INTENTION AS A FUNCTION OF OUTCOME EVALUATIONS AND BELIEFS: INFLUENCE OF ATTITUDES AND SUBJECTIVE NORMS ON BEHAVIORAL INTENTION TO ACQUIRE ORAL CONTRACEPTIVES OVER-THE-COUNTER
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1999
Raj esh Nayak
This manuscript is dedicated to the loving memory of my father, Vishwanath
Nayak, the man who inculcated in me the value of education, discipline and hard work. Some people inspire and motivate others more in their demise than when they are alive, in ways unfathomable to collective human wisdom. My father is one of them. He taught me more through his silence than through his words. I will always be indebted to him for his kindness, understanding, and his tacit approval of every decision I have made in my life. This work would never have seen the light of the day had it not been for his constant encouragement.
To my mother, Jenny Nayak, whose strength, character, and wisdom have made me what I am today. She taught me how to "go on," when faced with adversity, obstacles, and seemingly insurmountable challenges. I feel blessed to draw upon her unending affection and support to this date.
Finally, to my loving wife, Sarita, for standing by me during all the ups and
downs, facing the storm, for being there during the trying hours. I feel lucky to have her by my side, sharing in all the happiness and distress. I feel indebted to her for accompanying me through the dissertation phase, for constantly reminding that I could do it." My pursuit of intellectual capacity would never have been a reality without her support.
My profound gratitude goes to my major advisors, Drs. Carole Kimberlin and David Brushwood, for seeing me through the completion of this work. My association with these two individuals has taught me new things. I thank Dr. Kimberlin for helping me develop clarity in thinking process, and for infusing in me the true spirit of scientific inquiry. I thank Dr. Brushwood for helping me think critically about problems and for helping me realize that things kept short and simple are always better understood. I appreciate their continued support, valuable suggestions, constructive criticism, and above all, the opportunities they have provided to me for further learning.
I would like to thank Dr. Charles Hepler for his guidance in adopting the
appropriate conceptual framework for this study and for driving me to think logically through the research process. I appreciate his patience and invaluable guidance throughout this research endeavor. My heartfelt thanks go to Professor Paul Doering and Dr. David Miller for their advice, interest, and assistance from the beginning of this project. I remain indebted to them for their kindness and for their patience in dealing with frequently changing schedules and last-minute adjustments to help keep me on track, always with a smile on their faces. Meeting those dreaded deadlines would have been impossible without the cooperation of these individuals. I also thank Dr. Richard Segal for lending me his patient ears to hear my concerns as a graduate student and for addressing them in an appropriate manner.
I would also like to thank the staff of Pharmacy Health Care Administration,
especially Delayne, Debbie and Jennifer for all their help with the stuffing of envelopes, mailing and completing the necessary paperwork for this study.
Finally, I thank all my fellow graduate students for making the years I have spent here enjoyable, with special thanks to Peter and Avin forbearing the brunt of my frustration in the racquetball court whenever things did not appear to go well off the court.
TABLE OF CONTENTS
ACKN OW LEDGM ENTS ................................................................................................. iv
ABSTRACT ....................................................................................................................... ix
Problem Statem ent ...................................................................................... I
Background .................................................................................................. 3
Rationale for the Study ............................................................................... 6
Significance of the Study ............................................................................. 9
2 REVIEW OF LITERATURE
Introduction ................................................................................................ I I
Oral Contraceptives-Overview ................................................................. 11
H isto ry .................................................................................................. 1 1
Rx-to-OTC Switch and a Possible OTC Status for Oral
Contraceptives ................................................................................ 14
Pros and Cons of Switching OCs to Over-the-Counter ....................... 15
Contraceptive Beliefs, Attitutdes, and Behavior (Practice) ......................... 22
Attitude Theories Applied to Contraceptive Practices ......................... 27
M ethodological Considerations ........................................................... 29
Influence of Beliefs, Attitudes on Intentions and Use ......................... 35
Summ ary .................................................................................................... 40
3 THEORETICAL FRAMEWORK
Introduction ................................................................................................. 41
Theory of Reasoned Action ...................................................................... 42
Determ inants of Attitudes and Subjective Norm ................................. 44
Description of the M odel Constructs .................................................. 46
Research Questions .................................................................................... 49
Research Hypotheses ................................................................................. 51
Constitutive Definitions of M odel Constructs ........................................... 53
Introduction ................................................................................................ 54
Sampling .................................................................................................... 55
Data Collection Procedures ........................................................................ 57
Study Variables and Operationalization of Constructs .............................. 60
Instrument Developm ent and Validation ................................................... 75
Conclusions ................................................................................................ 89
Sample Selection ........................................................................................ 90
Data Analysis ............................................................................................. 92
Hypotheses Testing .................................................................................... 93
Lim itations ................................................................................................. 98
Summ ary .................................................................................................... 99
Introduction .............................................................................................. 100
Response to Survey .................................................................................. 100
Non-Response Analysis ........................................................................... 102
Characteristics of the Study Sample ........................................................ 103
Intrum mentation .......................................................................................... III
Hypothesis Testing ................................................................................... 114
Summ ary .................................................................................................. 137
6 DISCUSSION, CONCLUSION, RECOMMENDATIONS
Discussion ............................................................................................... 139
Conclusion ................................................................................................ 151
Study Implications and Recommendations for Future Research ............. 153
A STUDY QUESTIONNAIRE ......................................................................... 157
B COVER LETTER FOR PILOT STUDY ....................................................... 163
C M AIN STUDY COVER LETTER ................................................................ 164
D REM INDER POST CARD ............................................................................ 165
E SURVEY CENTER CALLSHEET ............................................................... 166
F COMPARISON OF MAIL AND TELEPHONE SAMPLES ....................... 167
G RESIDUAL PLOTS FROM MULTIPLE REGRESSION ............................ 173
H TEST OF MANOVA ASSUMPTIONS ........................................................ 175
1 MANOVA ASSUMPTIONS- HOMOGENEITY OF VARIANCE ............ 176
LIST O F R EFER EN C E S ................................................................................................. 177
BIOGRAPHICAL SKETCH ........................................................................................... 185
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
INTENTION AS A FUNCTION OF OUTCOME EVALUATIONS AND BELIEFS: INFLUENCE OF ATTITUDES AND SUBJECTIVE NORMS ON BEHAVIORAL INTENTION TO ACQUIRE ORAL CONTRACEPTIVES OVER-THE-COUNTER By
Raj esh Nayak
Chair: Professor Carole Kimberlin
Co-Chair: Professor David Brushwood
Major Department: Pharmacy Health Care Administration
The study examined the utility of the theory of reasoned action (TRA) for
explaining university women's intention to acquire oral contraceptives (OCs) over-thecounter (OTC). Results indicated that a woman's intention to acquire oral contraceptives without a prescription was mainly determined by her beliefs concerning the possible risks and benefits of using OCs and her perception of what her key referents (family, friends, and partner) would think about her acquiring OCs without a prescription. Physiological side effects and the pill's perceived benefits were the major considerations that entered into women's decisions to acquire or not to acquire OCs over-the-counter. All women in the sample (N=294) believed that using OCs would lead to negative outcomes (such as nausea, weight gain). There was no overall difference between women who intended to acquire OCs over-the-counter and those who did not in the way they evaluated the risks
and benefits of using OCs. Instead, the difference between the two groups stemmed from their perceived likelihood of the possible outcomes.
The amount of variation in intention explained by the indirect measures of attitudes and subjective norm was only 7 percent, whereas about 50 percent of the variation in intention was explained by the use of direct measures of these constructs. Only 37 percent of the respondents surveyed reported intention to acquire OCs over-thecounter. A majority of the women in the study expressed a favorable attitude toward the beneficial aspects of OTC oral contraceptives and negative attitudes toward the risky consequences arising from their use. A very low correlation was found (0.075, p=O. 199) between the perception of risk and intention scores, whereas a positive and significant correlation (0. 188, p=0.001) was found between perception of benefits and intention to acquire OCs OTC.
The study provided only a moderate support for the TRA. The findings of the study support the following general conclusions. First, women who express more favorable attitudes toward acquiring OCs OTC report stronger intentions to acquire OCs over-the-counter. Second, women's attitudes are a stronger determinant of intention to acquire OCs over-the-counter than are their perceptions of normative support among key referents. Third, women's attitudes toward the beneficial consequences of using OCs OTC are a better predictor of behavioral intention than are their attitudes toward risky consequences. Fourth, TRA seems to work best when attitudes and subjective norms are measured directly than when they are derived multiplicatively.
As oral contraceptives (OCs) enter their 3 8th year of prescription status, the question of whether women should be able to obtain them without a prescription has come to the forefront of issues related to women's reproductive health care. Currently 18 million American women are known to use birth control pills on a regular basis, with 80% of them expected to use OCs during their lives (Dawson, 1990). According to a survey conducted by Ortho Pharmaceuticals (1993), OCs are rated as the most preferred birth control method by women and they remain the most popular method in the United States. Consequently, the way in which OCs are made available to the general public has been a matter of interest to the policy makers, researchers, and educators. Health policy analysts have suggested a change in the existing system of prescription-only availability for oral contraceptives in the United States. They have suggested switching oral contraceptives to a nonprescription status (Grimes, 1993; Trussell et al., 1993; Potts and Denny, 1995; The Tan Sheet, March 29, 1993).
The available evidence indicates that women ate divided in their preference for obtaining birth control pills without a prescription. While some women are reported to favor close medical supervision over the acquisition of OCs, others are believed to be in support of a system that provides no mandatory medical supervision over the sale and use
of OCs (Nayak et al., 1994). Conflicting evidence in this regard, gathered mainly through national opinion polls (Gallup Poll, 1985, 1994; Taylor et al., 1994), provides no insight into why women may have varying preferences for obtaining birth control pills. At the core of the discussion concerning the availability of oral contraceptives without mandatory medical authorization are concerns related to safety and efficacy, dangers of side effects, economic costs, and perceived social barriers. Social interests such as patient autonomy, self-care, reduction in the incidence of teenage pregnancies, and prevention of promiscuity may also be viewed as significant to justify the prescription status for OCs. For example, some women feel that making OCs available without a prescription might lead to more women, especially adolescents, having sexual encounters, while others feel that such a step might even reduce teenage pregnancies in the US (Nayak et al., 1994).
Current evidence indicates that women have differing beliefs about the means of acquiring OCs (Nayak et al., 1994). Some beliefs about the biomedical aspects of using OCs relate to their direct effects on the body, such as experiencing side effects, or incidence of certain types of cancer. In addition, some women, when asked about an option of possibly acquiring pills over-the-counter (OTC), seem to hold beliefs that are somewhat different than when they are told of no such option. For example, some women believe that obtaining OCs without a prescription might lead to undesirable health problems (Nayak et al., 1994). OTC status might mean foregoing some routine physical examination, risking imperfect use, or not being able to take advantage of the professional oversight of their use (Nayak et al., 1994; Grimes, 1993; Trussell et al., 1993). These beliefs are considered indirect because they do not relate to the beliefs about the physical effects of OCs on the body. Instead, they relate to the psychosocial
consequences of acquiring them through a mode that is not considered traditional. Also viewed to be significant in the acquisition of OCs with or without professional supervision are women's beliefs about the cost and convenience of obtaining them from two different sources.
The social psychology literature holds that beliefs usually translate into
corresponding attitudes. Specifically, in the current context, women's beliefs regarding the means of acquiring OCs from different sources would be expected to shape their attitudes toward the same. This study, therefore, proposes to systematically investigate these attitudes and thereby explain any preference for a system of acquiring OCs in a certain way. In addition, the determinants of women's preferences for the current or the proposed new status for oral contraceptives are currently unknown. Therefore, this study seeks to examine the factors that relate to women's preferences for acquiring OCs from different sources.
Available evidence indicates that women are divided in their opinions about
obtaining oral contraceptives and contraceptive services from different sources (Gallup Poll, 1985, 1994; Taylor et al., 1994; Nayak et al., 1994). While some women are known to favor medical supervision over the acquisition of pills, others are not opposed to a system that provides them with greater control, rights and responsibilities. For example, our preliminary research in this area suggests this duality of preferences (Nayak et al., 1994). While some women were reported to prefer the status quo of medical authority, others preferred to obtain OCs under their own authority. The question of why women have differing preferences for the acquisition of OCs is unanswered at this time.
Preliminary findings suggest that women's preferences for acquiring OCs differ mainly due to the differing beliefs they hold. While some beliefs corresponded to their preference for a submission to medical supervision, others indicate a preference for autonomy.
Besides the beliefs referred to above by the women interviewed, the beliefs that are generally held by others are also viewed to be significant to OC acquisition in the US. A misconception about the use of OCs and their possible side effects seems to continue among the OC users and nonusers alike (Gallup Poll, 1985, 1994). Interestingly, according to these polls, beliefs about the negative aspects of OC use, such as the side effects of OCs, were found to be more prevalent among OC users than nonusers. The known side effects of the drug include pain in the legs, vision defects, breakthrough bleeding, weight gain, and sometimes even risks of stroke (Grimes, 1992). Despite their convenience, effectiveness, and reversibility of action, oral contraceptives in recent years have also sparked some debate about their link with cancer risk. One of the major risks of using pills, either real or commonly believed to be true, is the susceptibility of an individual to various types of ovarian and endometrial cancers, as well as breast cancer. OC users in particular, as the polls above indicated, seemed to hold more negative beliefs about the risks of cancer than did nonusers. The preliminary investigation also supported similar findings (Nayak et al., 1994). Conflicting scientific evidence regarding the incidence of cancer among OC users, especially those of ovaries and breasts, continue to be reported to this date. No comprehensive insight, however, is available today as to whether women's perceptions and attitudes regarding cancer risks are shaped by
scientific reports and/or personal experiences. The issue of whether or not OCs really cause certain types of cancer is still unresolved at this time.
Women's perceptions regarding the acquisition of OCs are also shaped by their beliefs about the economic costs of obtaining them. Some women fear that insurance policies that cover prescription drugs may no longer pay for OCs or OC-related services (Nayak et al., 1994). Yet, from the standpoint of costs, it might be less expensive to obtain OCs when they are sold through unrestricted channels because women may no longer have to pay large physician fees (Nayak et al., 1994). According to some health policy analysts, it might mean that nonprescription status for OCs might also threaten the survival of those nonprofit family planning clinics that rely heavily on the reimbursement associated with contraceptive distribution (Trussell et al., 1993). From the societal perspective, it is the general belief that unrestricted availability of OCs might actually lead to unprotected sex among those who use pills, especially the teenagers (Nayak et al., 1994). Preliminary findings also suggested that women who held positive beliefs about acquiring OCs in a particular way seemed to hold positive attitudes about acquiring them through that way. For example, women who believed that it was less expensive to acquire OCs without a prescription thought it was less risky to use them if acquired that way, and were more predisposed to acquiring them without a prescription. In contrast, those who harbored unfavorable beliefs seemed to hold unfavorable attitudes toward the means of acquiring them a particular way. In addition, some women who thought it was safe to obtain OCs without a prescription were willing to acquire them if the current situation made it possible to obtain them with no prescription. Thus, it seems that women's beliefs and attitudes about acquiring pills are sometimes at odds with those of
the society in general and policy makers. While some beliefs and attitudes are shaped by personal experiences, others are dictated by situational constraints.
It can be argued, based on the foregoing discussion, that a woman's beliefs about benefits of obtaining OCs in a particular way, her beliefs about the health risks involved, her beliefs about the cost of alternatives, and her beliefs about possible social consequences can be expected to correlate with her general attitude toward the act of acquiring them a certain way. While the current evidence points to the possible influence of some, if not all, of these belief components, it is not known which of these components are really important in choices a woman might make regarding obtaining OCs from alternate sources. Consequently, it is the purpose of this research to investigate the role of underlying factors that may shape and influence a woman's attitude toward obtaining oral contraceptives in a particular way.
