-G‘ . . . . .H . ... ,.....7H.__.w.w...._..r...!ur . . .. ., . 432%...» inhabit.” . . . . . u ... .. ...ww...a.m......wm.u._.¢.w..nw. . . . ,. . ‘ , wwwfifimfim .. _w...”,O/Jvurb.o.‘s..3.! ‘ . . ._ ._ ........,.».,§.L.(..h.£y . . . . ..Jdn.f....\nr.owu.r.umfi..m . _H .. .. . V .. ...‘.M.H..w_..,..m-..n‘...n....n . . . _ _ . . _. ..._........_........___...m~_w,firm}.34........mdmwm _ .. . _ . . . _. , ... ..3....“an”...Aw.._mw.......m;n.q“may. .an L... ... .. .... .. .. . . _ . . . . . . A . . . ._ . . . ‘ .. . . _ . Anny... I . . . . . _. R I . . . . . _ ... .. u M IR . . . . m . _ P f m . . . . . . . _ . . M 17.: ....3 ,4 .. ..qmn.fi....r..., 4.“.._r..»:u:,_..¢... E 0. w _ . .. 1., ”In .. ... . m! J L o)!..f.tl’.~z.u»$‘9.u m m e u . . . .. ._ 0 0 D . n u \ .. N E r... . N.. ..m . y . mw N N S .IH , . . A I III . . R N M.» m U .. w m . ‘ U . . 0 . v . .. . . . ._ . . ... .. .. . 4 . ......E .. ......'L¢-.. r»..l..vh.U ”...”..On 0 r. .o . ‘ 1 .Jd {if—"- m was ea 5%; ABSTRACT COMPARISON OF VOLUNTEER, PARTIAL VOLUNTEER AND NONVOLUNTEERS SUBJECTS FOR A FAMILY PLANNING EDUCATIONAL PROGRAM by Susan Christine Hedrick The literature findings of significant differences be- tween volunteer and nonvolunteer subjects on many demographic, personality, and other characteristics and on the threats these differences pose for the validity of inferred causal relation- ships and the generalizability of research findings were re- viewed. Possible limitations on the validity of experimental evaluations of the effectiveness of educational programs in the family planning area using volunteer subjects,.and on the utility of voluntary educational programs in making changes of any magnitude in societal family planning attitudes and behavior were specific concerns raised. In the present study, all women under 25 Years of age obtaining marriage licenses in the county during a certain time period were attempted to be contacted. Sixty % of these women, or 199 subjects were successfully contacted and given an inter- viewer-administered scale measuring variables of especial interest in the family planning area: interview process, demo- graphics, reproductive history, family planning attitudes and expectancies, knowledge and practice of birth control methods, and decision making influences and process. At the end of the Susan Christine Hedrick scale the Family Life Service Project, an educational program, was briefly described and the subjects asked if they would like to participate. Three groups were identified: the volunteers, the 110 women who answered'"yes"; the partial volunteers, the 39 women who answered "maybe"; and the non- vglunteers, the 50 women who answered "no." For the total comparative results, out of 104 total tests of significance, only fifteen reached acceptable signi- ficance levels. Different interviewers obtained significantly different rates of volunteering. Volunteers were found to have a lower income, have a husband with a higher educational level, be more likely to be planning to adopt a child, have a lower degree of familiarity/use of the IUD, obtain Pap smears more frequently, and finally to be more influenced on either their decisions regarding use of birth control or choice of a family size by the following sources of communication: physician, high school, hospital education program, family planning clinic, and television. These isolated findings do have some possible implications, however, the broad patterns of differences hypo- thesized were not found in the present study. The concept was postulated of multiple levels of volun— teering, including: those subjects who have such temporary residences that they cannot be located for even an initial contact, those who are contacted but refuse any data collection procedures, the "verbal" levels of volunteering compared in the present study, and the "behavioral" volunteers who fully participate Susan Christine Hedrick in the research or program. It was hypothesized that, given the existence of groups in the society with real differences in family planning attitudes and behavior, that the breaks be- tween these groups could come at any point along this volun- teer continuum. It was further suggested that the complex motivation underlying volunteering behavior could operate so that, for instance, the behavioral nonvolunteers could be com- posed of both those individuals who were most resistant to any personal or social change and those individuals who were so much the "already converted" that they felt no need for any further outside agency contact. The task of those attempting to evaluate educational program impact, and those attempting to investigate the phenomenon of volunteering behavior is thus tremendously complicated and important. COMPARISON OF VOLUNTEERS, PARTIAL VOLUNTEERS AND NONVOLUNTEERS FOR A FAMILY PLANNING EDUCATIONAL PROGRAM By Susan Christine Hedrick A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1975 To my parents 11 ACKNOWLEDGEMENTS For their patience and help in this long-term endeavor I would like to thank my committee: Dr. Louis G. Tornatzky, Dr. George W. Fairweather, Dr. Lawrence I. O'Kelly and Dr. Raymond Frankmann. Special appreciation goes to Lou Tornatzky, as my advisor, for his unflagging and good-humored inspiration, support and guidance. I would like to specific- ally acknowledge all the faculty and students, past and present, of the Ecological Psychology Program, who, under Dr. Fairweather's energetic inspiration, with their varied talents and friendship, make the work of learning experimental social innovation poss- ible, rewarding and even enjoyable. I would finally like to acknowledge that part of the financial support for this project came from Grant GZ-2179 as a part of the Special Projects in Graduate Education Program of the National Science Foundation, and from a Michigan State University Bio-Medical Sciences Support Grant. iii LIST OF LIST OF CHAPTER I. TABLE OF CONTENTS TABLES . FIGURES . INTRODUCTION . General Information Differentiating Characteristics Studies of Special Relevance to the Present Study . ' * Situational Determinants of Volunteering . Threats to Generalizability and Inferred Causal Relationships Implications for Present Study . Validity of inferred causal relationships . . Self—selection of audience Limits to the Application of the Literature to the Present Study . . . . . . . Hypotheses . Interview process . Demographics Reproductive history Family planning attitudes and expectancies. Knowledge and practice of birth control . iv . 13 . l6 . 21 . 22 . 23 . 24 . 26 . 26 . 28 . 31 32 . 34 TABLE OF CONTENTS (Continued . . .) CHAPTER Decision making influences and process . II. METHODS AND PROCEDURES . Subjects . Design . Procedures Measures . Interview process Demographics . Reproductive history . Family planning attitudes and expectancies . Knowledge and practice of birth control Decision making influences and process Volunteering measurement . III. RESULTS . Comparative Results . Interview process Demographics . Reproductive history . Family planning attitudes and expectancies . Knowledge and practice of birth control. Decision making influences and process Associative Results 1. Volunteering 2. Reproductive history V 52 52 54 62 62 71 83 89 93 98 TABLE OF CONTENTS (Continued ...) CHAPTER Pagg 3. Attitudes-expectancies ......................... 98 4. Religion and its correlates .................... 99 5. Sources Of influence......... .................. 99 IV. DISCUSSION .............................................. lOl Interview Process ....................................... 102 Demographics ............................................ l03 Reproductive History -- Family Planning Attitudes and Expectancies ...................................... l04 Knowledge and Practice Of Birth Control ................. 105 Decision Making Influences and Process .................. 106 Implications ............................................ 109 APPENDICES ..................................................... ll4 BIBLIOGRAPHY .................................................... 