. 3-27) NY and London:
Plenum Press.
Rainey, D., Stevens-Simon, C., & Kaplan, D.W. (1993). Self-
perception of infertility among female adolescents. AJDC. 147,
1053-1056.
Redelmeier, D.A., Rozin, P., & Kahneman, D., (1993).
Understanding patient's decisions: Cognitive and emotional
perspectives. JAMA. 270, 1, 72-76.
Reid, T.R. (1998). Feeding the planet. National
Geographic, October, 57-74
Ross, J.A. (1989). Short-term vs. long term failure rates.
Family Planning Perspectives. 21(6). 275-277.
Savonius, H., Pakarinen, P., Sjoberg, L., & Kajanoja, P.
(1995). Reasons for pregnancy termination: Negligence or failure
of contraception? Acta Obstet Gvnecol Scand. 74, 818-821.
Sawyer, R.G., & Pinciaro, P.J. (1998). College students'
knowledge and attitudes about Norplant and Depo Provera. American
Journal of Health Behavior, 22(3). 163-171.
Simon, H. (1959). Models of Man: Social and Rational. New
York, NY: John Wiley and Sons Inc..
Skinner, H.A., & Allen, B.A. (1983). Does the computer make
a difference? Computerized versus face-to-face versus self-report
assessment of alcohol, drug, and tobacco use. Journal of
Consulting and Clinical Psychology. 51(2), 267-275.
Smith, P., Weinman, M., Nenny, S.W. (1984). Desired
pregnancy during adolesence. Psvch Rep 227-231.
SOLO statistical system power analysis. Los Angeles: BMDP
Statistical Software.
SPSS/PC+, v. 4.0. SPSS Inc., Chicago, IL.
Steinlauf, B. (1979). Problem-solving skills, locus of
control, and the contraceptive effectiveness of young women.
Child development.50. 268-271.
76


Stevens-Simon, C., Dolgan, J.I., Kelly, L., & Singer, D.
(1997). The effect of monetary incentives and peer support groups
on repeat adolescent pregnancies: A randomized trial of the
dollar-a-day program. JAMA, 277 (12). 977-982.
Tanfer, K. (1994). Knowledge, attitudes and intentions of
american women regarding the hormonal implant. Family Planning
Perspectives, 26. 60-65.
Tanfer, K., & Horn, M.C. (1985). Contraceptive use,
pregnancy and fertility patterns among single women in their 20s.
Family Planning Perspectives, 17, 10-19.
Thorneycroft, I.H. (1993). Contraception in women older than
forty years of age. Perimenopausal Health Care. 273-275.
Trussell, J., Hatcher, R.A., Cates, W., Stewart, F.H., Kost,
K. (1990). Contraceptive failure in the United States: an update.
Studies in Family Planning, 21.(1). 51-54.
Tsui, A.O., de Silva, S.V., & Marinshaw, R. (1991).
Pregnancy avoidance and coital behavior. Demography. 28, 101-117.
Turner,C.B., & Darity, W.A.. (1980). Black couples and
birth control, knowledge, attitudes, and reported pregnancies.
Quarterly Health Education, 1, 30-34.
Vickery, D.M., & Lynch, W.D. (1995). Demand management:
Enabling patients to use medical care appropriately. JOEM. 37(5),
551-557.
Weinstein, N.D. (1987). Unrealistic optimism about
susceptibility to health problems: Conclusions from a community-
based sample. Journal of Behavioral Medicine, 10, 481-500.
Westhoff, C., Marks, F. & Rosenfield, A. (1993). Residency
training in contraception, sterilization, and abortion.
Obstetrics and Gynecology. 81, 311-314.
Whitley, B.E. Jr., & Hern, A.L. (1991). Perceptions of
vulnerability to pregnancy and the use of effective contraception.
Personality and Social Psychology Bulletin, 104-110.
Whitley, B.E. Jr., & Schofield J.W. (1986). A meta-analysis
of research on adolescent contraceptive use. Population and
Environment Bulletin. 8. 173-203.
Zabin, L.S., Stark, H., & Emerson, M. (1991). Reasons for
delay in contraceptive clinic utilization: Adolescent clinic and
nonclinic populations compared. Journal of Adolescent Health
Care. 12. 225-232.
Zelnik, M., Kantner, J.F, & Ford, K. (1981). Sex and
Pregnancy in Adolescence (Vol. 13). Beverly Hills, CA: Sage.
Zelnik, M., & Kantner, J.F. (1977). Sexual and
77


contraceptive experience of young unmarried women in the United
States. Family Planning Perspectives,?^ 55-61.
78


Full Text

PAGE 1

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 2

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 3

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 4

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 5

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 6

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 7

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 8

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 9

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 10

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 11

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 12

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 13

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 14

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 15

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 16

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 17

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 18

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 19

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 20

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 21

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 22

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 23

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 24

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 25

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 26

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 27

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 28

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 29

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 30

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 31

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 32

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 33

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 34

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 35

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 36

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 37

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 38

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 39

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 40

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 41

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 42

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 43

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 44

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 45

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 46

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 47

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 48

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 49

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 50

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 51

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 52

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 53

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 54

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 55

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 56

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 57

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 58

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 59

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 60

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 61

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 62

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 63

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 64

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 65

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 66

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 67

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 68

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 69

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 70

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 71

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 72

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 73

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 74

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 75

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 76

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 77

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 78

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 79

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 80

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 81

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 82

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 83

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 84

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 85

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 86

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 87

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 88

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 89

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

PAGE 90

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.