Rationale for the Study
Recently, a Harris Poll sponsored by a California foundation (Taylor et al., 1994) indicated support for the OTC pills among young women surveyed, whereas women surveyed by Gallup Organization (1994) reported a clear lack of support for an OTC pill. A poll commissioned by American College of Obstetricians and Gynecologists, conducted by Gallup Organization (March 1, 1985), points to some important discrepancies between available scientific evidence and public understanding of oral contraceptives. A similar poll conducted by Gallup 9 years later (January, 1994) reported somewhat similar findings, however, with a substantial shift in women's attitude toward OTC oral contraceptives. According to this survey, which was conducted through
telephone interviews of 997 women 18 and over, 86 percent overall and 91 percent of women on birth control pills didn't believe that OCs were safe enough to buy without first seeing a doctor. However, there was a substantial decrease in the percentage of women (compared to the 1985 poll), down to 54 percent from 76 percent, who felt there were substantial risks associated with the use of OCs. Furthermore, 29 percent of women cited cancer as the chief risk, as did 3 1 % in 1985, followed by birth defects, heart attacks and blood clots. Interestingly, while 41 percent of women polled did not believe in health benefits of OCs other than the prevention of pregnancy itself, only 6 percent were even aware of OCs' protection against certain types of cancer. The percentage of women who thought, rather unfortunately, that using OCs was more risky or as risky as childbirth stayed almost constant at 65 percent. However, the message that emerged clearly from this report was that, during the passage of time, fewer women actually thought oral contraceptives carried substantial health risks. The findings also reflect misconceptions about oral contraceptives among women, and their lack of understanding of how OCs work. In addition, when the Harris Poll (Taylor et al., 1994) asked women "how likely is it that they would buy the pill without a prescription if insurance or government subsidy did not cover it" women more frequently said "yes" to the question than in the earlier poll. The percentage of women responding to this question in positive was greater for pill users than for nonpill users. In the same survey, more women thought that they would buy OCs over-the-counter even if they were more expensive as long as OCs did not require a prescription. While this question did not relate to the safety aspects of oral contraceptives, it did indicate increased confidence on the part of women to use OCs if they were in fact available over-the-counter.
Notwithstanding the conflicting findings and methodological inadequacies of these surveys, the emerging mixed attitude among women about obtaining oral contraceptives from different sources is an issue for further investigation. Understanding how women make decisions about acquiring oral contraceptives is important for designing programs and health services to assist in preventing unwanted pregnancies and family planning. Changes in the status quo should take into account the views of women regarding the change, including their preference for and any perceived need for that change. The promotion of more effective means of providing OCs is an important public health issue. Thus it is important to study women's attitudes toward various means of access to oral contraceptives.
There is little research done today in areas of consumer perceptions of the need for and the safety of nonprescription products. Therefore, there is a need not only to assess any consumer preference for unrestricted access to nonprescription OCs, but also to identify and research factors that might actually deten-nine any preference they may have for different sources of pills available to them. In the current socio-political atmosphere, which recognizes the need to strengthen and promote research related to women's reproductive health care--a component of which also involves provision and acquisition of OCs--there exists a knowledge gap about issues of access and barriers to oral contraception. Therefore, it is the purpose of this study to address questions related to the preference for the prescription requirement for oral contraceptives and any preference for an alternative system in which to obtain OCs, and to identify factors relating to their stated preference. Broadly, the objective of the study is to seek answers to the following inquiries:
1. Given a new means of acquiring OCs, what factors differentiate women who express a preference for a system in which OCs could be acquired over-the-counter, from other women who do not express a preference for such a system?
2. What are the factors that underlie a woman's stated intention to acquire OCs over-thecounter? In what ways are the expressed beliefs different for different women regarding their choice of acquiring OCs over-the-counter?
3. Do women's beliefs about the acquisition of OCs correlate with attitudes about acquiring them? That is, are beliefs about the positive (negative) aspects of OC acquisition in a particular way correlated with the corresponding favorable (unfavorable) attitudes?
Significance of the Study
One of the important issues in social psychology research relates to whether
beliefs about objects, person, institutions, or events explain (or correlate with) a person's attitudes toward the same. Traditionally, this line of research has also investigated the role of attitude in influencing the performance of ultimate behavior. In the current context of OC acquisition, however, the relationship between beliefs and attitudes regarding the acquisition of OCs OTC is not clearly understood. It is not known whether beliefs held by women about the acquisition of oral contraceptives help shape their attitudes toward acquiring them. Would the attitudes toward acquiring OCs--favorable or unfavorable--explain any preference women might have for acquiring oral contraceptives in a particular way? Would the beliefs held by women regarding the use and the choice of means for acquiring OCs relate to their intention to acquire them in a particular way? Are behavioral beliefs predictive of behavioral intentions? The current investigation
seeks to address these questions. Are women's intentions to acquire OCs with or without a prescription predicted by their beliefs and attitudes? This study aims to investigate this issue further and explain any preference for acquiring OCs as a function of intention to acquire them at a future time. The study's important contribution lies in the fact that it provides an improved understanding of the relationship between OC beliefs, attitudes, intentions, and preferences. Such a relationship has not been explored in depth in the literature pertaining to reproductive health.
Besides theoretical ramifications, the study also has many practical implications. The ways in which contraceptives are obtained and used in this country are of immense interest to researchers, clinicians, and policy makers. Any policy directed at improving women's reproductive health care has far-reaching social, economic, and psychological consequences. For many years in the United States, oral contraceptives have been made available to women only under a physician's authority, whereas their counterparts in other countries have acquired them without any restriction. The socio-economic conditions and people's religious, political, and moral preferences, however, distinguish people in the US from those in other countries. Policy makers, in deciding whether or not to switch oral contraceptives from prescription to non-prescription status, will do well to examine women's beliefs and preferences regarding any changes. The significance of the study lies in the fact that the proposed switch of birth control pills to over-the-counter category not only seeks to improve women's reproductive health care, but also has implications for a rampant social problem in the US--teenage pregnancy.
REVIEW OF LITERATURE
This chapter is divided into two major sections. The first section is a brief introduction to the history behind the development of oral contraceptives and public perceptions regarding their use. Issues about safety, efficacy, and costs of using OCs when they are used with or without a physician's prescription are discussed in this section. This section also deals with issues related to Rx-to-OTC switches of oral contraceptives, including the social and economic ramifications of such a switch. Arguments in favor of or against switching birth control pills to an over-the-counter category are also reviewed in this section. The second section is devoted largely to the attitudes of women towards contraception in general and oral contraception in particular. The attitude theories relating to the use of contraceptives, including OCs, will be the major focus of this section. A brief introduction to the theory of reasoned action, a theoretical framework used for this study, will also be provided in this section.
The introduction of oral contraceptives in 1960 was a major medical achievement that changed approaches to family planning all over the world. Dr. Gregory Pincus and I I
his colleagues developed the oral contraceptive pill in 1960 for Planned Parenthood Federation. The OCs initially contained estrogen and progestin synthetic hormones, and worked primarily by suppressing the release of eggs from a woman's ovaries. These pills contained as much as 100 to 150 micrograms of estrogen and as much as 10 milligrams of progestin, a level high enough to cause some unwanted side effects. Since that time the level of estrogens in oral contraceptives has dropped steadily to a point where they are now considered by many to be safe enough to be moved over-the-counter (OTC). Although the oral contraceptive pill was widely accepted, concerns about its serious side effects started arising in the early '60s. Women from different countries reported blood clotting and stroke (Snider, 1990). Following ongoing research on side effects of OCs, it was concluded that high levels of estrogen were indeed linked to increased incidence of blood clots and heart attack. Subsequently, low-dose estrogen oral contraceptive pills were introduced into the market in the US. Contemporary oral contraceptives are considerably safer as they contain less estrogen and progestin. The risks of side effects are believed to have decreased for healthy, non-smoking women (Grimes, 1992).
One of the major uncertainties about OCs is their potential to cause breast,
cervical, ovarian, and endometrial cancer (Grimes, 1992; Gross et al., 1992; Thomas, 1991). While there are conflicting results among studies, a comprehensive review of oral contraceptive use and its possible link to different types of cancer has been published by the National Cancer Institute (Fact Sheet, June, 1985). Some experts contend that all the data available today reflect the effects of older, higher-dose pills and not the newer lowdose pills currently available in the market. In fact, birth control pills are even believed to have some beneficial health effects (Grimes, 1992; Gross et al., 1992). Numerous
studies show that OCs protect women from incidence of some ovarian and endometrial cancers, benign cysts of the ovaries and breasts, and pelvic inflammatory diseases (Grimes, 1992; Gross et al., 1992; Thomas, 1991). They are also believed to prevent heavy and irregular menstrual periods and a certain type of anemia (Grimes, 1992). Some would even argue that the risk of getting cancer from OCs is so small that benefits of taking them far outweigh the risks associated with their use (Trussell et al., 1993; Peterson and Lee, 1989). A few studies are currently under way to detect a link between cancer and OC use to see if such a link exists.
In any case, the oral contraceptives are not deemed safe for all women. Certain subgroups of women are at risk of serious illness and death if they use oral contraceptives. For instance, women who smoke, particularly those who are over 35, are at significantly increased risk of heart attack and stroke (Snider, 1990; Grimes, 1992). Women who are obese or having health problems such as diabetes, hypertension, or high cholesterol are often advised against using OCs. In addition, OCs are contraindicated in women with history of clots, stroke, heart disease, liver disease, or cancer of the breast or sex organs. Over the years, there have been more studies conducted on OCs to look for their serious side effects than any other medicine in history. However, misconceptions regarding the known and unknown effects of OCs continue mainly because of conflicting results from available studies. In one review, Peterson and Lee (1989) argue that the public misconceptions and controversies about OCs are mainly due to the failure in the efforts of health educators, health care providers, and the lay press. This assertion by the authors has been untested to date. They contend that legitimate disagreement among investigators regarding interpretation of data has also contributed to the confusion. This
assertion is particularly important in view of the current development in oral contraception research, which is advocating a "new" status for oral contraceptives. There is some further evidence of such misconception about the use of oral contraceptives in reproductive literature (Stenson, 1996).
In summary, it is not known whether the misconceptions referred to above contribute to the perceptions of risk among OC users. It is also not known if such perceptions change if OCs are bought and used under the circumstances that might actually alter the level of risks associated with their use. For example, the misconception that OCs cause cancer might mean lack of preference to acquire them with no medical supervision. Perceptions of risks of using OCs, either as a result of personal experience with them or as a result of misconceptions held about their use, are currently not well understood. The literature describing this issue is sparse, further emphasizing the need to conduct more research in this area.
Rx-to-OTC Switch and a Possible OTC Status for Oral Contraceptives
The FDA's OTC Review provided a major impetus for switching drugs from prescription to OTC in 1972. Following the Durham-Humphrey Amendment to the Federal Food, Drug, and Cosmetic Act of 1951 (Ch 578,65 Stat.648), prescription drugs were defined primarily as those unsafe for use except under professional supervision. Some of the major considerations in deciding whether or not a drug should be available without a prescription were its toxicity, its safety, and whether conditions for taking the drug are self-diagnosable conditions (Hayes, 1990; Rheinstein, 1985; Botstein, 1990; Wion, 1985). In reviewing switching, certain changes to the product labels and inserts
were made in order for the patient to understand the information clearly and easily (Hayes, 1990; Rheinstein, 1985; Botstein, 1990; Wion, 1985). Nonprescription drugs were thus deemed safe as long as consumers were able to follow directions and warnings on the label. The OTC-Review in 1972 made it possible to switch prescription drugs to nonprescription status and over 400 previous prescription-only ingredients have gone over-the-counter since then. The Nonprescription Drug Manufacturers Association estimates that more than 200 OTC drug products on the market today were available by prescription only a decade ago (Segal, 1991). Nine out of the 10 top selling drugs today are OTC products that were prescription-only not long ago. The FDA has reclassified antihistamines, nasal decongestants, sleep aids, pain relievers, cough medicines and antifungals among other therapeutic classes. It has been projected that OTC drug market would reach $28 billion in manufacturer sales by the year 2010. (Laskoski, 1992).
Pros and Cons of Switching OCs to Over-The-Counter
The possible switch of OCs to the non-legend category was suggested recently, amidst rising controversy, only to be postponed for later consideration (The Tan Sheet, March 29, 1993). The controversy over the matter related primarily to the safety of using OCs without medical supervision. The safety concerns mainly related to the possible adverse effects of using oral contraceptives without medical authority, and the possible risks of incidence of cancer. As pointed out earlier, the results from the national polls conducted during this and the last decade (Gallup Poll, 1985, 1994; Taylor et al., 1994) seem to support the controversy. In addition, our preliminary exploratory, qualitative study also supported the concern voiced by women during the previous polls. This study
(Nayak et al., 1994) involved 6 women attending the University of Florida Student Health Care Center, who expressed interest in learning about OTC oral contraceptives. Information from the participating subjects about their perceptions of safety of OCs was elicited through a face-to-face interview. Information about their preferences for acquiring oral contraceptives over-the-counter was also gathered. It was found that women were divided in their preferences for obtaining oral contraceptives from alternative sources. While some women preferred a more traditional, prescription-only method, some women indicated preference for an OTC oral contraceptive pill. The women, who indicated a preference for acquiring OCs under a physician's authority often cited safety concerns, particularly drug side effects. However, those who favored OTC oral contraceptives cited cost and convenience as being important considerations in their choice for acquiring OCs.
In most countries, including the USA, women are required to obtain a physician's prescription before they can use oral contraceptives, and they can usually buy them only from local pharmacies (see Paxman, 1980, Coeytaux and Allina, 1994, for review). A single argument in favor of deregulation of oral contraception has been that the prescription requirement for OCs made them geographically, economically, and sometimes culturally inaccessible to many women. In the United Kingdom, for example, it was argued that midwives, health care professionals, and even social workers could oversee the distribution of OCs (Anonymous, 1974). However, in the UK context, the possibility of making OCs available OTC was not entirely ruled out as long as it could be established that the currently used OCs containing less than 50 micrograms of estrogen were significantly safer than and as effective as the 50 microgram pill. In the USA, the
legacy of the Comstock Act of 1873, which made it a criminal offence to import, mail, or transport in interstate commerce any literature about birth control or any device designed to prevent conception or cause abortion, continued to medicalize the status of OCs.
Following the proposal for a change in New Zealand (Campbell et al., 1996) and the UK, and changes already in effect in countries such as India, South Africa, and some South American countries (see Coeytaux and Allina, 1994), similar proposals to give OTC status to OCs have recently gained momentum in the United States (The Tan Sheet, March 29, 1993). It appears that the concerns of the manufacturers and the prescribers in the US are very different. While manufacturers are concerned about the possible product liability lawsuits, physicians are skeptical about patient compliance with OTC pills, not to mention the foregone income from patient visits (Conkling, 1993, Goldstuck et al., 1987; Oakley et al., 1991). Increased profit motives on the part of family planning clinics may also be contributing largely to the current prescription status for OCs (Trussell et al., 1993). Thus, there seems to be a complete lack of consensus among health care professionals and others regarding the appropriate status for pills. Recently, a major proponent of the switch has even gone on record by reversing his position on the issue. (See editorial: Grimes, 1995). However, the above findings do suggest that while the positions of researchers and policy makers regarding the status for OCs are known (as researchers do not all have a similar position in this issue), we do not yet completely know the opinions of the ultimate users of OCs.
The evidence reviewed above suggests that women's beliefs about the purported status for OCs hinge mainly on their perceptions of risk associated with their use. A woman's perception of risk in this case may be broadly classified as one of two types.