123 vi TABLE 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES Comparison of Volunteer Status Groups for High Active and Low Active Interviewers . Comparison of Volunteer Status Groups for Most Active Interviewers . Means and Analysis of Variance - Number of Minutes Spent in the Interview . . Chi-Squares - Demographic Variables . Means and Analyses of Variance - Demographic Variables . . . . . . . . . . . . . . Means and Analyses of Variance - Reproductive History Variables . . . . . . . . . . . . . Current Pregnancy Status Means and Analyses of Variance - Attitude Toward Birth Control . . . . . . . . . . . . Means and Analyses of Variance - Family Planning Expectancy Items . . . . . . . . . . . . . . Means and Analyses of Variance - Attitudes Regarding Ideal Family Planning Behavior Means and Analyses of Variance - Other Attitudinal Items . Adoption Plans . Knowledge of Time of Ovulation Chi- -Squares - Familiarity/Use of Birth Control Methods . . . . . . . . . . . . . . Means and Analyses of Variance - Familiarity/Use of Birth Control Methods . . Chi-Squares - Practice of Birth Control . vii 55 56 57 60 63 64 65 67 68 7O 72 73 75 77 80 LIST OF TABLES (Continued . . .) TABLE 17. 18. 19. 20. 21. 22. 23. 24. Means and Analyses of Variance - Practice of Birth Control and Related Behavior . Familiarity and Use of Family Planning Clinic . Decision Making Process and Influence Sources Within the Family . . . . . . . . Other Interpersonal Decision Making Influences. Group-educational or Institutional Sources of Influence . Media Sources of Influence Cluster AnalySis . . . . . . . . Correlation of Raw Cluster Scores With Each Other . viii . 82 . 84 . 85 , 86 , 88 , 9O , 94 97 FIGURE 1. LIST OF FIGURES Design Used in the Study . ix CHAPTER I INTRODUCTION An issue of great and increasing recent importance in psychology is that of artifactual bias in behavioral research. A great deal of the concern has centered around the possible non-representativeness of subjects used in the vast majority of research studies due to their volunteer status. This non- representativeness has been seen as posing threats to the valid- ity of inferred causal relationships and to the generalizability of the research findings. The growing concern over the ethical issues of right to privacy and informed volitional consent for research subjects should continue to increase the interest in this topic. The ethical guidelines adapted by the American Psychological Association urge that potential subjects be informed of "all features of the research that reasonably might be expected to influence willingness to participate, and. . . all other aspects of the research about which the participant inquires" (American Psychological Association, 1973, p. 79). One warning was given that "a science of informed and organized volunteer sophomores might be on the horizon" (Rosenthal and Rosnow, 1975, p. 127). Accentuating the importance of the study of the characteristics of the volunteer is the fact that in the near future not only all research subjects but all participants in service programs 1 2 of all types may be involved only by their own completely in- formed volitional consent. The work in the area of volunteering behavior has been summarized at various stages by Robert Rosenthal and Ralph Rosnow (Rosenthal, 1965; Rosenthal and Rosnow, 1969, 1975; Rosnow and Rosenthal, 1970). Their latest effort (Rosenthal and Rosnow, 1975) presents l. ‘ggmg general information on the phenomenon, the data concerning 2. characteristics differentiating volunteers from nonvolunteers, data detailing the 3. situational determinants of volunteering and findings regarding possible 4. threats Eg‘ggneraliZability and inferred causal relationships from this artifact. Our review of Rosenthal and Rosnow's work will be organized by these categories. General Information That volunteer status is an important sOurce of res- ponse bias is attested to first by the great variation in res- ponse, reported as ranging from 10% to 100% at one university, for example (Rosenthal and Rosnow, 1975). In one study when a recruiter extending the same invitation to participate in the same experiment solicited female volunteers from different floors of the same dormitory he received response rates vary- ing from.74 to 26% (Marmer, 1967). Rosenthal and Rosnow also present several studies that indicate that volunteering is a reliable act. When individuals were asked to volunteer a second time, either for the same or a different experiment, the reliability coefficients ranged 3 from..97 to .22, with a median of .52. For a study by Martin and Marcuse of sexual hehavior (1958) (the topic most closely related of the studies presented to family planning, the topic of the present study) the reliability was .67. Differentiating characteristiCs To summarize their review of the literature on the characteristics differentiating volunteers from nonvolunteers Rosenthal and Rosnow have developed a list of 22 "conclusions" concerning the association between volunteering and the characteristics. This list of conclusions is ordered by the degree of confidence in the association of the characteristic and volunteering. The definition of the degree of confidence involved "an arbitrary, complex multiple cutoff procedure in which a conclusion was felt to be more warranted when (1) it was based on a larger number of studies, (2) a larger percent— age of the total number of relevant studies significantly favored the conclusion and (3) a larger percentage of just those studies showing a significant relationship favored the conclusion drawn" (Rosenthal and Rosnow, 1975, p. 87). Four groups of characteristics were given, those with: maximum confidence (a minimum.of 19 total studies and 86% of significant studies favoring the conclusion), considerable confidence (17 total studies and 73%), some confidence (3 studies and 61%) and mini- mum.confidence (3 studies and 50%). This list of conclusions will be presented with minor abbreviations as the clearest statement of the literature 4 findings. Findings of particular relevance to the present study, i.e. female non-college population survey data, commun- ity based studies, or topics related to family planning, will be distinguished in two ways. First, any elaborations or clarifications of Rosenthal and Rosnow's conclusions by the present author will be presented in brackets. Secondly, immediately following the quoted list of conclusions, the studies with special relevance to the present study will be separately discussed. Rosenthal and Rosnow's list of con- clusions (1975, pp. 88-90) is as follows: ' Conclusions Wa'r’r'a‘n’t‘in'g‘ Maximum Confidence l. Volunteers tend to be better educated than nonvolun- teer.... l’when there is no personal contact between investigatErs and respondent, e.g., mailed question- naire, the magnitude of this relationship is somewhat reduced. Most of the work on this variable has been done in the area of survey research, because of the small degree of variance in educational level in the typical undergraduate population used in laboratory experiments.;T 2. Volunteers tend to have higher social-class status than nonvolunteers.... ‘l—when the task is "quite risky in a biological or psthological" sense (i.e., personality, hypnosis, and sleep research) this finding may be reversed. There may also be a tendency for females to show less of this relationship than males, although this could be confounded by the hazardous nature of the tasks for the four studies involving females only. 7 3. Volunteers tend to be more intelligent than non- volunteers when volunteering is for research in general, but not when volunteering is for somewhat less typical types of research such as hypnosis, sensory isolation, sex research, and small-group and personality research. 4. Volunteers tend to be higher in need for social approval than nonvolunteers. 5. Volunteers tend to be more sociable than nonvolunteers. 5 [in questionnaire studies, this finding seems stronger for non-college samples] 'ConcluSionS'warranting;considerab1e Confidence 6. 10. ll. Volunteers tend to be more arousal-seeking than non- volunteers, especially when volunteering is for studies of stress, sensory isolation, and hypnosis. . Volunteers tend to be more unconventional than non- volunteers, especially when volunteering is for studies of sex behavior. [Conventionality in these studies was defined in terms of sexual behavior, attitudes or views of sex roles] . Females are more likely than males to volunteer for research in general, but less likely than males to volunteer for physically and emotionally stressful research (e.g., electric shock, high temperature, sensory deprivation, interviews about sex behavior.) Volunteers tend to be less authoritarian than non- volunteers. CThere is some indication that this is more true for females than males; Jews are more likely to volunteer than Protestants, and Protestants are more likely to volunteer than Catholics. Volunteers tend to be less conforming than nonvolunteers when volunteering is for research in general, but not when subjects are female and the task is relatively "clinical" (e.g. hypnosis, sleep, or counseling research.) ' CO'n'clU'S'i’Oh'S' War-ranting 'S'O'IIIe‘ Confidence . 12. 13. 14. 15. Volunteers tend to be from.smaller towns than non- volunteers, especially when volunteering is for ques- tionnaire studies. Volunteers tend to be more interested in religion than non-volunteers, especially when volunteering is for questionnaire studies. Volunteers tend to be more altruistic than nonvolunteers. Volunteers tend to be more self-disclosing than non- volunteers. 16. 17. 6 Volunteers tend to be more maladjusted than.non- volunteers especially when volunteering is for potentially unusual situations (e.g., drugs, hypnosis, high temperature, or vaguely described experiments) or for medical research primarily psychiatric employing clinical, rather than psycho- metric, definitions of psychopathology. [When recruitment was for studies of sensitivity training or sex behavior, volunteers were found to be better adjusted, however] Volunteers tend to be younger than nonvolunteers... especially if they are female. (This finding may be somewhat less true for survey research and could also be a confounding with the higher status of older subjects in some of the studies] C'On'CIUS‘i'Oh‘S' Wa'r‘r‘a‘n’t‘i‘n‘g‘ Minimum ‘C‘on‘f'i’d'ence 18. 19. 20. 21. 22. Volunteers tend to be higher in need for achievement than nonvolunteers, especially among American samples. Volunteers are more likely to be married than non- volunteers.... Firstborns are more likely than laterborns to volun- teer... O Volunteers tend to be more anxious than nonvolunteers, especially when volunteering is for standard, non- stressful tasks and especially if they are college students. [The two studies of sexual behavior having findings on this variable found no significance differ- ences, however] Volunteers tend to be more extraverted than non- volunteers when interaction with others is required by the nature of the research. Studies of special relevance to the present study. Many studies have been done of the possible bias of volunteer samples in research on sexual behavior. The Kinsey survey research findings (Kinsey, Pomeroy, and Martin, 1948; Kinsey, Pomeroy, Martin and Gebhard, 1953) were the impetus for the first of these studies. Maslow (1942) had earlier found that 7 in a female sample his measure of self-esteem correlated .71 with unconventional sexual attitudes and -.61 with virginity. He found in a study designed with Kinsey that female volunteers for a Kinsey survey at his college‘wggthigher in self-esteem (Maslow and Sakoda, 1952). Siegman (1956) in a study of both male and female subjects found no differences in self—esteem between volunteers and nonvolunteers, however. No differences in anxiety, rigidity or defensiveness were found. He did find volunteers to be significantly more unconventional, however. For instance, 92% of the volunteers as opposed to 42% of the non-volunteers advocated "sexual freedom” for women. Martin and Marcuse (1958) in a study of college students who signed up for a study of sexual attitudes found no differ- ences in anxiety, sociability, ethnocentrism, dominance, authoritarianism, or intelligence though there were differences on most of these variables between volunteers and nonvolunteers for studies of personality or hypnosis. Diamant (1970) found college-age male volunteers to report more sexual intercourse and to have more permissive attitudes than nonvolunteers. Kaats and Davis (1971) compared three groups of intro- ductory psychology students: the control where 98% of the class filled out a questionnaire during their class session; the volunteer completion group where a class was instructed to begin the questionnaire in class, but, because of lack of time, to finish filling it out at home and return it the following day (62.5% did); and the volunteer selection group where 8 students had an opportunity to sign up for the study as a part of the research credit for the class, the usual procedure for obtaining undergraduate subjects. There were no differences on any variable between the control and volunteer completion groups. The volunteer selection group, however, as compared to the control group had more permissive attitudes regarding both male and female sexual behavior, more frequent dating, although younger were more committed in their current dating relationship, and had more non-coital sexual experience. However, all three groups, it was found, had very similar non-virginity rates, the main variable of interest in this study. Kaats and Davis report that the variables with signi- ficant differences above are usually correlated with and can be thought of as plausible predictors of deciding to engage in intercourse. They hypothesized therefore that the main motivation for the subjects volunteering to participate in the study was to obtain additional information to help them make that decision (p. 32). Barker and Perlman (1972) using Jackson's Personality Research Form."play" scale found volunteers to be significant- ly more arousal-seeking. In a study not reviewed by Rosenthal and Rosnow, Bauman (1973) compared volunteers: students who when contacted by letter came to a study to complete a ques- tionnaire of sexual knowledge, attitudes, and behavior (20% of the total females and 15% of the total males contacted) and a "main sample": a random sample from the same population 9 where the students were visited in their dorms and reportedly because of special procedures assuring their anonymity all completed questionnaires. When a mean sexual knowledge score was used, the male volunteer sample was more knowledgeable than the main sample. When the percent distribution on the sexual knowledge test was considered, a chi-square procedure found female volunteers to be lower in sexual knowledge. There were no significant differences in attitude, virginity, number of sexual partners, number of coital events, or contraceptive practice (Bauman, 1973, p. 29). For independent variables, volunteer men were less likely to be religiously devout, volunteer women dated less frequently, and were less likely to be dating one person exclusively. There were no differences in self-judged sexual knowledge. Bauman further analyzed the data by comparing the correla- tions between subject (or demographic) characteristics and sexual knowledge, attitudes, and practice for the volunteer group with those same correlations for the main sample. When the p <.05 level was used, different decisions were made (significant for one group but not in the other) for about a third (7 out of 23) possible comparisons for the men, and half of the comparisons for the women. For instance, religious devoutness correlated significantly with permissive attitudes for the volunteer but not the main sample. No direction was given for the relationship. There were several further studies which were related to the tOpics or methodology of the present study. The first 10 studied child-rearing attitudes in a non-student sample (Cohler, WOolsey, Weiss and Grunebaum, 1968). Mothers of one or more children under the age of five answered a newspaper ad and filled out a maternal attitude scale and completed