The first type of risk is an inherent risk of minor and serious side effects of the drug itself, regardless of the type/level of supervision--regarded as direct effects of the drug. The second type of risk--termed indirect risk--relates to the therapeutic failure as a result of inadequate or lack of professional oversight and general health risks associated with foregoing certain preventive health services.
The first type of risk, the risk of side effects, stroke, and cancer, has been already discussed in detail. The second type of perceived risk is often the result of imperfect use of OCs without proper clinical counseling. This type of risk relates to the indirect effects of using OCs acquired possibly from alternative sources. Examples of indirect effects include possible noncompliance with oral contraceptives (an efficacy issue) and loss of health screening. In a review article, Trussell and his coauthors (1993) argue that safety considerations, either from the biomedical aspects of OC use or from the standpoint of health risks accruing from foregoing preventive health services, do not justify prescription status for oral contraceptives. As for the efficacy of OCs, they contend that efficacy among perfect users will not be affected by prescription status. The level of noncompliance with OCs, often cited as an argument against switching OCs to the nonprescription status, is also believed to vary depending on the way they are acquired. These researchers argue that the type and extent of imperfect use determine the compliance with OCs during typical use. Imperfect use, they contend, may include missing pills and failure to use a backup method of contraception if pills are missed, if antibiotics or anticonvulsants are taken, or if vomiting or severe diarrhea occurs. They believe that the missed pill noncompliance, although a serious and a common problem, and the subsequent risk of pregnancy may be effectively minimized by proper counseling
and monitoring provided routinely by the health personnel. Trussell and his colleagues (1993) also cite instances of missed pill noncompliance even among women who visit clinics regularly to obtain birth control pills. They assert that improper use of OCs is widespread despite the current prescription requirement (see Oakley, 1994; Oakley et al., 1991; Potter, 1991).
An indirect component of this risk, the general health risk, stems from women's foregoing certain preventive health services and screening for cancer. The status quo requires that women submit themselves to certain preventive health services like pelvic examination and pap smears to screen for cancer, even though the health conditions they are screened for are unrelated to decisions about oral contraception. A part of this attendant service is also the testing for sexually transmitted disease (STI)s). The idea that OCs may be available through other unregulated channels may mean foregoing such tests and exposing oneself to higher levels of risk a risk to oneself and a risk to the society as well. Some experts view contraceptive prescription as being held hostage to such unrelated preventive services and argue in favor of conducting such examination separately from the prescription for oral contraceptives (Grimes, 1993; Trussell et al., 1993; Waldron 1990). In light of the current opinion that such examinations should either be deferred or held independent of the oral contraceptive prescription, the FDA has moved recently to recommend physical examination to be delayed before a physician prescribes oral contraceptives (The Blue Sheet, May 26, 1993).
Arguments in favor of OTC status, for drugs in general, also include cost savings, convenience, choice of providers, and the availability of free information and advice. Typically, drugs cost less as nonprescription than prescription medicines, although
initially the nonprescription cost may be higher than the prescription price (Schondelmeyer and Johnson, 1994; Trussell et al., 1993). Temin (1992) studied the costs and benefits of switching cough-and-cold medicines in the United States and found that visits to doctors for common colds fell by 110,000 per year (from 4.4 million) from 1976 to 1989. This trend mainly coincided with the switching of medicines. Among the factors that he considered and rejected as possible explanations were decreased number of potential patients, a general decrease in visits to physicians, and a relative increase in the cost of visits to physicians. Having ruled out other possibilities, he concluded that the decrease in physician visits was attributable to the switching of these drugs (1983). His estimate of the overall saving to the population was $70 million per year.
Regarding the costs associated with the acquisition of OCs, a major drawback of the prescription requirement for OCs is the cost of obtaining them. Included in the overall cost are the price of the drug, cost of physician visits, cost of time lost from work, and the cost of prescription refills that require repeated consultation with the physician. Assuming that low-dose OCs are safe enough to buy OTC, the issues of self-diagnosis and dosing adjustments to meet patient needs are currently being debated widely in the health care community. However, making OCs available OTC is believed to present some obstacles for certain women. Some fear that poor women who obtain OCs inexpensively, say, from family planning clinics, may now pay higher prices because of the possible discontinuation of price discounts by companies to these clinics. While some women depend on Planned Parenthood and family planning clinics to get their OCs free of cost, some worry that insurance policies that covered prescription drugs may no longer pay for OCs (Trussell et al., 1993; Nayak et al., 1994). Thus tied to the cost of the
OC is lack of insurance coverage. Women who are covered under the Medicaid prescription drug program are likely to favor the status quo as it substantially reduces outof-pocket cost to the recipient. There is very limited information currently available describing the cost-related issues involved in the acquisition of oral contraceptives in the US. The available evidence, however, indicates substantial cost savings for most women if they were to acquire OCs with no prescription from a physician.
As noted by Trussell et al. (1993), above all the arguments, the strongest one in favor of a nonprescription status for OCs is its purported role in minimizing unplanned and unwanted pregnancies, especially among teenagers. Increased adolescent sexual activity, mainly in the past decade, has resulted in a sharp increase in the rate of pregnancy, childbirth, abortion, and even increased incidence of STIs in the US. The Gallup Poll (1994) reported that 84% of the women polled were not aware that OCs posed fewer risks than childbirth in women aged 55 and under. Removing barriers to the availability of OCs, therefore, is expected not only to dispel the misconception about their use, but actually help prevent unplanned and unwanted pregnancies. Also a lesser known fact about OCs is their effective use as post-coital contraceptives (Trussell et al., 1992; Owen, 1993). Recently, the FDA has endorsed the use of OCs as postcoital contraceptives (The Pink Sheet, March 3, 1997; The Green Sheet, July 8, 1996). Experts believe that only the increased availability of pills through sources other than a physician will remove any barrier to their widespread use (Trussell et al., 1993). A review by a team of experts contends that if postcoital hormonal treatment were widely available, the number of unintended pregnancy in the United States could fall by 1.7 million each year, and the number of abortions could be reduced by 800,000 annually (Trussell et al., 1992).
Although few women and even doctors know of this option, the possible sale of OCs through unrestricted channels will no doubt be of benefit to those women who wish to use them as moming-after pills, in addition to those who use them for regular contraception.
As a solution to the problems discussed in the preceding section, the proponents of change have made some suggestions of change to the current system. Recommendations include labeling changes, revision of patient package inserts (PPI), and provision of new options that include, among other things, sale of OCs through pharmacies, telephone authorization from nurses, installation of toll-free helplines for consultation, and acquisition of OCs by answering self-administered questionnaires that screen women for contraindications (Trussell et al, 1993).
Contraceptive Beliefs, Attitudes, and Behavior (practice)
The literature is replete with research that shows the effects of demographic
variables, such as socio-economic status, education, religion, age, race, etc. on effective contraceptive use. Typically, women of higher status and better education are believed to be effective contractors, even if there exists no trend data to examine the effects of these variables over a period of time. While a few studies have investigated the influence of access on contraceptive practice (Hanna, 1994; Eisen et al., 1985), there is a lack of information about the role of access to contraception in determining the choices women make about their birth control options. Unlike in other countries where access to contraception has always been a major problem, in the United States, the issues of access have been discussed only in relation to hormonal contraceptives and devices. Therefore,
it is not known if a woman's access to contraception, such as family planning services, will actually determine her preference for a particular method of contraception. Evidence also suggests that despite the widespread availability of contraception, a large proportion of sexually active adolescents have failed to use contraception effectively (Durant et al., 1984; Scher et al., 1982). It is also not known if any barrier to access to contraception or related services might be responsible for such a failure. Due to the limited predictive ability of a contextual variable like access, the research on contraceptive attitudes and beliefs appears to have focused instead largely on numerous aspects of sexuality, pregnancy, and family structure. In order to explain contraceptive practice, recently researchers have focused their attention increasingly to psychological variables like sense of control, self-esteem, and self-efficacy. In addition, the researchers, recognizing reproductive behavior and contraception use as being a joint enterprise, have begun to explore interpersonal interactions, such as those between man and woman, and physician and patient (Weisman et al., 1991; Oakley, 1994; Keller, 1996.). Interestingly, some major studies carried out in the United States, each employing the now familiar KAP (knowledge, attitude, practice) model, have failed to establish any significant relationship between psychological variables and contraceptive practice.
The majority of the studies reviewed here has concentrated on adherence to
contraceptive regimen and compliance with contraceptive therapy among sexually active female adolescents. These studies indicate that the level of adherence in female adolescents was largely influenced by their contraceptive perceptions. The findings relating to attitude regarding contraception have generally been consistent. Most of the studies, however, are based on college student samples. It is possible that selection
variables related to university environment result in sample biases that could substantially affect research results. For the same reason, the generalizability of these studies may also be limited. The research on women's perceptions about oral contraception in particular has focused largely on benefits and detriments of using OCs (Weisman et al., 1991; Wernier and Middlestadt, 1979; Herold and Goodwin, 1980). However, only a small number of such studies that incorporate psychological variables within their framework have been published recently (Durant et al., 1984).
A few studies have also examined the role of locus of control in contraceptive
behavior (Herold et al., 1979; Visher, 1986; Blignault and Brown, 1979). Several studies describing women's attitudes about the benefits and barriers of contraception in general have also been discussed in the literature. Among attitudes expressed, besides clear contraceptive benefits such as preventing pregnancy (Eisen et al., 1985; Kalmuss et al., 1987; Scott el al. 1988), and demonstration of responsibility (Eisen et al., 1985; Kalmuss et al., 1987), having control over one's own life (Kalmuss et al., 1987) appears to be an important variable determining contraceptive attitude. In addition to benefits, women are also reported to perceive barriers to contraception. Some personal barriers include negative perceptions about possible harmful health effects, (Eisen et al., 1985; Kalmuss et al., 1987; Luker, 1975; Scott et al., 1988), perception that they are difficult to obtain and use (Eisen et al., 1985; Luker, 1975), that they involve planning for sexual activity (Eisen et al., 1985; Luker, 1975), that they are against religious beliefs (Eisen et al., 1985), that they may be expensive (Luker, 1975), are embarrassing (Eisen et al., 1985), and are ineffective (Kalmuss et al., 1987). There are other perceived barriers to contraception which included parental disapproval and partner factors (Luker, 1975).
While some of these studies examined the role of contraception in general, a few also looked at the barriers in the use of oral contraceptives. While it can be argued that some of the barriers and benefits related to different forms of contraception apply also to oral contraception, no conclusive evidence is available to support this argument at this time. More research is warranted to identify such perceived benefits and barriers among oral contraceptive users.
Kathleen (1994) reported an oral contraceptive study that sought to somewhat address the knowledge gap mentioned above. This study described 12 female adolescents' oral contraceptive perceptions that were considered independently and then after facilitation by a nurse. Oral contraceptive perceptions were measured through the use of a decision balance sheet and a structured interview, each eliciting 44 and 31 descriptions respectively. The content analysis of the data indicated that the majority of the participants perceived OC benefits related to pregnancy prevention and others' approval and barriers related to side effects and others' disapproval. Even though this study fails to describe the term "facilitation" in the context of the study, it supports the findings of the earlier studies (Eisen et al., 1985; Luker, 1975) about perceived benefits and barriers to contraception. The structured interview part of this study is noteworthy in that it reveals preference on the part of some participants to seek autonomy and what was described as self-approval in oral contraceptive use. This finding also partly confirms previous results regarding subjects' wish to gain more control over their lives (Kalmuss et al., 1987). Thus, the review presented here indicates that there is substantial literature relating to the influence of psychological factors on contraceptive practices in general.
However, the literature pertaining to a similar influence on the use of oral contraceptives is scarce and sketchy.
Forman et al. (1997) reported an OC study that determined female college
students' attitude toward over-the-counter availability of oral contraceptives. A survey containing 28 items to assess beliefs of young women about OC acquisition OTC was administered to 290 female undergraduate students at an urban women's liberal arts college in Boston, Masscahusetts. The investigators sought to determine students' beliefs about OC availability and examine factors associated with these beliefs. A dichotomous question was asked of these students assessing whether or not they believed that OCs should be available without a prescription.
Students who believed that OCs should be available OTC were asked to rank
three reasons for this belief from a list of I I choices. Those who did not believe that OCs should be available OTC were asked to rank three reasons from a list of nine choices. The results of the study indicated that 65 percent of women who responded to the survey felt that OCs should not be available without a prescription. Women cited two reasons frequently for not wanting to buy OCs without a prescription. They were: (1) physiological side effects that could have been prevented by a health care provider (59%), and (2) women would not go to their providers for regular checkups (56%). The most commonly cited reason for wanting OTC availability of oral contraceptives was that there would be fewer unwanted pregnancies (74%). Further, logistic regression performed by the authors revealed that both younger age and previous pregnancy were the important factors that related to the beliefs about acquiring OCs over-the-counter.
While the majority of the women in the study felt that OCs should not be available OTC, this study had several limitations. First, the study design used convenience sampling method to approach their subjects. Second, the overall response rate to the survey was low at about 29 percent. The study provided no scientific evidence regarding the reliability and validity of the instrument that was used in the study. Finally, the investigation lacked a sound theoretical framework to base its conclusions on. Thus, while the conclusions made by the authors were important from the standpoint of OC availability OTC, the generalizability of the study results to other populations is questionable. However, this is the only empirical research on the issue of OTC availability of OCs that is known to the author of this manuscript and is thought to be very relevant to the current investigation.
Attitude Theories Applied to Contraceptive Practices
The attitude theories are considered to be a part of a larger family of cognitive, individual choice models, which together are usually referred to as expectancy-value models. The prediction and explanation of human behavior rests on the assumption that people make a reasoned choice of behavior to enact based on instrumentalities, valences and expectancies associated with the behavior (Kanfer, 1990). Several socialpsychological models have been used in the past for the prediction of contraceptive behavior. The history of attitude-behavior consistency in the area of contraceptive practice is believed to parallel closely the developments in the theoretical literature on consistency.
Early studies done in the area of attitude-behavior relationship have generally demonstrated a weak relationship between these variables. According to Fishbein (1975), the relative neglect of the relationship between attitude and behavior could be in large part attributed to the widespread acceptance of the assumption that there is a close correspondence between the ways in which a person behaves toward some object and his beliefs, feelings, and intentions with respect to that object (p.336). Traditionally, however, given the way in which the term attitude was defined (as a learned predisposition to respond to an object in a consistently favorable or unfavorable manner), the view among researchers has been that there is a strong link between attitude and behavior. Further, this traditional view also held that any stimulus object came to elicit an attitude which mediated or determined all responses to the object. According to this view, any response could be used to infer a person's attitude. As an explanation to the weak attitude-behavior relationship, however, Wicker (1969) noted in his review of 42 studies that taken as a whole, these studies suggest that it is considerably more likely that attitudes will be unrelated or only slightly related to behaviors than attitudes will be closely related to action" (p.65). More recently, there have been attempts to specify variables that moderate the attitude-behavior relationship and to identify those situations under which attitudes will predict behavior. For example, Fishbein and Jaccard (1973) studied the situation and personal factors that render motivational (and, hence, attitudinal) variables irrelevant to behavioral prediction. It is against this backdrop that we will examine studies that investigated attitude-behavior relationship in the area of reproductive behavior in general and contraception in particular.