some demo- graphic items. Then their "cooperation was requested for a subsequent study to be held approximately one month later" (Cohler et al., 1968, p. 605). Sixty-four percent of the women attended the second session. These revolunteers or second- level volunteers were then compared with the original volun- teers. This study is especially relevant to the present study not only because of the topic related to family planning but because of this method of assessing volunteer bias. With non- student populations it is usually not feasible to obtain a captive population for baseline data to use in later comparing volunteers and nonvolunteers. Cohler et al. see this volunteer - revolunteer procedure as a feasible alternative. "Since these revolunteer subjects are familiar with the demand characteristics of the study . . . they seek further partici- pation entirely on the basis of their own assessment of the situation, being in possession of information which is some- what similar to that of the volunteer subjects in the captive subject design" (Cohler et al., 1968, p. 604). There were no differences between the volunteers and re- volunteers on wife's education, husband's occupation, or Hollingshead's index of social position. The revolunteers did have a significantly greater number of children. There 11 were significant differences on three out of thirteen of the scales on the maternal attitude questionnaire and on two out of three of the scale-based factors. The volunteers, as come pared to the revolunteers, had "greater feelings of fear and helplessness about labor and delivery and greater uncertainty and discomfort in learning to meet the infant's needs. These mothers also felt less responsive to their infants and found the mother-infant relationship to be less enjoyable" (Cohler, et al., 1968, p. 610). In addition, they had less awareness of the child's bodily concerns, e.g. felt that children should not be curious about the body and how it functions; and they had more stereotyped sex-role attitudes. It would seem, Cohler et a1. hypothesized, that since the revolunteer mothers found childrearing to be more enjoyable, they had more interest in the completion of the questionnaire; or possibly, since they had more children, they simply had a greater desire to get out of the house. In a study carried out to evaluate the Salk polio vaccine (Meier, 1972), it was found that mothers who volunteered to have their children vaccinated were higher in education and social class. Fairweather (1969) considered the question of the effect of the volunteer status of the participants in a research pro- ject he conducted to evaluate different after care systems for mental patients. He found that volunteers and nonvolunteers for an innovative after care system were significantly differ- ent on three variables. Nonvolunteers had, on the average, 12 l. spent less time in mental hospitals, 2. held employment in a higher occupational category, and 3. were more frequently social drinkers as opposed to heavy or non-drinkers. Non- volunteers were then matched on all hiatorical variables, including these three where significant differences existed, to a group of subjects who had volunteered for the innovative system but who had been assigned to the control condition sys- tem. The intention of the matching was to eliminate historical differences between the samples. When matched, the subjects were measured at five six-month intervals on thirteen follow- up measures. These measures included the social change variables of importance in his study: cumulative days in the community and days of full-time employment; respondent evalu- ation of the patient's association with friends, symptom behavior, drinking behavior, etc. and patient self-evaluation of his satisfaction with his living condition, leisure activity, and community living. 0f the 65 chi-squares, only three were significant. In this study therefore, subjects matched so that the only measurable difference between them was that of volunteering or not volunteering, did not exhibit any differ- ences in the variables of interest in this study. Finally, Parlee (1974) found that the levels of estrogen and progesterone might be significant correlates of volunteer- ing in women. These studies on tOpics related to family planning or using populations and methodologies related to the present study seem to support several of the conclusions reached by l3 Rosenthal and Rosnow, especially with regard to the greater un- conventionality of volunteers. However, these studies do seem to account for more than their share of contradictory or other- wise distinctive findings. Situational‘DeterminantS‘of'VOlunteering The findings on a variety of "more or less" situational variables that possibly influence volunteering were summarized by Rosenthal and Rosnow in a list of 10 conclusions divided into four groups and presented in order by the degree of con- fidence in the association of the variable and volunteering. To order these variables, similar criteria to those employed in ordering the volunteer characteristics were used. A discussion of the first and second conclusions, of particular relevance to the present study, will follow this listing. Rosenthal and Rosnow's conclusions are as follows (1975, pp. 118-119): 'ConclusionS'warranting;MaximumgConfidenCe 1. Persons more interested in the topic under investi- gation are more likely to volunteer. 2. Persons with expectations of being more favorably evaluated by the investigator are more likely to volunteer. 'Conclusions:warranting;considerable'Confidence 3. Persons perceiving the investigation as more import- ant are more likely to volunteer. 4. Persons feeling states at the time of the request for volunteers are likely to affect the probability of volunteering, especially when contact with the un- intended victim can be avoided and when the source of guilt is known to others. Persons made to "feel good" or to feel competent are also more likely to volunteer. 5. Persons offered greater material incentives are more likely to volunteer, especially if the incentives are l4 offered as gifts in advance and without being con- tingent on the subject's decision to volunteer. Stable personal characteristics of the potential volunteer may moderate the relationship between volunteering and material incentives. 'ConcluSions'warranting;some‘Confidence 6. Personal characteristics of the recruiter are likely to affect the subject's probability of volunteering. Recruiters higher in status or prestige are likely to obtain higher rates of volunteering, as are female recruiters. This latter relationship is especially modifidable by the sex of the subject and nature of the research. 7. Persons are less likely to volunteer for tasks that are more aversive in the sense of their being painful, stressful, or dangerous biologically or psychologically. Personal characteristics of the subject and level of incentive offered may moderate the relationship between volunteering and task aversiveness. 8. Persons are more likely to volunteer when volunteering is viewed as the normative, expected, appropriate thing to do. ‘ Conclusions warranting Minimum Confidence 9. Persons are more likely to volunteer when they are per- sonally acquainted with the recruiter. The addition of a "personal touch" may also increase volunteering. 10. Conditions of public commitment may increase rates of volunteering when volunteering is normatively expected, but they may decrease rates of volunteering when non- volunteering is normatively expected. The finding that those more favorable toward or more interested in the topic are more likely to volunteer is based on research on many different topics. For instance, those respondents owning radios and using them more were more likely to answer questions or respond to mail questionnaires earlier (Rosenthal and Rosnow, 1975, p. 111). "Early responders to a 15 survey on the use of airlines were more than twice as likely as later responders to have flown" (Rosenthal and Rosnow, 1975, p. 111). In studies of public policy "respondents may be overrepresented by individuals with strong feelings against the proposed policy" (Rosenthal and Rosnow, 1975, p. 112). In studies of sex behavior, the greater sexual experience and/or permissiveness of volunteers found in some studies was taken as evidence for the hypothesis. In a recent study not included in Rosenthal and Rosnow, Keith (1974) manipulated the degree of perceived similarity between a stimulus organization and community organizations on the dimensions of activity, scope, value and operational orientation, perspective, and size. It was found in contrast to the above findings that the manipulation had no significant effect on the likelihood of the community organization to attend a work-conference run by the stimulus organization. When "the subject's interest is defined in terms of activity in, and affiliation with formal organizations" (Rosenthal and Rosnow, 1975, p. 112), the relationship is still supported. This is also the case when interest was defined by "the sub- ject's favorableness to treatment procedures as measured by his continuation in, or benefit from, his treatment" (Rosenthal and Rosnow, 1975, p. 113). These latter studies concerned a mental health outpatient clinic, a vocational guidance program and an alcoholism treatment program. "In terms of the magnitude of the effect, the interest or involvement of the subject 16 appears to be one of the most powerful determinants of volun- teering" (Rosenthal and Rosnow, 1975, p. 113). The second of the situational determinants, the expecta- tion of being favorably evaluated, conCerned studies where subjects were in a position to make favorable self-reports about their educational or occupational achievement, or psychiatric or gender-related adjustment. There is also one study reported where the probability of being favorable evalu- ated was specifically manipulated as an independent variable and the hypothesis was supported (Olsen, 1968). As Rosenthal and Rosnow (1975, p. 117) point out: Olsen's study is especially important because in so many of the other studies summarized it is not possible clearly to differentiate the variable of expectation of favorable evaluation from the variables of subject interest, intelligence, education, and social class. Subjects may be more interested in things in which they have done well, and we are usually in no position to say whether it is their interest or their having done well that prompts them to respond by participating in our research. Better-educated, more-intelligent subjects and those classified as higher in social class are by common cultural definition in a position to say ”better" things about themselves than those classified as lower in social class. The very clear results showing higher rates of volunteering by those more interested in the research area, by those who are better educated or more intelligent, and by those classified as higher in social class may, in a sense, provide additional support for the hypothesis that volunteering rates increase with the increased expectation of favorable evaluation. Threats to_generalizability and inferred causal relationships Given these findings, the methodological implications of volunteer status could be quite extensive. Rosenthal and Rosnow (1975) sketch on an intuitive level some possible 17 examples of volunteer bias. For instance, if a researcher relied on volunteer subjects to standardize norms for an IQ test, given the finding of higher intelligence for volunteers, it follows that the estimated population parameters could be seriously in error, e.g. an artificially inflated mean. Rosenthal and Rosnow point out (1975, p. 119): we might expect that routinely sampling just volun- teer subjects in behavioral research should jeopardize the robustness of any research conclusions if the subjects' educational background, social class, intelligence, need for social approval, sociability, and so on could be indicted as pertinent differentiat- ing characteristics related to the estimation of para- metric values for the research problem in question.... In clinical research, nonconformity may be suspect; in 'medical research the subjects' psychological adjustment; . and so on. Rosenthal and Rosnow further illustrate how the volun- teer status of subjects could lead us to make errors in either overestimating or underestimating the effectiveness of an experimental procedure. For instance, suppose an investigator wanted to test the effectiveness of a persuasive appeal by exposing one group to the appeal and comparing their reactions to an unexposed control group. Their conclusion states that volunteers are likely to be high in need for social approval. Crowne and Marlowe (1964) have found that people high in this need tend to be more readily influenced than low-scorers. This could lead the exposed subjects, by the fact of their volunteer status, to overreact to the appeal, thus distorting the true magnitude of the difference between them and the control group and thereby leading the experimenter to overestimate the 18 persuasiveness of the appeal, or to a "type I"1error. An example of a situation in which the opposite erroneous conclusion, or a "type II"1error, could be made is one in which an experimenter is testing the effect of an edu- cational program designed to reduce rigidity in thinking. In this case, it might be hypothesized that volunteers are less rigid in their thinking due to their hypothesized low authoritarianism, and thus all the subjects would already be low in the rigidity of their thinking, and the experimenter would be led to underestimate the effect of the educational program.due to the artificial minimizing of the difference be- tween the experimental and the control subjects. Rosenthal and Rosnow have conducted a series of studies designedto test experimentally the effects of voluntarism which they conceptualize as an artifact-independent variable. Their first studies (Rosnow and Rosenthal, 1966; Rosnow, Rosenthal, McConochie and Arms, 1969) demonstrated that volun- teers may be more accomodating to what they perceive as the demand characteristics of the situation. For instance, volun- teers reacted more positively than nonvolunteers to a positively slanted communication and more negatively than nonvolunteers to 1It should be noted here that Rosenthal and Rosnow's usage of these terms differs from the strict statistical usage of rejecting the null hypothesis when it is true or not reject- ing the null hypothesis when it is false, respectively. Their usage relates rather to the validity of the experimental find- ings when generalized to other than volunteer populations. 19 a negatively slanted one. When a two-sided presentation was used to obscure the directionality of the demand characteris- tics, the stimuli were generally ineffective for both groups. The next study (Rosnow and Suls, 1970) was conducted to explore the hypotheses of "type I" and "type II" confound- ing postulated above. The research also explored the idea of pretest sensitization as a source of invalidity in attitude change research. Since our exploratory studies suggested that volun- teers may be more compliant with demand characteris- tics than nonvolunteers, it could be hypothesized that the effect of using a before-after design on volunteers should be in the direction of a type I error, whereas the effect with nonvolunteers should be in the type II direction. (Rosenthal and Rosnow, 1975, p. 151.) A Solomon 4-group research design was used, and the main finding was that: Pretested volunteer subjects were more acComodating, and pretested nonvolunteers less accomodating, to an attitude manipulation. This suggests that using be- fore-after design can lead to overestimates of the attitudinal effects of persuasive communications when the subjects are motivated positively pursuant to demand characteristics and to underestimates when they are ca tive nonvolunteers (Rosenthal and Rosnow, 1975, p. 155). In the next experiment (Goldstein, Rosnow, Goodstadt and Suls, 1972), using a standard verbal operant conditioning procedure, a biasing effect of voluntarism was also demonstrated, along with a finding that there might