Wicker (1969), in his review of 42 studies obtained an average attitude-behavior correlation of approximately 0. 15 for these studies. Similarly, the investigations aimed at contraceptive behavior and attitudes, and those that studied family planning behavior have typically reported weak attitude-behavior relationship (for reviews, see Mauldin, 1965; Werner, 1977). Mauldin (1965) states, following his review of some of the KAP (Knowledge, Attitude, Practice) studies pertaining to family planning behaviors, "we know in a general way that verbal and non-verbal behavior often are not closely correlated (p.98). According to Werner (1977), the reason for a weak relationship between attitudes and behavior was that the measurements techniques used were faulty. He recommends that more attention be devoted to improving measurement procedures in family planning research or programs. Methodologically speaking, Davidson and Morrision (1983) argue that the majority of the studies involving contraceptive behavior were conducted using across-subject designs. They demonstrate in their study (1983) that within-subject models indeed provided a more adequate explanation of behavior from attitudinal constructs, and hence more accurate prediction of behavior from attitudes than did the across-subjects models. They argued that the difference in prediction reflected an important distinction between the two approaches in their orientation to behavioral prediction. According to them, in the across-subjects approach, what a person does is best described in comparison to what others do, whereas from the within-subjects perspective, what a person does is best described in comparison to what the person could have done but did not do.
Across-subj ects approach has been the common approach used by the majority of the studies reviewed here. These studies have reported somewhat weak attitude-behavior correlations, while leaving considerable room for improvement. For example, even when specific measures of attitudes were employed, the attitude-behavior relationship was found to be weak, as reported in some of the studies above. This situation has led to the development of theories that included variables other than attitude in the prediction of behavioral intention or behavior. Fishbein and AJzen (1975), for example, suggested that norms might also influence a person's decision to perform a behavior. They called their framework the Theory of Reasoned Action, which is one of the most widely tested theories in social psychology (Fishbein and AJzen 1975). This theory proposes that an individual's attitude toward a behavior and his subjective norm expectations of appropriate behavior are the immediate determinants of his behavioral intention, which precedes behavior. There is a substantial body of research that has examined the role of attitudes in the prediction of contraceptive behavior, family planning behavior, etc using the Theory of Reasoned Action (See Jaccard and Davidson, 1972; Adler et al., 1990). The results of these studies have generally supported the utility of this model in understanding the reproductive behavior.
Methodologically speaking, Davidson and Morrison (1983) point to a deficiency in traditional research. Traditional research, in their opinion, has relied heavily upon the prediction of behavior based on the (a) the actor's attitude toward performing the behavior, (b) the attitudes of other respondents toward performing the behavior, and (c) relative attitudinal intensity among the respondents. This also seems to be the approach adopted by a majority of the studies that examined various reproductive behaviors
(Kothandapani, 1971; Jaccard and Davdison, 1972; Werner and Middlestadt; 1979). These studies, involving mostly across-subjects approach, have held that individuals most likely performed the behavior when they possessed positive attitudes toward performing that behavior. In contrast, the approach used by Davidson and Morrison (1983) adds another variable to the earlier model--a person's attitude toward alternative or competing behavior and the relative intensity of these attitudes. Thus, the inclusion of a competing behavioral alternative seemed to add to the correspondence between the behavior and attitudes.
A methodological drawback in a majority of the studies reporting weak attitudebehavior relationship was that the instruments used to measure attitudes and behavior often did not take into account the multidimensional nature of attitudes, including cognitive, conative, and affective components. Thus, single attitude scores could not adequately represent all of these attitudinal components and thus could not predict behavior accurately. Ajzen and Fishbein (1973) point out that the traditional measures of attitudes, while relevant to the prediction of behavior, are insufficient. Further, they contend that most treatments of attitudes have not dealt systematically with additional variables (such as personality characteristics, social norms, habits etc.) as determinants of behavior. Instead, they argue that researchers have viewed these other variables as sources of error variance. Further stressing this point, to explain the low attitudebehavior relationship, particularly with respect to the fertility related behaviors, Fishbein and Jaccard (1973) contend that the investigators have been unable to recognize the role non-demographic factors. To resolve this issue, Fishbein (1972) has proposed an intention-based model to account for 'nonattitudinal' factors. In this model, the intention
to perform a behavior was proposed to mediate the relationship between attitudes and behavior. In addition, another component--called subjective norm--represented a social or normative influence on social behavior. Thus, it appears that, in order to obtain a substantial attitude-behavior relationship, one needs to take into account variables that are not strictly attitudinal in nature.
Experts (Fishbein, 1972; Davidson and Jaccard; 1975) argue that most of the
large-scale studies have not been based on any underlying or organizing theory. Instead, such studies are restricted to the testing of a "large number of interesting, but often unrelated hypotheses." Therefore, it is perhaps possible to achieve larger magnitude of relationship between attitude and behavior if one formulates a body of theory to guide the research.
As pointed out earlier, a concern with regard to the attitude-behavior research in the area of reproductive behavior was that very few studies actually examined the influence of variables that moderated the relationship between attitudes and behavior. For example, personality variables, prior behavioral experience (Regan and Fazio, 1977; Bentler and Speckart, 1979), including level of moral reasoning (Gorusch and Ortberg, 1983; Schwartz and Tessler; 1972) have been shown to influence the magnitude of the attitude-behavior relations in studies investigating nonreproductive behaviors. Such studies are, however, rare in the reproductive social psychology domain. A majority of the studies seems to work under the assumption that there is a direct relationship between attitudes and behavior, ignoring the role for external variables. There is no systematic research conducted to define the role of these "other" variables in influencing attitudebehavior relationship. For example, it is not known whether these variables moderate the
attitude-behavior relationship or they influence the relationship in other manner (e.g, as antecedent variables), particularly in the area of reproductive behavior research.
In partial agreement with Wicker's (1969) conclusion, Fishbein and Ajzen (1974) show that the low values are typically obtained for correlations between general attitudes (e.g., attitude toward religion) and individual behaviors (e.g., particular religious behavior such as praying before meals). These researchers, through aggregation analysis (1974), demonstrate that simply aggregating behaviors can substantially increase the correspondence between these variables. Behaviors that are aggregated establish a measure of behavior that corresponds in generality to the attitudes that are assessed. Thus, by designing multiple-act criterion which aggregates behavior over an appropriate sample of differing acts carried out in varying contexts and on differing occasions, one could actually produce relatively high correlations between an attitude and behavior (1974). However, a good majority of the studies dealing with contraceptive behavior-both prior to and after Fishbein and Ajzen's aggregation analysis--seem to have adopted single-act criterion, thus further contributing to a low attitude-behavior correspondence.
Pagel and Davidson (1984) made an effort to further evaluate attitude-behavior correspondence employing a comparison of multiple attitude models. These investigators compared the predictive validities of three prominent social psychological models of attitude and behavioral decisions, applied to contraceptive behavior. They compared Rosenberg's (1956) instrumentality-value model, Fishbein's (1963, 1967) beliefevaluation model, and Beach's adaptation model of the subjective expected utility (SEU) model (Beach et al., 1979) in the prediction of contraceptive behavior. The components of each of the models were rated by a sample of 70 female undergraduates and their
attitudes and behavioral plans toward using three different contraceptive methods were assessed. The authors reported that each of the expectancy-value models significantly predicted subjects' attitudes and behavioral plans.
There is a large body of research reported on attitude-behavior relationships, primarily in the area of contraceptive alternatives, family planning etc. (Cohen et al., 1978, Davidson and Jaccard, 1975; Jaccard and Davidson, 1972, 1979; Kothandapani, 197 1). The literature is also replete with work on the attitude-behavior relationship with respect to condom use and contraceptive utilization. However, the studies on attitudebehavior consistency concerning the use of OCs are reported less frequently in the literature.
Further examination of the studies from the methodological standpoint reveal that, while a majority of the studies seemed to employ adequate sample sizes, some did not utilize adequate numbers of subj ects (Aj zen and Fishebein, 1972; Werner and Middlestadt; 1979), thus diminishing the power. Depending on the type of research design and conceptual frameworks used, a large number of studies used correlational statistics, including multiple regression, bivariate correlations etc. In any case, certain observations are in order regarding the procedures used in many studies involving the measurement of contraceptive behavior. First, a good number of studies using Fishbein's behavioral intention model reported the prediction of behavioral intention from attitudes rather than using the actual overt behavior itself as a criterion. Some of the studies assessed self-reported, retrospective reports of behavior in the same session in which attitudes were measured. As a consequence, a desire on the part of respondents to appear consistent or a tendency to infer their attitudes from their reported behavior might have
contributed to the relatively successful predictions that were obtained. A majority of the studies examined attitudes toward single behaviors, using a large variety of theoretical frameworks available at the time. Also, a good number of studies reported reliability and validity estimates but a few studies failed to report these values (Ajzen and Fishbein, 1969; Davidson and Jaccard; 1979). The reliability and validity issues were adequately addressed in some studies through item analysis and such, especially when self-report attitude scales were constructed. However, those studies involving behavioral observation seemed to fall short of reporting good reliability estimates. Finally, as should be expected, a large number of studies utilized a fairly homogeneous sample of female respondents, particularly college students or adolescents. Therefore, generalizability of these studies to a larger population may be questionable.
Influence of Beliefs, Attitudes on Intentions and Use
The earlier literature in social psychology makes no clear distinction between
intention and attitude. Fishbein and Ajzen (1975; p.288) defined intention as a person's location on a subjective probability dimension involving a relation between himself and some action. A behavioral intention, therefore, referred to a person's subjective probability that he will perform some behavior. Fishbein and Ajzen (1975) point out that the more favorable a person's attitude toward some object, the more he will intend to perform positive behaviors (and the less he will intend to perform negative behaviors) with respect to the object. However, they contend that when attitudes and intentions are measured at different levels of target specificity, a high correlation is usually not
obtained. As the measure of intention becomes more specific (in terms of the behavior, situation, or time), its relation to attitude will tend to increase.
Currently, little is known about how women's beliefs about contraceptive
methods relate to either their intention to use them or their actual use. It is not known if women's intentions and actual use of OCs is affected by the relative desirability of available alternatives. Women who prefer other methods of contraception may have weaker intentions or be less likely to use the pill, even if they hold positive beliefs about their effectiveness. Consequently, it appears that beliefs alone about OC effectiveness are not sufficient to affect behavior. Instead, intentions to use OCs may also be related to other factors of OC use, such as cost, interpersonal consequences of their use, or social factors.
Providing insights into the possible relationship between belief and intention, a study by Moore et al. (1996) reported that female adolescents' beliefs about the contraceptive pill were predictive of their initial intentions to use the pill and their actual pill use over the course of a year. This study used a sample of 345 adolescents attending an adolescent health care clinic in the San Francisco area. These adolescents were interviewed about their beliefs regarding the consequences of using the pill and about their intentions to use OCs and other contraceptive methods. A year later, a report on their sexual activity and OC use over the course of the year was also obtained. Concerns about health and physical appearance differentiated subjects who intended to use the pill from those who did not. This study found that several beliefs reported by subjects did not relate to intentions or to actual contraceptive use. For example, the belief that using the pill would prevent pregnancy was not associated with intentions or with use. The
investigators inferred, based on this finding that the effectiveness of the pill as contraceptive was not in fact irrelevant. Rather, they contended that belief in its effectiveness did not discriminate between those who intended to use it and those who did not. Instead, perhaps there were other beliefs that discriminated between two groups more directly than did the beliefs about pill's effectiveness. This finding may be supportive of the assertion made earlier that factors other than the beliefs about OC effectiveness also accounted for intentions to use them in the future. One of the major drawbacks of this study was that it did not use any conceptual framework to base their results on.
Jaccard and Davidson (1972) conducted a study particularly relevant to the foregoing discussion about women's attitude toward using oral contraceptives. The authors asked the subjects (a college sample) to rate the concept of "using birth control pills" on a set of evaluative semantic differentials. They found that the measure of attitude was highly related to the subjects' beliefs about the consequences of using OCs and their evaluations of those consequences. An individual interview with an independent sample of 22 women was carried out to elicit 15 beliefs and 12 referents relevant to the use of OCs. For example, they were asked where they would go for more information about OCs and whether there were any particular individuals or groups who would approve or disapprove if they used birth control pills. The 12 most frequently mentioned referents that were included in the study were: mother, father, husband/boyfriend, zero population growth, women's magazines, and the 'religion I was brought up in." Normative beliefs concerning each referent and their motivation to comply with the referents were assessed using appropriate scales. These two measures
were multiplied and the resulting products for the 12 referents were summed to provide a measure for the normative component. Subjects also indicated their intention to use OCs on another 7-point scale. Then a multiple regression analysis was performed. A multiple correlation of 0. 83 5 was obtained for the prediction of intention to use OCs from the attitudinal and normative components of the theory. However, a major shortcoming of this study was that the investigators did not analyze various components of the scale and failed to provide information on scale reliability.
Another study, methodologically similar to the previous one, by the same researchers (Davidson and Jaccard, 1975) investigated the factors that underlie a woman's intention to use or not to use oral contraceptives. The respondents for this study consisted of a stratified random sample of 270 married women almost all of whom had children. The questionnaires used in the study consisted of measures of behavioral beliefs, evaluations of outcomes, normative beliefs, and motivations to comply with each referent. The measures were also taken of intentions to use birth control pills and attitude toward personal use of birth control pills. Approximately one year later women were reinterviewed and asked to report their use or nonuse of the pill during the past year. Consistent with expectations, a strong intention-behavior relationship (r--0.85) was observed, reflecting the fact that most women (93%) behaved in accordance with their intentions. Attitudes and subjective norms were in turn accurately predicted from beliefs based on salient outcomes and referents. With regard to normative beliefs, the women's major concerns centered on the expectations of their husbands or boyfriends and doctors. The women were highly motivated to comply with these referents. The women who believed that their husbands or boyfriends and doctors thought that they should use the
pills intended to do so. Conversely, the women who believed that these two referents opposed their use of birth control pills formed intentions not to use them. Only moderate motivations to comply were observed with respect to the remaining referents such as mother, father, friends and the religion they belonged to.
Almost all of these studies examined the influence of beliefs and attitudes on intentions to use OCs or other forms of contraception. None of the studies actually examined this relationship with reference to the choice of or intentions to choose contraception of any kind. Therefore, there exists a knowledge gap in contraception research that looks at the role of beliefs and attitudes on intention to choose (rather than use of) contraception of a particular kind. However, these studies did indicate that contraceptive behavior and other fertility-related intentions might be understood in terms of the theoretical model of reasoned action. Furthermore, the results of these studies indicated that a woman's intention to use a certain method of contraception is ultimately determined by her beliefs concerning the advantages and disadvantages of using the method in question and by her beliefs that relevant others think she should or should not use that method. Thus, it can be argued based on the evidence gathered above that a woman's intention to acquire oral contraceptives in a particular way should then be dependent on her attitudes toward the same and her perceived expectations of those who are important to her. Her beliefs, attitudes, and her subjective norm should also shape any preference a woman may have for acquiring OCs by different means. The argument above is therefore central to the adoption of a theoretical model of behavioral intention that incorporates these components.
The evidence from the reviewed literature suggests that contraceptive behavior (practice) can be better understood by examining the beliefs that underlie the attitude toward contraception. Attitudes toward contraception in general, and oral contraception in particular, seem to be based on the perceived advantages and disadvantages associated with the method in question. In addition, the available evidence indicates that a better representation of the relationship between contraceptive attitude and behavior is obtained when the mediating role of intention between the attitude toward the behavior and the ultimate behavior itself is taken into consideration. Any model that examines women's oral contraceptive acquisition behavior needs to take into account the relevant literature and consider the limitations of some of the models reviewed in the literature in explaining the OC acquisition behavior, particularly when the ultimate behavior itself is unobservable from the practical standpoint. The next chapter presents a theoretical framework that seeks to explain how oral contraceptive acquisition over-the-counter is understood in terms of women's intentions to engage in this behavior.