be a curvilinear relationship between demand clarity and compliance. The level of compliance might be highest when the demand characteristics are neither so blatant as to cause reactance nor so subtle as 20 to not be clearly received. In the neXt study (Rosnow, Goodstadt, Suls, and Gitter, 1973), evidence was provided that the strength of ego- defensive motives is a possible mediating variable affecting compliance for volunteers and nonvolunteers. A conflict was set up between two sets of cues, one related to demand characteristics and the other to the possibility of unfavor- able evaluation. To establish the ego-defensive cues, an actor was hired to pose as another subject in the group and remark that he had heard from a friend in another class that, depend- ing on the condition, "only people with really high IQ's" or "only people with really low IQ's" would perform the task in the direction indicated by the demand characteristics. In this study, the volunteers and nonvolunteers were both likely to conform to the ego-defensive cues rather than the demand characteristics when they were in conflicting directions. In the last study reported (Aiken and Rosnow, 1973), it was found that the role expectations that subjects associ- ated with research participation were altruistic (advancing science, and helping the experimenter), and evaluative (being evaluated on some psychological dimension) in that order, with obediance to authority having less pertinence. The major difference between volunteers and nonvolunteers was that non- volunteers put "significantly heavier emphasis on a work- oriented dimension. . . whereas the volunteers placed more ‘weight on an affective dimension" (Rosenthal and Rosnow, 1975, p. 171). 21 Rosenthal and Rosnow conclude their comprehensive and fascinating work with a proposed theoretical model where volunteer status was seen as one of many artifact-independent variables that affect studies' outcomes by impinging on the behavioral states of compliance, counter-compliance, and non- compliance at any of three mediating points: receptivity, motivation, and capability. Various experimental procedures for "circumventing the artifact problem or for teasing out any biasing effect when the difficulty cannot be avoided" (Rosenthal and Rosnow, 1975, p. 302) were also presented. Implicationsfor'the'Present'Study The purpose of the present study is to determine the differences between volunteers, partial volunteers and non- volunteers for a family planning educational study. Partial volunteers are those subjects who respond that they might attend the program. The distinguishing characteristics of the present project are that recently married younger women from an areaewide and not an exclusively college popuIation were personally contacted; interviewed at their homes about their family planning knowledge, attitudes, practice and related items; and then asked to participate in an educational program, called the Family Life Service Project, identified as being sponsored by Michigan State University. Since different con- tents and channels of messages were to be evaluated, no speci- fic thme, place, format or topic was mentioned. Because of the content of the pretest, it will be assumed that in Spite of the rather nebulous title of the project, it was perceived 22 as being concerned with birth control methods and family size determinants and that the subjects' decisions regarding participation were based on that perception. The findings compiled by Rosenthal and Rosnow and their experimental studies raise two main concerns for any study of family planning educational programs: the validity of any inferred causal relationships and self-selection of the audience. Validity of inferred cansal relationships. The family planning educational program was experimentally evaluated as to the effectiveness of different messages and channels of communication (Taylor, 1975). Volunteer subjects have been reported as being higher in need for social approval and there- fore more readily influenced and more compliant to demand characteristics. Experimental evidence was also reported (Rosnow and Suls, 1970) to support the hypothesis that these characteristics would lead in the direction of a"type I"error, overestimating the effects of a persuasive communication. In the present study there are many factors that would lead to this type of error. The demand characteristics would probably be consistently in the pro-birth control, low family size direction. The pretest sensitization, as a before-after design was used, and the message content, no matter how care- fully balanced the presentation, would probably have this effect. Furthermore, it is probable that a program and people associated with a university could be generally perceived as 23 having these "liberal" pro-ecology, pro-birth control, etc. values. The ego defensive cues, also found to be an important determinant of compliance (Rosnow, Goodstadt, Suls, and Gitter, 1973) would, because of the perception of the values of the program personnel, also lead in this direction. The present study cannot directly address this question of the validity of the experimental findings. However, it is hoped that the present study's measurement of the extent, direction, and magnitude of volunteer-nonvolunteer differences on variables of interest in the family planning area will give some guidance to the interpretation of any possible bias in the experimental evaluation of treatment programs in the family planning area. Any serious amount of bias could cause the experimental findings to be generalizable only to populations recruited in a very similar manner to the present study. Self-seleCtion of audience. The second concern raised by the findings in the literature is that the voluntary nature of program participation might mean that the program would involve only a very select subsample of the general population with different general characteristics as well as possibly very different knowledge of, attitudes and expectancies toward, and practice of birth control, the main variables of interest in this area. We could, for instance, find support for M. Brewster Smith's contention that "the resistance of many segments of society is such that self-selection operates so that audiences become restricted to those already informed and 24 converted" (Smith, 1966, p.332). If indeed voluntary programs reach only audiences who are most susceptible to them, most interested in and knowledgeable about the subject matter, and already "converted", then educational programs with voluntary participation could have only very limited utility as effec- tive change agents of any magnitude in society. Limits to the Applicability of the Literature to the Present Study The present study was designed and carried out in 1970-71 when Rosenthal and Rosnow's compilation of findings and experimental studies were in a preliminary stage. Their findings are therefore not specifically incorporated into the designs or measures used. There are several issues concerning distinctive characteristics of the present study and restric- tions on the findings compiled above, that should be noted. The continuing need at the present time is to start building a reliable index of volunteer-nonvolunteer differences for edu- cational programs in the family planning area and on the independent and dependent variables of particular utility or relevance in that area. The question of situational specifi- city of findings is important here. As noted the topics most related to that of family planning showed a particularly high incidence of findings that were somehow distinguishable from the other studies compiled. The population studied in the present research: non-college women, and the techniques used: survey research in the community, were also characteristics of 25 many of the studies that were somehow distinguishable from the other studies compiled. ' The Rosenthal and Rosnow findings could also be less applicable because the present study compared second-level volunteers or revolunteers with an original volunteer sample, a methodology that was not the standard one used in the studies compiled. Different "types" of volunteers could have very different characteristics. There are two further factors that need to be mentioned in a consideration of the applicability of Rosenthal and Rosnow's findings. The first is that the interviewed subjects were asked to participate in an educational program, the Family Life Service Project, which probably did not have as clear an identity as a research study as most of the studies compiled, which concerned volunteering for laboratory experiments. One can hypothesize that the reported altruistic motivation, "helping science" or even helping the persons who contacted them, who were probably perceived clearly as interviewers and .not researchers, would be less salient in the present situation. Given that this was the most important motivation located by Rosenthal and Rosnow, then some of the findings based on studies of individuals volunteering for those reasons would not apply to a study where the volunteering was for an educa- tional program, which could be perceived as offering the potential for more personal benefit than the traditional laboratory experiment. 26 A final point to be raised here concerns the type of measures used in the wide-range of studies analyzed by Rosenthal and Rosnow. Many of the variables were measured with pen and pencil tests and projective tests, e.g. the Rorschach, MMPI, Marlowe-Crowne, California-F, etc. These measures could have questionable reliability and validity for establishing the kind of behavioral and attitudinal differences of interest here (Forsyth and Fairweather, 1961). Hypotheses The variables to be measured in the present study will be discussed throughout as organized into six conceptual groups: interview process, demographics, reproductive history, family planning attitudes and expectancies, knowledge and practice of birth control and decision making influences and process. It should be noted that there is a wide variance in the amount of support for the hypotheses to be made. It should also be noted that the partial volunteers will not be specifically mentioned in the hypotheses below but will always be understood as having values on the variables between the values of the volunteers and nonvolunteers. Interview process. Rosenthal and Rosnow's compilation of the situational determinants of volunteering included the finding that the characteristics of the recruiter did influence the rates of volunteering with high status and female recruiters obtaining higher rates. A prior acquaintance with recruiter 27 or some ways of adding a "personal touch" were also found to increase volunteering. It could be that other characteristics of the interviewer including the friendliness, perceived similarity of status, etc. could also influence volunteering. It could be that longer contacts between the inter- viewer and the respondent could increase the volunteer rate because of increased opportunity for a personal contact and identification with the interviewer to be established. The reported higher need for social approval and greater conformity to demand characteristics (being a "good" subject) of volun- teers would lead to their being more cooperative with the interview process, e.g. giving more complete, thought-out answers requiring more time. A longer contact could also be an index of the "personalness" of the contact, respondents talking more freely to interviewers with whom they felt more rapport, and making more curt replies to interviewers they felt less comfortable with. The interviewer could similarly extend or curtail contacts with subjects depending on how comfortable the contact was. This effect it would seem would be parti- cularly strong with an interview on a very personal topic such as family planning. A final index of the degree of comfort- ableness of the interview process would be the reaction of the subjects to the questionnaire content itself. we can therefore hypothesize that: Interview Process HypOthesIS‘l. The particular interviewer contacting the subject will influence the rate of volunteering. 28 IntervieW‘ProceSS'HyPOtheSIS'2. The time spent in the inter- ViéW‘Will be longer for the volunteer than for the nonvolunteer. Interview Pr0cess HypOthesis'3. The volunteer will resent fewer items on the questionnaire form than the nonvolunteer. DemographicS. Rosenthal and Rosnow's (1975) conclu- sions that volunteers were higher in educational level and social class status were those with their maximum amount of confidence. Though income was not specifically dealt with, it can be considered as associated with these first two vari- ables. There are several additional factors that could support these hypotheses for the present study. First, it could be that highly educated people would feel less apprehension about participation in an educational program sponsored by a univer- sity because of their greater familiarity with and probably more successful experiences in an educationalsetting. Secondly, their educational and social status could lead them to expect to be more favorably evaluated in the research sett- ing (an important determinant of volunteering) just because of those demographic characteristics as discussed by Rosenthal and Rosnow (1975, p. 117), or because of that greater experi- ence in educational settings. It could also be that high social status-income people have more leisure time for edu- cational pursuits. It should be noted at this point that many of the variables in the study were measured both for the respondent and her husband. Not having any evidence to the contrary, the 29 same hypotheses will be made for both. The following hypo- theses will therefore be made: Demggraphic HypOthesiS'l. The volunteers and spouses will have a higher educational level than the nonvolunteers. Demographic Hypothesis'Z. The volunteers and spouses will have a higher social class status than the nonvolunteers. Demographic HyPOthesis 3. The volunteers will have a higher income than the nonvolunteers. As far as race/ethnic group status is associated with the above demographics, and given the fact that the white interviewers and university identification would not give black women any special assurance of the programis relevance to their needs, we can hypothesize that: Demggraphic HypOthesis 4. Volunteers will be most likely and nonvolunteers least likely to be white rather than non-white. Rosenthal and Rosnow (1975) assert with "some confidence" the conclusion that volunteers are younger than nonvolunteers. Differences in religious preference is a con- clusion with "considerable confidence". For the topic of the present study it would seem that age and religious preference would be at least as salient as for the average laboratory study. we will therefore state the following hypotheses: Demographic Hypothesis 5. Volunteers and spouses will be younger than nonvolunteers. Demographic Hypothesis 6. Volunteers and spouses will be more likely to be Jewish than Protestant and more likely to be 30 Protestant than Catholic, while nonvolunteers will show the opposite relationship. There is no direct evidence on the effect of women's working status on volunteering behavior. It is felt, however, that working (full or part time) women could be more in con- tact with and involved in the "outside world" and therefore less apprehensive about and more interested in becoming in- volved in something outside the home routine. It could also be that the topic of birth control could be of more interest to them because of its importance in allowing them to continue working. The hypothesized higher need-achievement of volun- teers could also support this hypothesis. It is felt that these factors could outweigh the opposing factors of working women possibly being of lower social status and thus less likely to volunteer, having less free time for outside pursuits, or having less need to get out of the house than full-time house- wives. It will be hypothesized therefore that: Demographic Hypothesis 7. Volunteers will be more likely than nonvolunteers to be employed. The effect of student status on volunteering