The theoretical framework and constructs that underlie the current research is briefly described in this chapter. This study utilizes the theory of reasoned action (Fishbein and Ajzen, 1975) in conceptualizing the research problem. A description of the theoretical model and hypothesized relationships among variables is presented here. The research questions that will be addressed by the study are also presented.
The theory of reasoned action is traditionally used for the purpose of predicting a specific behavior based on a concept known as behavioral intention. The theory holds that the behavioral intention is derived from a person's attitude toward performing the behavior in question and social norms that result from a person's perceptions of others' expectations toward performing that behavior. The attitude towards the behavior in turn is conceptualized to be a function of certain beliefs that a person holds regarding the performance of the behavior. The specific behavior in question is the acquisition of oral contraceptives over-the-counter. While all the model constructs to be discussed in this chapter are expected to underlie this specific behavior, the primary response variable of interest to this research is the behavioral intention--intention to acquire OCs over-thecounter. The beliefs regarding the acquisition of OCs over-the-counter, attitudes derived from these beliefs, and social norms are conceptualized to predict the intention to acquire
OCs over-the-counter. The literature reviewed in the previous chapter indicated that nonattitudinal variables (such as social expectations) were found to add substantially to the prediction of behavior when included in the model along with attitudinal variables. Thus, the theoretical framework to be used in the prediction of intention to acquire OCs over-the-counter will comprise both attitudinal and social constructs.
Theory of Reasoned Action
The model central to this study, the Fishbein model (Fishbein and Ajzen, 1975; AJzen and Fishbein, 1980), is designed to predict an individual's intention to perform a behavior. Figure. 1 displays the schematic representation of the theory of reasoned action (TRA). This model is predicated on the assumption that most human behavior is to some degree volitional in nature and hence guided by the behavioral intent of the individual. Further, the model assumes that human beings make rational decisions and also make systematic use of information available to them. An individual's intention, according to the theory, is the immediate determinant of his/her behavior. The model contends that the strength of one's intention to perform a behavior is a function of two factors: (a) beliefs and attitudes about the consequences or outcomes of performing the behavior, and the evaluations of those consequences, and, (b) subjective normative beliefs about what others think the actor should do and the actor's motivation to comply with those beliefs. Thus, behavioral intentions (BI) are held to be a joint function of attitudes toward performing a particular behavior in a given situation and the social norms perceived to govern that behavior. Other variables, such as individual characteristics or situational aspects in which action occurs, are treated as external to the model. The model holds that
all such external variables influence behavioral intentions indirectly, that is, if and only if
they influence one or more of the model's predictors. Fishbein (1972) emphasizes that
the attitude being assessed is the individual's attitude toward the performance of that
behavior, and not an attitude toward a given object, value, person, or situation.
Variables_______Attitude toward Belief that the Attd
Targets behavior leads to twr
certain outcomes behavior
Evaluation of RelativeB
Deorahcthe outcomes importance ofE
Age, sex, occupation componentsI
education Belief that specificR
shul rshud______________________ nt pefor th
Authritriansmomly witshol Dominanceo spefcrfrentse
Figre : Teor ofReaoneaAc io ASeuadbjhbn c180tige 7.1,prm
Pexsnlte ra ariabeAlvralsntcosdrdbshhoy
Deten-ninants of Attitudes and Subjective Norm
An individual's attitude toward a behavior, according to the model's proponents, is a multiplicative product of the individual's beliefs about probable consequences of engaging in that behavior and the individual's subjective evaluation of these consequences. On the other hand, the multiplicative product of the individual's normative beliefs (belief that specific individuals or groups think he should or should not perform the behavior) and motivation to comply with these beliefs, determine the subjective norm of an individual. The model can be expressed symbolically as follows:
(B) & BI = (AB) wl + (SN) w2
B Overt behavior
BI Behavioral intention
AB- Attitude toward performing the behavior in a given situation
SN Subjective norm
W I, W2 Empirically determined weights
First, AB, is the actor's attitude toward performing the behavior in question under a given set of circumstances. Here, a person's attitude toward a specific behavior is proposed to be a function of the perceived consequences of performing that behavior and of the person's evaluation of those consequences. Thus,
AB = Ibiei with i =1 through n
Where b is the belief that performing behavior B leads to consequence or outcome I; e is the person's evaluation of outcome 1; and n is the number of beliefs the person holds about performing the behavior B.
The second or normative component of the theory, SN, refers to the influence of social environment on the behavior. The subjective norm is the person's perception that most people who are important to him think he should or should not perform the behavior in question. According to the theory (Fishbein and Ajzen, 1975), the general subjective norm is determined by the perceived social expectations of specific referent individual's or groups, and by the person's motivations to comply with those expectations. Symbolically,
SN = Ebimi
Where bi is the normative belief (i.e., the person's belief that reference group or individual i thinks he should or should not perform behavior B); mi is the motivation to comply with referent i; and n is the number of relevant referents. According to the theory, the bXm products are computed for each relevant reference group and summed. The sum is viewed equivalent to a "generalized normative belief," i.e., subjective norm
According to the proponents, the empirical weights wl and w2 are expected to
vary with the kind of behavior which is being predicted, with the conditions under which the behavior is to be performed, and with the person who is to perform the behavior. Some behavioral intentions may be influenced more by normative considerations (expectations of friends, family etc.) than by attitudinal considerations (the expected outcomes of the act). The reverse may be true for some other behaviors. Fishbein and Ajzen (1975) point out that the weights for the attitudinal and normative components are estimated by using standardized regression coefficients resulting from multiple regression
analyses. Thus the present equation represents a multiple regression with two predictors AB and SN, with the criterion being 1, the behavioral intention under consideration.
As pointed out earlier, any additional variable is held to influence BI if, and only if, it affects one or more of the model's predictors. Thus, according to Fishbein and Ajzen, situational variables, personality characteristics, etc. will influence a person's behavioral intentions only if they are related to AB or SN, or if they influence the relative weights of the two components. It should be noted here that the attitudinal and normative components of the model have a direct influence on BI, which in turn is expected to account for most of the behavioral variance. Thus, according to the theory, if one can predict BI, one can also predict B with only slight attenuation.
Description of the Model Constructs
According to Ajzen and Fishbein (1980), the attitude toward a behavior is a person's positive or negative evaluations of performing that behavior. Although the behavior in question in this study is that of acquiring oral contraceptives over-the-counter (Figure 2), the intention to acquire OCs over-the-counter is the predicted variable of interest in this research. This model essentially holds that the relationship between the OC acquisition behavior and attitude toward acquiring OCs over-the-counter is mediated by the intention to acquire OCs over-the-counter. This behavioral intention is a function of a person's attitudes toward performing the behavior (personal component) and subjective norms about performing the behavior (normative or social component). Further, a component called intention to acquire oral contraceptives OTC mediates the relationship between attitudes and the behavior. A woman's attitude toward performing a
behavior, such as acquiring pills over-the-counter, is determined by her salient beliefs about doing the same. People usually believe that performing a given behavior will lead to both positive and negative consequences. Their attitude toward the behavior correspond to the favorability or unfavorability of the total set of consequences, each weighted by the strength of the person's beliefs that performing that behavior will lead to each of the consequences (Ajzen and Fishbein, 1980). For example, a belief concerning the acquisition of birth control pills OTC could be that a woman might miss her regular health examination, which she typically receives on a visit to a health care clinic which she might consider to be an unfavorable negative consequence. Similarly, another belief about acquiring oral contraceptives OTC may be of that of convenience and cost savings, a favorable positive consequence. Given this set of beliefs, the theory holds that a person's attitude toward a behavior can be measured by multiplying her evaluation of each of the behavior's consequences by the strength of her belief that performing the behavior will lead to that consequence and then summing the products for the total set of beliefs (Ajzen and Fishbein, 1980).
The subjective norm is the individual's perception that most people who are
important to him or her think s/he should or should not perform the behavior in question (Fishbein and Ajzen, 1975; p.302). The general subjective norm is determined by the perceived expectations of specific referent individuals or groups, and by the person's motivation to comply with these expectations. It is believed that in the behavior relating to the acquisition of OCs over-the-counter, expectations of a person's family or friends may play a significant role. However, in other instances, the expectations of the society at large may be most influential. Typically, the expectations of more than one reference
group will have to be considered. In addition, it is also necessary to measure the individual's motivation to comply with each of the relevant groups. According to the equation above, the bXm products are computed for each relevant reference group and summed.
Some of the consequences of performing a given act may please or displease
relevant reference individuals or groups, and it may lead to reward or punishment from a given referent. Fishbein and AJzen (1975, p.3 06) acknowledge that the concept of motivation to comply has been interpreted differently by researchers. They point out that motivation to comply at times may not be behavior-specific. That is, a person may be generally motivated to comply with his or her referent regardless of that referent's particular demands. Alternatively, the motivation to comply could refer to referent's wishes concerning the particular behavior or behavioral domain under consideration. However, they contend that, referring to both theoretical and empirical grounds, motivation to comply can be best conceived as the person's general tendency to accept the directives of a given reference group or individual. If a person's general tendency is to comply with her referent concerning a much larger behavioral domain--such as reproductive behavior, taken as a whole, as opposed to just OC acquisition--then she is much more likely to accept the directives coming from that referent. Consider, for example, the normative belief that my mother thinks that I should not use any form of contraception." A person can hold this belief about others' perceptions independent of her own beliefs regarding the behavior per se, such as I believe buying birth control pills over-the-counter is safe." Here, even when a person holds positive beliefs about the action, the subjective norm is contradictory to such beliefs. More over, it can be seen that
the individual's perceptions of referent's beliefs related to a much larger behavioral domain contraception in general, whereas, the individual's beliefs toward performing the behavior related more specifically to the acquisition of OCs OTC. Another example: 46my partner thinks I should obtain birth control pills OTC to make family planning easier (a subjective norin)," while "I think it is not safe for me to use birth control pills bought OTC for family planning since they cause major side effects." Here, too, the perception of what a particular referent might think is at odds with one's own personal beliefs. Yet, such perceived expectations, regardless of one's own beliefs, may govern. the ultimate choice of behavior. Thus, the expectations of others regarding one's behavior in question may in fact influence one's own attitude toward the behavior. In other words, beliefs about a referent's reactions may have different effects on the attitudinal and normative tendencies of a person and, consequently, on his/her behavioral intentions.
Fishbein and Ajzen (1975) note that normative beliefs may be inferred from the referent's perceived attitude toward performing a given behavior. If the referent is perceived to have a favorable attitude toward performing the behavior, or more specifically, toward the person's performing the behavior, the normative belief may be formed that the referent thinks the person should perform the behavior in question. Thus, the framework suggests that the oral contraceptive behavior may in fact be understood in terms of the theoretical model of behavioral intention.
Several research questions and hypotheses are derived by examining the relationships of variables in the model specified in Figure 2.
OCs OTC Intention
to acquire Acquisition of
Figure 2: Theory of Reasoned Action Applied to OC Acquisition
In writing these questions, a hypothetical situation that individuals have legal access to OCs over-the-counter is assumed. The research questions examined are: I.What is the predictive power of attitude toward acquiring OCs over-the-counter in the prediction of a woman's intention to acquire OCs over-the-counter in the hypothetical situation that it is the legal means of acquisition?
2. What are the beliefs that will describe the differences in intention between women who intend to acquire oral contraceptives over-the-counter and those women who do not?
3. Are attitudes regarding perceived risks and benefits of acquiring OCs over-the-counter predictive of a woman's intention to acquire them through that means?
4. What is the relationship between a woman's subjective norm, including motivation to comply with the expectations of her significant others, and her intention to acquire OCs over-the-counter?
5. What is the relative contribution of attitudes and subjective norms in predicting intentions to acquire OCs over-the-counter?
Even though the nature of this investigation is largely exploratory, several
hypotheses have been developed here. These hypotheses are based on the model and empirical studies discussed in the accompanying literature review. The hypotheses specifying the relationships among variables are consistent with the theoretical framework and concepts incorporated in the model. Furthermore, these hypotheses address the research questions stated above as well as the study objectives outlined previously in the proposal.
The first two research questions are directed at the explanatory power of the theory of reasoned action in the prediction of behavioral intention to acquire oral contraceptives over-the-counter. The following hypothesis is proposed in support of the first research question. The second research question is exploratory in nature and no hypotheses have been proposed to support it.
H I: A woman's intention to acquire oral contraceptives over-the-counter may be predicted from (a) her attitude toward acquiring them OTC and (b) her normative
beliefs, weighted by her motivation to comply with those perceived norms.
The next research question addresses different factors -- beliefs about biomedical, social, and economic aspects of OC use -- that influence an individual's decision to acquire oral contraceptives OTC. Although there exists limited empirical evidence to support these hypotheses, the major thrust for the statement of this hypothesis comes from the preliminary examination of OC-related attitudes carried out by the authors (Nayak et al, 1994). In addition, opinions expressed by the experts (Grimes, 1993;
Trussell et al., 1993; Potts, 1995) and communications appeared in academic journals further justify the generation of the hypothesis stated below.
H2: There will be a positive correlation between favorable attitudes associated with the use of OCs acquired OTC and intentions to acquire OCs through that
The fourth research question addresses a woman's perceptions of general social
expectations regarding the behavior as a function of the person's set of beliefs about what significant other people want the person to do. With respect to the current research, the Fishbein model posits that women's perceptions of subjective norms regarding the choice of means to acquire OCs will be associated with the perceived expectations of her partner, mother, father, and other relevant people regarding the method, weighted by her motivation to comply with those expectations.
H3: Perceptions of the wishes of other people, weighted by the woman's
motivation to comply with those wishes, will be related to a woman's intention to
acquire OCs over-the-counter.
F14: The directly measured perceptions of subjective norms more supportive of
OC acquisition over-the-counter will be associated with stronger intentions to
acquire OCs over-the-counter.
The last research question examined in this study is purely exploratory in nature. Therefore, no hypotheses have been proposed to address this question in the study.
Constitutive Definitions of Model Constructs Attitude toward behavior. A person's attitude toward behavior is his or her beliefs about the probable consequences of engaging in a behavior multiplied by his or her subjective evaluation (e.g. the degree of "goodness" and "badness") of each consequence. Behavioral intention. An assessment of an individual's own subjective probability that he or she will perform a given behavior. A psychological construct distinct from attitude, it represents the person's motivation in the sense of his or her conscious plan to exert effort to carry out a behavior.
Motivation to comply. A measure of how important it is to an individual that s/he behave in the way that those important to him or her think s/he should behave. Normative belief. This is a belief underlying a person's subjective norm that specific individuals or groups think s/he should or should not perform the behavior. Subjective Norm. This is the person's perception that most people who are important to him or her think s/he should or should not perform the behavior in question, also known as "generalized normative beliefs".
This project required an instrument that would measure attitude toward acquiring OCs over-the-counter, subjective norms regarding the acquisition of OCs over-thecounter, and intention to acquire OCs over-the-counter. The mail questionnaire consisted of measures on women's beliefs about acquiring OCs over-the-counter, evaluations of these beliefs, subjective norms regarding the acquisition of OCs OTC, and intention to acquire OCs over-the-counter. In addition, direct measures of general attitude toward acquiring OCs over-the-counter and subjective norms were also included in the questionnaire as a method to provide validity checks to the measures.
The survey research procedures outlined in Dillman (1978) were used to guide the development of the mail questionnaire and follow up telephone interviews. Following the development of the initial version of the survey instrument, it was pilot tested and revised for use in the research being presented. As part of the main research, the follow-up telephone interviews of subjects who did not respond to the mail questionnaire was conducted approximately five weeks after the initial bailout of the questionnaire. This chapter reports on sample selection, data collection procedures, measurement of the independent and dependent variables, and validation based on the results from pilot studies, and the data analysis strategy for testing research hypotheses.
The study population consisted of women attending the University of Florida at the Gainesville campus. Two randomly selected samples of women registered for the Summer and Fall, 1998 semesters were used for the pretest and for the main study respectively. All women aged 18 and over and enrolled in the University of Florida were eligible for inclusion in the study. University of Florida students are geographically and culturally diverse. They come from more than 100 countries, all 50 states and each of Florida's 67 counties. In Fall 1997, University of Florida had a total enrollment of 41,713, of which 2561 were African American students, 3767 were Hispanic students, 2381 were Asian-American students, and 1751 were international students. Some 46 percent of the students were women. Of the nearly 42,000 students enrolled in the Fall of 1997, some 77 percent were undergraduates, 17 percent were graduate students and 6 percent were in professional programs (including dentistry, law, medicine, pharmacy and veterinary medicine).
The use of contraception of any kind was not used as a criterion for inclusion. It was reasonable to expect in any given population particularly in a campus population for some women who have not used any form of birth control in the past to be contemplating the use of OCs or another form of contraception in the future. Regardless of the history of current use or nonuse of contraception, women were expected to hold certain views regarding the acquisition of oral contraceptives.
A listing of female students who were enrollees at the University of Florida was obtained from the Registrar's office. This list constituted the sampling frame for the study. A systematic sampling strategy outlined in Agresti and Finlay (1986) was employed to select subjects for participation from the list provided. According to this method, a systematic random sample is one in which (a) a member is chosen at random out of the first k names in the sampling frame, and (b) every kth member after that one is selected. Here k = N/n; where, N is the female student population size, and n is the desired sample size derived from the pilot study.
Procedures to Enhance Participgji on
Efforts were made by the investigator to establish trust with subjects by stressing the confidentiality aspect of their response to the questionnaire. In addition to assuring the anonymity, the subjects were told of the value of their input to this research endeavor in order to emphasize the importance of their contribution to health care research. They were made aware that their participation in the study contributed to the society by helping the researchers gain improved understanding of the problems and issues facing the policy makers regarding the use and sale of OCs in the US. Questionnaires were made brief, their format attractive, and their content as easy to understand as possible. The procedures recommended in Total Design Method (TDM) proposed by Dillman (1978) to enhance participation in mail surveys were adopted in the current study. The financial costs associated with stationary, postage, telephone interviews etc. were paid by the funds
allocated by the Division of Sponsored Research, University of Florida, in support of doctoral dissertations.
Sample Size Estimation
The proposed research was considered to be largely exploratory, thereby rendering a priori predictions of the distribution of values of the study variables impossible. Therefore, the determination of the sample size was done following the pilot study, which validated the instrument used in the study. The approximate sample size needed to perform the proposed statistical analyses in the study was based on the test statistics obtained and the distribution of values of study variables resulting from the pilot test of the instrument. For example, the sample size estimation was based on the association statistics such as bivariate correlation coefficients and/or multiple correlation statistics, as reflected in the research hypotheses.
Data Collection Procedures
A new questionnaire, specifically developed for the purpose of measuring women's attitude toward acquiring OCs over-the-counter was used to collect data (Appendix A). The research data was collected in two steps. The preliminary step involved the collection of data for the pilot testing of the instrument. The questionnaire containing measures of OC attitudes and intention to acquire OCs over-the-counter were used in the pilot study to obtain data to establish instrument reliability and validity. The campus Institutional Review Board permission was obtained through appropriate documentation.
Procedures for the Written Questionnaire
On July 23, 1998, a thousand questionnaires were mailed to the previously
selected respondents along with a self-addressed, stamped envelope. A cover letter was designed for informing the potential respondents of the study objectives and requirements (Appendix B). The data collected in this step was used for further validation of the instrument.
The second step of the study involved data collection for the purpose of testing research hypotheses, following instrument revalidation. On September 16, 1998, a mailing procedure similar to the one described above for the pilot-testing of the instrument was used for the final study sample. A randomly selected sample of 500 female students was used for the final study. However, for this sample, a combination of mail and telephone follow up procedures was used to enhance participation in the study. The questionnaires were marked with code numbers for the identification purpose only, linking them to the subject's name and address. The respondents' names were then checked off from the list when a completed questionnaire was received. The file linking code numbers, address, and telephone numbers was kept in a password protected file accessed only by the investigator. Subjects were informed of the objective for number coding in the cover letter (Appendix C) and were assured of complete confidentiality of their response to the questionnaire. A reminder postcard was mailed to the nonrespondents on September 30, 1998, exactly two weeks after the original bailout (Appendix D). An additional copy of the questionnaire was sent to those who had not received or lost the questionnaire, following telephone or e-mail requests to the
investigator. This consisted of a cover letter, including restatement of basic appeals from the original cover letter, a replacement questionnaire, and another return envelope.
The written questionnaire was 5 pages long and had 48 items. It was expected that the subjects would need about 10- 15 minutes to complete the questionnaire. Written questionnaires consisted of independent and dependent variables, both measured at the same time. Subjects mailed completed questionnaires to a post office box number.
Procedures for the Telephone Survey
Where possible, follow-up telephone calls were made to the nonrespondents,
urging them to respond to the survey. The telephone interview was conducted during the first week of November 1998, approximately one month after the nonrespondents received the reminder postcards. For the purpose of telephone administration of the survey, the investigator sought professional assistance. The services of Florida Survery Research Center (FSRC), a not-for-profit, on-campus survey organization affiliated with the University of Florida, were used to approach the subjects who had not yet responded to the survey. A list of nonrespondents containing telephone numbers of the subjects was provided to the Center by the investigator. The IRB exemption letter was used to initiate the process and the protocols outlined in the IRB manual were adopted to assure confidentiality of responses and subject anonymity. Specifically, the FSRC survey services included the following:
1. Preparation of survey questions in the format used for telephone inter-views;
2. Training of interviewers and other necessary steps to prepare for the
3. Efforts to contact all of the individuals in the sample of students who had not
completed the mail questionnaire;
4. Three attempts, including two call backs in an attempt to complete the
5. Maintenance of records of each contact /attempted contact (Appendix E)
including reasons for refusing to participate;
6. Careful review of each completed instrument for accuracy;
The information available at the time on contact numbers obtained from the Registrar's office included a few non-working numbers, business numbers, as well as residential telephone numbers. The average length of the interview was 12 minutes. Where possible, qualitative information that was available from the respondents was recorded by the interviewers.
Study Variables and Operationalization of Constructs General Variable Description
Table 4.1 lists the variables examined in the study and information on their
operationalization and measurement. Table 4.2 depicts a semantic hierarchy of variables representing the relevant model constructs and their measurement. The predictor variables for this study were attitude toward the behavior, namely, the attitude toward acquiring OCs over-the-counter, and subjective norms regarding the acquisition of OCs OTC.
Table 4.1 Description of Study Variables Variable Measurement
Age Years of age
Race By categories: African American, white, Asian American,
Religion Categories: Catholic, Protestant, Jewish, Other
Other variables of interest
Previous use of OCs Dichotomous: l=yes O=no
Current use of OCs Dichotomous: lyes 0--no
Duration of OC use Categories scale: months of use
Type of health professional seen for Categories: Physician, NP, Midwife, Other
Plans to use OCs in the future Categories: Yes, No, Don't know
i) for MD visits Categories: Yes, No, Don't know
ii) for prescription or nonprescription drugs Categories: Yes, No, Don't know
a. Attitude toward acquiring OCs over-the-counter (Indirect 5-point Likert type of scale Measure) (9 items)
Evaluation of Consequences Very important (+2).very unimportant(-2)
Beliefs about acquiring OCs OTC Very likely (+2) ........ very unlikely (-2)
b. Subjective norms regarding the acquisition of OCs OTC 5-point Likert type of scale(3 items) (Indirect Measure)
Very likely (+2) ........ very unlikely (-2)
Normative beliefs Very likely (+2) ........ very unlikely (-2)
Motivation to comply
c. General attitude toward acquiring OCs over-the-counter(Direct 5-point Likert type of scale (2 items) Measure) Extremely good (+2) ..... Extremely bad(-2)
Extremely favorable(+2) ... Extremely
d. Subjective norms regarding the acquisition of OCs (Direct 5-Point Likert scale (One item) Measure) Very likely (+2) ........ very unlikely (-2)
Predicted Variable 5-point Likert type of scale (One item)
e. Intention to acquire OCs OTC Very likely (+2) ........ very unlikely (-2)
Table 4.2 Semantic Hierarchy in OC attitude research
Constructs/ Attitude Toward Behavior Attitude Behavioral
Concepts (Indirect Measure) Toward Intention
Attitude toward acquiring OCs over-the- Behavior
counter (Direct measure)
Behavioral Beliefs Subjective Norm acquiring OCs overthe-counter
Beliefs about risks of using Normative beliefs about Attitude toward Intention to acquire
OCs acquiring OCs OTC acquiring OCs over- OCs over-theBeliefs about benefits of using Motivation to comply with the-counter counter
OCs. referents regarding the
acquisition of OCs OTC
Variables (Eevaluation of beliefs X belief
strength = attitude)
(5-point differential (5-point Likert- (5-point Likert (5-point
scale) type scale) scales) Likert-type
A. Evaluation of Beliefs A.Normative beliefs 1. "In my opinion, following a
I .Indicate on a scale below the acquiring OCs over- hypothetical
degree of "importance" you I. If you decided to use birth the-counter without an scenario .......
Measurement think the effect of experiencing control pills, how likely is it MD's prescription is"
(observables) nausea as a consequence of that your family would think I .Assuming that you
Examples using pills would have for you that you should buy them Extremely good
without a doctor's .... Extremely bad birth control pills, how
Very important.... very prescription? likely is it that you
unimportant would acquire them
Very likely .... Very unlikely 2. "My opinion about without a physician's acquiring OCs over- prescription?
B. Subjective probability 2. If you decided to use birth the-counter without an
(5-point Likert scale) control pills, how likely is it MD's prescription is:" Very likely ..... V1ery
that your partner would unlikely
1. What is the likelihood that think that you should buy Extremely favorable -you will experience minor side them without a doctor's .----------Extremely
effects such as mild nausea prescription? unfavorable
when you have acquired pills
without a doctor's prescription? Very likely.... Very unlikely.
Very likely ...... Very unlikely B.Motivation to comply
2 What is the likelihood that you "Generally speaking, I will will avoid unplanned pregnancy do what my family thinks I when you have acquired pills should do." without a doctor's prescription?
Very likely ...... very unlikely Strongly Agree -----Strongly disagree
These variables were used to predict the individual's intention to acquire OCs
over-the-counter. The purpose of each variable is discussed in detail in the following
section. Consistent with the theoretical formulation and the literature reviewed, the study
instrument also consisted of direct measures (as opposed to multiplicatively derived
measures) of both attitude and subjective norm. The items corresponding to these constructs provided a direct measure of women's attitudes and subjective norms regarding the acquisition of OCs over-the-counter, These measures served as validity checks for the theoretical constructs embedded in the TRA. For example, based on the theory, the measures of these constructs, derived indirectly by the multiplication of the model components, should correlate positively with the measures taken directly. The statistical check performed this way further provided us with the means to confirm the theoretical validity of the model being used. Therefore, it was a point of interest to the current investigation to include both indirect and direct measures of model predictors.
In sum, consistent with Fishbein formulation, both direct and indirect measures
for attitude toward acquiring OCs over-the-counter and the subjective norm regarding the same were developed in this study. Direct measures were those that incorporated attitude and subjective norm statements on a 5-point scale designed to measure response differentials such as 'very important' and 'very unimportant,' 'extremely favorable' and extremely unfavorable, 'very likely' and 'very unlikely' etc. For example, an item, "My attitude about obtaining birth control pills without a doctor's prescription," measured directly a woman's attitude toward acquiring OCs over-the-counter on a 5-point, 'extremely favorable- extremely unfavorable' scale. Similarly, the subjective norm regarding the acquisition of OCs over-the-counter was measured using the item, "Most people who are important to me and whose opinion I value think that I should obtain my birth control pills over-the-counter, without a physician's prescription," on a 5-point 'very likely-very unlikely' scale. Indirect measures were derived as the multiplicative products of belief strength (i.e., perceived likelihood) and evaluation of consequences
(for attitude measures), and as multiplicative products of normative beliefs and motivation to comply (for subjective norm measures).
Attitude Toward Acquiring OCs Over-The-Counter (Indirect Measure)
According to Aj zen and Fishbein (1980), the attitude toward a behavior is a person's positive or negative evaluations of performing the behavior. The behavior in question in this study was that of acquiring oral contraceptives OTC, and the response variable of interest was the individual's intention to acquire OCs over-the-counter. A woman's attitude toward performing a behavior in question is determined by her salient beliefs. People usually believe that performing a given behavior will lead to both positive and negative consequences. Their attitude toward the behavior corresponds to the favorability or unfavorability of the total set of consequences, each weighted by the strength of the person's beliefs that performing that behavior will lead to each of the consequences (Ajzen and Fishbein, 1980). A person who believes that the acquisition of OCs over-the-counter leads to a positive consequence, such as easier fertility control, may find it desirable to buy them OTC. As a result, she is likely to hold more favorable attitudes toward acquiring OCs over-the-counter. Similarly, beliefs about negative aspects of acquiring OCs over-the-counter--such as minor side effects--may lead to more negative, less favorable attitudes toward acquiring them over-the-counter. According to the theory, a person's attitude toward a behavior can be predicted by multiplying her evaluation of each of the behavior's consequences by the strength of her belief that performing the behavior will lead to that consequence and then summing the products for the total set of beliefs (Ajzen and Fishbein, 1980). The salient beliefs in the proposed
research can be represented in terms of the following categories: (i) her beliefs about risks associated with using OCs, (ii) her beliefs about benefits of using OCs. Each one in turn is discussed in detail below.
The salient beliefs incorporated in the study instrument basically related to the
beliefs about possible outcomes of using oral contraceptives acquired OTC. According to the theory, these are termed behavioral beliefs. Belief statements representing the categories above were presented to each subject and her degree of evaluation (i.e., important/unimportant) of each possible outcome when using OCs was measured. This part of the scale is referred to as the evaluation component of the scale. Then the responses to items that measured subjects' perceived likelihood (or belief strength) of each outcome resulting from the use of OCs when they are acquired OTC as well as with a physician's prescription were scored on the questionnaire. This part of the scale is referred to as the belief component of the scale. In other words, the instrument consisted of a subscale that measured evaluations of the probable outcomes of using OCs in general, and subscales that measured the perceived likelihood of each of these outcomes resulting when OCs are acquired OTC or with a physician's prescription. The scores on each item on the evaluation and belief components of the scale were then multiplied and summed across all the items for each individual. This score represented the attitude toward acquiring OCs over-the-counter. Thus, attitude scores were calculated for each individual, representing the OC-attitude for that individual. More details on the belief categories represented in the scale are as follows.
Beliefs about risks
An individual's beliefs about risk in this case may be one of two types. The first type of risk perception occurs due to the direct effects of the drug on the body. These are inherent risks of minor and serious side effects of the drug itself, regardless of the type/level of supervision. The second type of perceived risk occurs due to a) the risk of therapeutic failure as a result of lack of professional oversight, or b) a general health risk associated with the foregoing of certain preventive health services. Such perceptions may arise due to what can be called indirect effects associated with the use of the drug. These perceptions are usually psychosocial in origin.
One of the major perceived risks of using the pills is the susceptibility of an
individual to various types of ovarian and endometrial cancers, breast cancer, and stroke
- often trained as direct effects of the drug on the body. Some minor side effects such as pain in the legs, vision defects, breakthrough bleeding, weight gain etc. also generate similar risk perceptions. The second type of perceived risk is often the result of imperfect use of OCs without proper clinical counseling. This risk factor is likely to figure into the judgments of those women who might consider choosing only that option which offers some kind of medical safeguards over the use of OCs. OCs are contraindicated in some women with conditions that rule out the use of pills. Women who are aged over 35 and who smoke, and those who are pregnant, or those who breast-feed are the unlikely candidates for oral contraceptives. Also, myriad brands and different strengths of OCs (often considered to be more of a marketing decision than one of any therapeutic relevance) will possibly enhance the perception of risk women may associate with a given option.
The perceptions of general health risk stem from women's belief that acquiring
OCs from a source that involves little or no professional supervision may mean foregoing certain preventive health services and screening for cancer. The status quo requires that women submit themselves to some preventive health services like pelvic examination and pap smears to screen for cancer, even though the health conditions being screened for are unrelated to decisions about oral contraception.
Another perception of risk, neither direct nor indirect, is the risk of having to pay more for OCs and related services if they are sold OTC. It is believed that making OCs available OTC may present some obstacles for certain women. Some fear that poor women, who obtain OCs inexpensively from family planning clinics, may now pay higher prices because of the possible discontinuation of price discounts by companies to these clinics. A possible lack of private insurance and/or Medicaid coverage for OCs is believed to present problems to women who may wish to acquire OCs without a physician's prescription. Thus, a woman faced with options for acquiring OCs by different means is likely to weigh the benefits and risks involved in choosing an option and hold intentions for acquiring OCs through only that means which poses minimal risk to herself and to the society. A more detailed discussion of both direct and indirect effects of using OCs is presented in the section on literature review. Specifically, five items representing women's beliefs about risks of using OCs per se were incorporated in the questionnaire. These items were expected to tap into women's perceptions of risks--both direct and indirect--stemming from the use of OCs.
Beliefs about benefits
The most important benefit of using oral contraceptives is the prevention of pregnancy and the regulation of menstrual cycle. Other important beliefs about the possible benefits include benefits accruing from improved compliance, professional counseling and possible convenience depending on the way in which OCs are acquired. Perceptions of benefits will also include beliefs about cost savings when OCs are acquired OTC. A woman may strongly believe that seeing a physician in order to receive her birth control advice is not an effective means of exercising her reproductive option, either because she feels competent and knowledgeable enough to care for herself or because it is not worth her time, and thus may attach moderately negative value to it. At the same time, a woman may believe that seeing a physician is necessary because it is likely to ensure her overall health, an outcome to which she is likely to attach a very high value. The OCs are believed to protect against several forms of cancer a benefit so potent that some women are even strongly recommended to use OCs on a regular basis regardless of their need for contraception. Also of benefit to a woman is the possible use of OCs as post-coital or emergency contraceptives (morning-after-pill). In addition, our preliminary investigation (Nayak et al., 1994) also indicated the convenience aspect of OC acquisition. In this study, some women when told of an OTC option for the acquisition of pills, felt that OCs were more convenient to acquire OTC since this option entailed no physician visits or long appointments. Some women also felt OCs were convenient to get, particularly from family planning clinics and the student health care clinic on campus, since this type of acquisition did not entail parental consent.
A more detailed account of monetary aspects of acquiring OCs from different
sources is presented in the section on literature review. The reviewed literature indicates that OCs may be cheaper to acquire OTC for those who obtain their pills from officebased practitioners and those who do not have prescription insurance coverage. These women are better off receiving them over-the-counter perhaps after an initial visit to the physician if felt necessary. Such women are also believed to save money on unnecessary physician visits, particularly for refill authorizations. Thus, conditions mentioned above were conducive to shaping a woman's beliefs about costs associated with each of the options available to her and influence her evaluations of each of the possible consequences. Therefore, beliefs about benefits contained items measuring the following beliefs: beliefs about prevention of pregnancy and regulation of menstrual cycle, beliefs about convenience, and beliefs about emergency contraception. Four items on the instrument were expected to capture the perceptions of benefits of using OCs.
In sum, the evaluation component of the scale consisted of nine items
representing nine different possible consequences when using oral contraceptives, each corresponding to the beliefs categorized above. The items were generated based on the literature reviewed on contraceptive behavior and preliminary work conducted by the investigator. Women rated each of the nine consequences in terms of its importance: how important (very important) or unimportant (very unimportant) it would be if using OCs had that consequence for them. For example, the respondents would rate on a 5-point scale the degree of importance, ranging from very important to very unimportant, each consequence--such as minor side effects, prevented pregnancy, regulation of menstrual cycle, and so forth--has for them. A scale ranging in value from +2 (Very important) to -
2 (very unimportant) was created for this purpose. Thus, higher scores for an individual represented more positive evaluations and lower scores signified more negative evaluations. Furthermore, the items I through 4, and item 9 on the evaluation subscale-all representing negative consequences of using OCs--were reverse scored to capture the negative evaluations.
The belief component of the model (also known as subjective probability or
perceived likelihood) was then measured by having women rate how likely they thought it was that they would experience each of the outcomes if they chose to use OCs acquired OTC, and with a physician's prescription. Thus the subjects were asked to rate the likelihood of each consequence occurring on two sets of scales representing acquisition of OCs over-the-counter and with a physician's prescription respectively. On a 5-point scale created for this purpose, women rated their likelihood estimates ranging from very likely to very unlikely. For example, women were asked how very likely or very unlikely they thought they were to avoid unplanned pregnancy or gain access to OCs as emergency contraception pills etc. if they acquired OCs over-the-counter, or with a physician's prescription. The scale values ranged from +2 (very likely) to -2 (very unlikely) to determine the subjective probability of occurrence of each of the nine consequences of using OCs listed previously. More positive scores for a woman represented the higher likelihood that she thought the consequence in question would result.
As specified in the theory of reasoned action, the evaluation of a given
consequence was then multiplied by the estimate of the likelihood that using OCs acquired OTC would result in that outcome. The resulting scores were summed across
the consequences to provide a "belief X value" (Ibiei) score for acquiring OCs over-thecounter and with a physician's prescription. Thus, the OC-attitude score for each person was derived as a summated product of these terms.
General Attitude Toward Acquiring OCs Over-the-Counter (Direct Measure; A-act)
General attitude toward acquiring oral contraceptives OTC was measured by
having subjects indicate on 5-point bipolar scales, their favorability for obtaining them OTC. For example, the respondents were asked to indicate their opinion on a 5-point scale, ranging from extremely good (+2) to extremely bad (-2), and from extremely favorable (+2) to extremely unfavorable (-2), regarding their assessment of the appropriateness of acquiring OCs over-the-counter, with a neutral value anchored at zero. The experts in the literature (Cone and Foster, 1995; p. 169) have critiqued the use of single-item measures, especially when reliability of such measures is not available. Therefore, the attitude items were presented twice to obtain a more reliable measure of the construct. The general attitude toward a method was then calculated by averaging the two ratings. Higher scores represented more favorable attitudes toward acquiring oral contraceptives OTC.
Subjective Norm Regarding the Acquisition of OCs Over-the-Counter (Indirect Measure)
Consistent with the TRA, both direct and indirect measures for subjective norm were taken. The subjective norm is the perception that most people who are important to the individual thinks she should or should not perform the behavior in question (Fishbein and Ajzen, 1975 p.302). Further, it refers to the person's perception that important others
desire the performance or nonperformance of a specific behavior. According to the theory, the subjective norm was determined by the perceived expectations of specific referent individuals or groups, and by the person's motivation to comply with these expectations. The theory holds that, the more a woman perceived that others who are important to her thought she should buy her OCs over-the-counter, the stronger were her intentions to acquire them OTC.
Normative Beliefs and Motivation to Comply
Each respondent was asked to rate how she believed each of the people important to her would feel about her using OCs acquired OTC. Scales were created separately to measure a woman's perceived expectations of others regarding the acquisition of OCs over-the-counter and her motivation to comply with those expectations. Three items each constituted subscales for normative beliefs and motivation to comply. These items were written to measure a woman's perceptions of what four other individuals or groups important to her will think regarding her intention to acquire OCs OTC. On a 5-point scale women were asked to indicate how each individual or a group of individuals (i.e., family, close friends, and partner) would feel about her choosing to acquire oral contraceptives OTC. The scale ranged from very likely (+2) to very unlikely (+2) that a particular person would feel that she should use OCs acquired OTC. Similarly, motivation to comply with each referent regarding the acquisition of OCs was measured on a 5-point scale ranging from strongly agree (+2) to strongly disagree (-2) that she will comply with the perceived expectations of the important person. The scale values were anchored similarly as before. For each person rated, scores on a woman's motivation to
comply with each referent was then multiplied by her perceived expectations of that referent regarding the OC acquisition OTC. The scores for these three referents were then summed, providing a single score for the subjective norm regarding the acquisition of OCs over-the-counter.
Subjective Norm (Direct Measure)
This was a direct measure of subjective norm. Consistent with the
recommendation made by the proponents of TRA, the scale consisted of basically one item (Ajzen and Fishbein, 1980), measuring women's responses on a 5-point scale ranging from 'very likely' (+2) to' very unlikely' (-2), with scale values anchored similarly as before. For example, a person may perceive that most people who are important to her and whose opinion she values think that she should or should not perform the behavior in question. The scale developed for this purpose measured subjects' perceptions of what others think she should do. The rating on the single item was considered as a direct measure of subjective norm for OC acquisition OTC. More positive scores represented stronger normative influence on the person regarding the acquisition of OCs over-the-counter. A positive correlation of this measure with that obtained indirectly for subjective norms was to provide a validity check for the TRA model used in the study.
Intention to Acquire OCs Over-The-Counter
This was the response variable of interest in the study. A scenario depicting a
plausible situation for obtaining oral contraceptives OTC was built into different sections
of the questionnaire. This scenario described a new means for acquiring oral contraceptives, including the means that offered highest professional supervision (physician's prescription) and the means that offered least professional supervision (OTC). Subjects' response in intending to acquire OCs over-the-counter was then measured using a 5 -point scale, with the scale values ranging from very likely (+2) to very unlikely (-2) it is that the person would intend to obtain oral contraceptives OTC. Larger values for behavioral intention represented stronger intentions for acquiring OCs over-the-counter.
Demographic and Other Variables
Table. 4.1 provides a list of demographic variables that were of interest to the study and their operationalization. These variables provided a description of the study sample. Subjects' age was measured as a continuous variable. In addition, the information about subjects' race, religion, duration of current use of OCs in months, plans to use OCs sometime in the future, and the type of health professional seen for OCs was obtained at the categorical level. This information was obtained for descriptive purposes only. The current reality indicated that women obtained their OCs and related services from professionals other than physicians. Therefore, it was a matter of interest to this study to get an idea as to the type of professional supervision involved in the acquisition of OCs among those who had used them in the past. Variables describing the current and past OC use were measured on a dichotomous scale. Categorical variables measuring insurance coverage for prescription and nonprescription drugs and also for MD visits were included in the questionnaire. The past or current use of OCs, if any,
reflected the level of experience women had with this form of contraception. Further, it was also a matter of interest to examine if there was any difference in OC intentions for women who had used oral contraception in the past and those who had not.
Instrument Development and Validation
A pilot test of the instrument was carried out first. This was to assess the
reliability and validity of the instrument for further refinement before data collection. First, a panel of expert judges, comprising mainly the faculty of the Department of Pharmacy Health Care Administration, University of Florida, was requested to check the readability and face validity of the questionnaire. A group of graduate students and scholars at the department was also asked to review the scale for errors and, poorly worded, misleading or confusing items. Next, questionnaires were mailed to a sample of randomly selected female students enrolled in the University of Florida to identify any construction defects. Clarity and relevance of the items were assessed depending on the responses to different items in the questionnaire. This process was also expected to help identify problems in instructions, formatting, item wording and so on.
One of the primary objectives of this investigation was to develop and validate the OC-attitude scale. An exploratory factor analysis was conducted to verify previously identified domains on the scale pertaining to beliefs about risks and benefits of acquiring OCs, thus contributing to establishment of construct validity. Principal component analysis with varimax rotation was employed to look at the pattern of correlations among variables in a correlation matrix. Eigenvalues resulting from principal component analysis were identified to calculate communality estimates (proportion of multivariate
variance explained) of the orthogonal vectors. All factors achieving eigenvalues greater than 1.00 were retained. The resulting orthogonal factors were then subjected to varimax rotation to gain further insights. The goal here was to verify the existence of two subscales of behavioral beliefs (i.e., beliefs about risks and beliefs about benefits) which are conceptually reasonable and which have acceptable psychometric properties. The factor loadings of less than .3 0 were considered unimportant.
In addition, the internal consistency of the scales was calculated to estimate their reliability using Cronbach' s alpha. Mean item and total score on the 0 C-attitude subsccales were separately calculated. Partial alphas for each item were inspected and item-total correlations determined. An item analysis to determine item-remainder correlation was also performed. Items were considered for deletion if (i) the item to total correlation was less than .20, and (ii) the partial alpha indicated that the Cronbach's alpha would increase substantially if the item were deleted. Since our scale consisted of two previously identified dimensions, coefficient alpha for each dimension was calculated. Variability on responses and intercorrelation between items were also assessed. Items were retained in conceptually defined subscales depending on the correlation between items in that scale and also with items in other subscales. Thus, independent and internally consistent subscales were established for separate behavioral belief domains. The revised instrument was then used for the testing of hypotheses on the larger sample as detailed in the section following pilot study.
The readability and face validity of the questionnaire was initially established by using a panel of expert judges consisting of faculty and graduate students at the department of Pharmacy Health Care Administration and experts at the departments of Pharmacy Practice and Foundations of Education. Next, the questionnaire was pilot tested among female students registered for summer semester at the University of Florida to identify construction defects. The records at the Office of the Registrar indicated that there were 12450 female students registered for the 1998 summer semester. 1000 female students were selected from the listing provided by the University Registrar's Office using systematic sampling technique. For this purpose, every 12th student from the list was drawn to obtain a sample of 1000 students (Agresti and Finlay, 1986). Since the primary aim of the pretest was to aid the evaluation and revision of the instrument, no reminder postcards or follow-up telephone calls were used. The questionnaires were mailed out to students' residential addresses on July 23,1998,
The questionnaire was composed of an evaluation scale, a set of two likelihood scales (for physician and OTC sources respectively), a subjective norm scale, and items to measure intention to acquire OCs OTC and OCs in general. The evaluation scale sought to measure subjects' evaluation of the possible consequences of using oral contraceptives. As described in the previous section, the items included in the questionnaire represented beliefs about benefits and risks of using OCs. The two likelihood scales tested respondents' subjective probability (or perceived likelihood) of experiencing the listed consequences if OCs were acquired with a physician's prescription and OTC respectively. The subjective norm component of the instrument
consisted of normative belief and motivation to comply subscales. Furthermore, direct measures for both attitude toward the behavior and subjective norm were included in the questionnaire. In addition, items eliciting responses about subject demographics and OCrelated experiences were also included in the test.
In two weeks following the initial bailout, 25 1 questionnaires were returned,
constituting a response rate of 25 percent. Nine questionnaires were returned undelivered. Since the bulk mailing procedure was used in this step, there was no guarantee that the mail was forwarded to those students who had moved to a new address. The response rate for the pretest study was somewhat low probably due to the questionnaire being received during the final week of surnmer classes, and due to the bulk mailing service. The low response rate could also be attributed to the fact that there was no follow-up bailout of questionnaires or reminder cards for the pilot study. However, the sample obtained above was adequate for instrument validation. Excel 97 (SR-1) was used to set up a spreadsheet database of responses. SPSS (Release 7.0) and JMP IN (Version 3.02) statistical programs were used to run statistical analysis on the data imported from Excel.
Table 4.3 describes the pretest sample. The mean age of the sample was 24 years. Of the total sample, about 48% were 21 years or under, 42% were between the ages of 22 and 31. 50% of the sample reported the current use of OCs, while 78% of the sample reported having used OCs in the past. About 54% of those who used OCs used them for two years or more, 23% used them for more than a year, and about 19% had used them for only a few months. About 67% of those who are currently using OCs, or those who used them in the past, saw a physician first for OC prescription, while about 3 1 % saw nurse practitioners first for their OCs.
Table 4.3 Pretest Sample Characteristics
Sample Characteristics* Percent Mean/Std. dcv
Age 24 yr/6.7yr
African Amherican 7.1
Asian American 1.8
Possible OC use in the future
1 don't know 34.8
Insurance for MID visits
1 don't know 4.4
Insurance for Rx and NonRx drugs Yes 68.0
1 don't know 5.7
Intention to acquire OCs OTC Mean ---0.03/1.6
Very likely (intenders) 26.9
Somewhat likely (intenders) 18.1
Somewhat unlikely (non-intenders) 19.4
Very Unlike ly(non-intenders) 27.8
Intention to use OCs
Very likely 63.0
Somewhat likely 17.2
Somewhat unlikely 4.4
Very Unlikely 8.8
The instrument validation was achieved by first performing the factor analysis
and then by the estimation of internal consistency. The confirmation of the presence of
two previously conceptualized domains, namely, beliefs about benefits and beliefs about risks, within the evaluation and likelihood subscales was the reason for performing factor analysis first. The factor analysis was also performed to establish construct validity of the theory of reasoned action by confirming the presence of two constructs, namely, attitude toward the behavior and subjective norm, conceptualized by the model. Following factor analysis, reliability estimation for each of the subscales previously identified by the factor analysis was done. The section below describes the analytical procedures used to validate the instrument.
Primarily, construct validity of the study instrument was established by two methods: 1) Factor Analysis, and 2) Correlation statistics. Construct validity of the instrument is determined by examining the extent to which the measures correlate with other measures designed to measure the same thing, and whether these measures behave as expected (Churchill, 1979). If the proposed relationships among the TRA model constructs are confirmed as predicted by the theory, both the constructs and the instrument that measure them are said to be valid (Crocker and Algina, 1986). With regard to correlational studies, Crocker and Algina (1986) state that there are no generally recognized guidelines for what constitutes adequate evidence of construct validation. However, correlations computed between measures that are appropriate and conceptually reasonable served as evidence of construct validity of the study instrument. For example, significant correlations between directly measured subjective norm and attitudes, and
their indirectly measured counterparts expected to provide evidence of construct validation.
Factor analysis was one of the procedures used in this analysis for construct validation of the instrument. The factor extraction procedure employed principal component analysis for initial unrotated factor solution. Then an orthogonal rotation procedure (Varimax) was used to examine rotated, uncorrelated factors. Factor extraction procedures were employed on data obtained for all the variables representing the TRA model constructs. Specifically, the summated, multiplied products of evaluation and likelihood items (i.e., Xei bi ) represented the attitude toward the behavior variable. These items were expected to reveal the presence of two subsets of items representing beliefs about benefits and beliefs about risks. Next, the items representing indirectly computed subjective norm (i.e., InbXmc), obtained by multiplying normative belief and motivation scores, were also included in the analysis. The expectation here was that the subjective norm component in the factor analysis would account for the extraction of a separate factor, representing three sources of influence--family, partner, and friends. Results of the analysis are summarized in the following section.
Following the rule of thumb of at least 5-10 subjects per item (Cone and Foster, 1995; Gorusch, 1983; Nunnally, 1967), the sample size of 251 obtained was considered adequate to perform factor analysis. Table 4.4 presents a 3-factor solution, describing factor loadings for all the items representing the constructs from the TRA. The pattern exhibited by the factor loadings was as predicted by the theory. The analysis procedure
extracted three factors having eigenvalues of 2.3, 1.9, and 1.6 respectively, while the rest of the factors recorded eigenvalues of less than 1. Furthermore, the factor structure derived from the analysis confirmed the presence of two domains within the attitude component of the model (beliefs about benefits and beliefs about risks), as conceptualized by the investigator. The Scree plot further confirmed the existence of three prominent factors within the TRA instrument.
All the items on the 'beliefs about benefits' subscale (convenience, avoidance of unplanned pregnancy, regulation of menstrual cycle, and emergency contraception) loaded on the first factor. Each of these items had a factor loading of 0.5 or higher. Similarly, all the items on 'beliefs about risks' subscale (nausea, breast cancer, weight gain, and pain in the legs) loaded on Factor 2, with the exception of 'Personal expenditure' item.
Table 4.4 Rotated Factor Matrix for the TRA Constructs FACTOR 1 FACTOR 2 FACTOR 3
Nausea .61385 -.08133 .16549
Wgt Gain .70862 -.16204 .00534
Leg Pain .81241 -.04047 .02081
Br Cancer .72666 .16521 -.02088
Pregnancy .00021 .72025 -. 10991
Mens Cycle -.14858 .71711 .18198
Morng-aft-pil .12342 .53696 -.12751
Convenience -.12894 .66317 -.00381
Expenditure .01066 -.41023 -.06166
NBXMC Emly .19801 .02813 .80271
NBXMCFind -.09522 -.10072 .79779
NBXMCPrtnr .05896 .05874 .58664
Factor 1: Beliefs about risks (individual eXb), Factor 2: Beliefs about benefits (individual eXb), Factor 3: Subjective norm (individual nXb).
Further examination of factor loadings indicated that this item loaded heavily on a four-th factor by itself (0.88), with the factor eigen value of 0.9. It appears that 'personal expenditure' works as a separate factor. Furthermore, commonality estimates for this item and other items in both the subscales did not suggest a need for removal of any item.
In sum, the factor structure above confirmed the presence of two domains within the 'attitude toward acquiring OCs OTC' component. It also empirically verified the presence of the other dimension conceptualized by the TRA-subjective norm. Almost all the items loaded strongly (factor loadings of 0.6 or greater) on the corresponding factors. The item involving 'personal expenditure' appeared to be somewhat problematic. A separate, independent factor was found to correlate highly with this item, however, with a factor eigenvalue of less than 1. This further suggested that 'personal expenditure' item worked as a separate factor by itself.
As a further evidence of construct validity, Pearson correlation coefficient was computed for a bivariate association between a) directly and indirectly measured subjective norm variables, and b) directly and indirectly measured attitude variables. A correlation of 0. 130 that is significant at the 0.05 level was obtained for the former, while a significant correlation of 0.385 at 0.01 level was obtained for the latter. This further confirmed both the construct and predictive validity of the TRA, as proposed by the model proponents.
Following the confirmation of the conceptualized domains within the TRA instrument, the internal consistency for each subscale was then determined. First,
Cronbach's alpha was computed for two subscales within the evaluation component of the instrument. Next, a similar analysis was conducted for the likelihood (OTC) scale and subjective norm (indirect measure) component of the scale. The consistency with
which the subjects performed across items measuring the same construct was obtained by calculating coefficient alpha. A positive contribution to internal consistency of the subseale was one of the criteria considered for retaining the items in the analysis. The evaluation component of the attitude scale consisted of two subscales: I) beliefs about risks and, 11) beliefs about benefits of using oral contraceptives. The internal consistency for both subscales of the evaluation scale is presented in Table 4.5 and Table 4.6.
Table 4.5 Item-Total Statistics (Evaluation of Benefits Subscale) Benefit items Corrected item-total Alpha if item
Convenience .5503 .5063
Emergency Contraception .3557 .6563
Regulation of menstr cycle .4283 .5844
Avoidance of pregnancy .4470 .5854
Alpha = 6501
Table 4.6 Item-Total Statistics (Evaluation of Risks Subscale) Risk items Corrected item-total Alpha if item
Breast Cancer .3350 .5935
Personal Expenditure .1153 .695 8
Nausea .4249 .5482
Pain in the legs .5168 .5001
Weight Gain .5446 .4803
Alpha = .6260
Note: These items were reverse scored to reflect beliefs about risky consequences of using- OCs
Table 4.7 Item-Total Statistics (With 'Personal Expenditure' Item Removed) Risk items Corrected item-total Alpha if item
Breast Cancer .3359 .7 134
Nausea .4722 .6410
Pain in the legs .5724 .5750
Weight Gain .5568 .5835
Alpha = .6978
Table 4.8 Item-total Statistics (Normative Belief Subscale) Sources of influence Corrected item-total Alpha if item
(normative belief) correlation deleted
Family .8265 .8398
Partner .8140 .8509
Friends .7739 .8829
Alpha = .9008
Table 4.9 Item-Total Statistics (Motivation to Comply Subscale) Sources of influence Corrected item-total Alpha if item
(motivation to comply) correlation deleted
Family .5266 .5265
Partner .4323 .6494
Friends .5107 .5554
Alpha = .6743
A similar analysis was also performed on the likelihood (OTC) subscale of the instrument. Tables 4.8 and 4.9 report the internal consistency for normative belief and motivation subscales of the subjective norm component of the instrument. An acceptable level of alpha of 0.50 to 0.60 (Nunnally, 1967) was used for retaining or eliminating the items. Given these cut-offs, none of the subscales seemed too problematic. Table 4. 10 summarizes the information on internal consistency for different subscales of the study instrument. The alpha for all the subscales were found to be greater than or equal to 0.60.
To further determine instrument reliability, two parameters were examined
through item analysis. First, item-to-total correlation was examined. Next, inter-item correlation for each subscale was examined to validate the results of item analysis. Item analysis a computation and examination of any statistical property of respondents' response to an individual item (Crocker and Algina, 1986) was employed to identify the final set of items for each dimension of the instrument. The parameter employed for this purpose is called item reliability index.
Table 4. 10 Reliability Estimates of Pretest Subscales Subscale Number of Items Reliability
Beliefs about benefits
Beliefs about risks 4 0.65
Beliefs about risks (minus 'personal 5 0.62
expenditure') 4 0.69
Beliefs about benefits
Beliefs about risks 4 0.62
Beliefs about risks (minus 'personal 5 0.74
expenditure') 4 0.82
Subjective norm component
Motivation to comply 3 0.90
According to Crocker and Algina (1986, p.320), item reliability index is the
parameter that should be used in lieu of the simple correlation between item and criterion because the item variance actually weights the relative contribution of a particular item to overall test score reliability. Consequently, the correlation between item score and the total score was examined in this analysis. The corrected item-to-total correlation was calculated to correct the spurious value that results when item score contributes to the total score. Thus, it is the correlation between an item score and the total of the
remaining items. Some researchers (Zaikowski, 1985; Shimp and Sharma, 1987) advocate a corrected item-to-total correlation of 0.50 or greater as a rule of thumb for retaining an item in its subscale. For the benefit subscale of the evaluation component of the questionnaire, three items correlated less than 0.50 with the assigned subscale (Table
4.5). Three items of the risk subscale of the component had correlations less than 0.50 with the assigned subscale (Table 4.6). However, normative belief subscale of the instrument had all the item-to-total correlations of greater than 0.50 (Table 4.8), while the motivation subscale had only one item that correlated less than 0.50 with the assigned subscale (Table 4.9). Each of these subscales is described in detail below.
Beliefs about benefits subscale. Table 4.5 shows that three items had correlations
under 0.5. These items had correlations of 0.35, 0.42, and 0.44 respectively.
These items were not considered problematic, however, for two reasons. First,
removal of these items would have lowered the reliability considerably. Second,
the measured correlations were not sufficiently low to render them useless.
Instead, all the three items were found to be moderately correlated with the total score. However, the second item (emergency contraception), having the lowest
correlation, seemed to be in need of revision. It appears that the concept of
morning-after-pill as emergency contraception is perhaps not well understood by the respondents. In addition, deletion of this item from the subscale did not seem
to influence reliability estimates much. However, revision of this item, rather than deletion, seemed to be the more appropriate step in instrument validation.
Furthermore, this step seems necessary given the importance this item has for OC
use OTC and for its practical implications. It is expected that all these items would perform better in the main study involving a larger sample. Beliefs about risks subscale. Of the three items that had less than 0.50 correlations, only item 2 (personal expenditure) had a poor correlation with the total (0. 11). This was consistent with our previous findings about the item performance during the factor analysis. The lower correlation could be attributed to the possibility that the item personal expenditure' worked as a separate factor. It did not seem to belong to the risk subscale. As a further check, the variance of this item in relation to other items in the subscale was also examined. From the variance stand point, the item did not look problematic. The item variance was found to be comparable to all the other items in the subscale. Motivation to comply subscale. Only one item (Partner) had a correlation of less than 0.50. However, the correlation of 0.43 for this item is acceptable for three reasons. First, the correlation of 0.4 is sufficiently high to justify its retention in the subscale. Second, this item is expected to perform better with a larger sample. Third, it is logical to believe that major decisions involving contraception receive substantial input from partners, whereas inputs from friends and parents may not receive much weight. This is particularly so in the case of OC acquisition, where input from a partner might have received considerable weight. For this reason, it can also be argued that the perceptions about what a partner might think about acquiring OCs OTC--and hence the motivation to comply with partner's wishes-may be somewhat skewed, thereby reducing item variance. Such perceptions may also be in conflict with similar perceptions regarding what friends and family
might think about buying OCs over-the-counter. A closer examination of the
response to this item also revealed that almost all the women in our sample
responded to this question (i.e., 99.9% item response), indicating that all the
women who responded to the questionnaire did in fact have a partner who they
could think of at the time of survey. Thus, this item was retained in the
Inter-item correlations for each subscale revealed that almost all the inter-item correlations were greater than 0.30. As a further evidence of internal consistency, correlation coefficients were computed for the two items that sought to directly measure attitude toward acquiring OCs OTC (section VI, items 2 and 3 on the instrument). A twoitem measure of attitude toward acquiring OCs OTC was incorporated in the study instrument, with the mean of both items being used for data analysis. These items were found to correlate very highly (0.918) with each other at the 0.01 level.
Overall, the study instrument was found to be reliable and valid. Only two items were found to be problematic.'Personal expenditure' and 'emergency contraception' items appeared to be in need of further examination. In experts' opinion, further revision of these items was not necessary because it was believed that any revision would result in little change in the factor structure of the instrument. A substantial item-to-total correlation (0.3557) and a reasonably strong factor loading (0.54) for the 'emergency contraception' item indicated that this item did not perform poorly. While the removal of the item on 'emergency contraception' from the benefit subscale would not have enhanced
the reliability considerably, the same was not the case with the 'personal expenditure itern.' In fact, the results showed that the removal of this item from the risk subscale resulted in better reliability for the risk subscale (Table 4.7). Therefore, the consensus among the experts regarding the 'personal expenditure' item was that it did not belong to the risk subscale. Given the importance this item had for the OC acquisition OTC, the item was retained in the instrument, with the understanding that it worked as a separate factor by itself. The rest of the items did not seem to need any revision. A larger sample to be employed in the main study may in fact result in better item performance. Therefore, no further changes were made to the instrument. The main study and hypotheses testing were then conducted with the same instrument.
The list provided by the Office of the Registrar at the University of Florida served as the sampling frame for the main study. A list of all the female students registered for the Fall 1998 semester was obtained. The actual sample size needed to test the research hypotheses was determined using three different methods. The strategy to analyze the data basically involved the use of multiple regression analysis. According to one estimate (Green, 1991), for studies involving multiple regressions, the minimum sample size could be determined as follows. The minimum sample size is estimated as a function of the effect size as well as the number of predictors to be used in the prediction of the dependent variable (Green, 199 1). Based on this methodology, the minimum sample size
(N) required to evaluate multiple correlation coefficients is = L/172, with a power of 0.80 (alpha-.05).