could be influenced by many different variables. The hypothesized higher educational level and younger age of volunteers would both support the hypothesis of their being more likely to be students. Also student status could mean that subjects would be generally interested in educational pursuits of all kinds, would identify more with a program associated with the major [I’ll 31 educational institution in this area, and in general would be more at home in educational settings. Therefore, the hypo— thesis will be made that: Demographic Hypothesis 8. Volunteers and spouses will be more likely than nonvolunteers to be students. The number of organizations belonged to could, similarly to working status, be seen as an index of involvement and interest in outside pursuits. A greater number of organiza- tions belonged to could also increase the chances of a subject belonging to some organization with a connection to the subject of family planning, ecology, etc. we can thus hypothesize: Demographic HypOthesis 9. Volunteers and spouses will belong to more organizations than nonvolunteers. ReproduCtive'HiStory. The only direct finding of the relationship between the actual fertility level of subjects and volunteering was the study of child rearing practices discussed earlier (Cohler, et al., 1968). This study found volunteers to have more children than nonvolunteers. It is felt that this one study is outweighed by the findings of the standard family planning literature (Fawcett, 1970; Pohlman, 1969; Rainwater, 1965; Whelpton and Kiser, 1966). Fertility is shown in these studies to be negatively correlated with education, income, and social class. Given volunteers' hypothesized higher values on those variables we would expect volunteers to have lower fertility levels. 'We will thus hypothesize: ReproduCtive’Hiatory‘HyPOthesiS'l. Volunteers as compared to 32 nonvolunteers will have fewer: pregnancies; stillbirths, abortions and miscarriages; children born live; living child- ren; and children living in the home. Reproductive Hiatory‘HyPOthesis 2. Volunteers will be less likely to be pregnant than nonvolunteers. Family planning attitudes and expectancies. The relationship between family planning attitudes and expectancies and volunteering would seem to again be addressed by the standard family planning literature which indicates that higher income, social class, and educational levels would lead to mpre liberal attitudes toward birth control, and abortion and possibly to higher educational aspirations for children, a secondary determinant of lower desired family size. According to the literature, however,‘pp difference between social classes on actual desired family size could be supported. It is felt that the lower classes desire as small a family as the upper classes but because of problems regarding knowledge about and access to methods of contraception cannot always achieve this goal (Fawcett, 1970; Pohlman, 1969). There are further factors that would seem.however, to influence the family size desires in the direction of volunteers having smaller desired families. The first is the general idea of self selection operating so that the volunteers are an already converted audience. The belief that the university is perceived as "liberal" pro-birth control, low family size etc. could lead to only those women with those attitudes feeling that they would be favorably evalu- ated and thus feeling free to volunteer. The family planning 33 literature supports the idea that the above hypothesized differences in religious preference would lead to more liberal attitudes and lower family size desires among volunteers. Finally, volunteers' reported nonauthoritarianism and un- conventionality could lead to more liberal attitudes and ex- pectancies. The hypotheses to be made are therefore: Family Planning;AttitudeS‘and‘Expectancies Hypothesis 1. Volunteers and spouses will have more positive attitudes toward birth control and abortion than nonvolunteers. Family PlanningyAttitudeS‘andExpectancies Hypothesis 2. Volunteers will desire and expect fewer children than non- volunteers. Family,Planning;AttitudeS'and‘ExpectancieS‘HypOthesiS‘3. Volunteers will consider ideal for the average couple in the U.S. a smaller number of children than nonvolunteers. Family Planning Attitudes and‘Expectancies Hypothesis 4. Volunteers will consider ideal a longer time interval between marriage and the birth of the first child and between children than the nonvolunteers. Family Planning_Attitudes and Expectancies Hypothesis 5. Volunteers not already having children will be planning a longer time interval between marriage and the birth of the first child than the nonvolunteers. Family Planning Attitudes and Expectancies Hyppthesis 6. Volunteers as compared to nonvolunteers will consider a smaller number of children as "too many" and a.smalhr number of children as "too few" for the average U.S. couple. 34 Family Planning AttitudeS'and ExpecrancieS'HypOthesiS 7. Volunteers will be more likely than nonvolunteers to be plann- ing to adopt children. FamilylPlanningyAttitudeS'and'ExpeCtancieS'Hypothesis 8. Volunteers as compared to nonvolunteers will consider having a male heir as less important and will have higher educational aspirations for their children. Knowledgg andpractice of birth control. We could generally expect that volunteers' hypothesized higher education- al level would lead them to be more knowledgeable about many matters, including reproductive physiology and the methods of contraception. Through their higher educational level they could also be more exposed to the media and other educational sources and thus have access to more information in that way. The hypothesized higher educational level, social class and income could all be seen as giving volunteers greater access to sources of medical care and contraceptive devices. It could also be that people who had some information about and experience with a topic would be more interested in it, a potent situational determinant of volunteering according to Rosenthal and Rosnow (1975). Similarly to the points raised above concerning attitudinal items, those people with some information and the "correct" contracepting behavior might expect to be more favorably evaluated and thus be more likely to volunteer. In support of the opposing hypothesis, however, it could be that the more informed people, and those successfully 35 contracepting, could feel that they have all the information they need about family planning, and therefore do not need to attend a program. This would create an audience of those who feel a lack of information from.other sources and who want to be better informed. From the standpoint of program.utility, this latter situation would be the most favorable kind of bias. The only study reported above that specifically tested for differences in contraceptive behavior (Bauman, 1973) -- with a single college-age population -- found no differences. The same study did find female volunteers to have less sexual knowledge. Maintaining a consistent framework of volunteers being an already converted audience, we will hypothesize that: Knowledge and Prentice of Birth COntrol Hypothesis 1. Volun- teers will be more likely than nonvolunteers to name correctly the most fertile time of the menstrual cycle. Knowledge and‘Praetice'of'Birth‘ControlHypothesiS'Z. Volun- teers will have more familiarity/use of all the methods of birth control than nonvolunteers. Knowledge and Practice of Birth Control Hypothesis 3. Volun- teers will be likely to be currently using a method of contra- ception and to be using a more effective method than non- volunteers. In agreement with the more favorable attitudes, greater knowledge, and greater use hypothesized above, it was felt volunteers would also differ from volunteers on the related items concerning their feelings about the methods of contra- ception and utilization of related types of medical care. It 36 will be hypothesized that: ‘Knowledge‘and'PraCtice‘of'Birth Control HyPOthesiS'4. Volun- teers using a method of contraception will use it more often, consider it to be more effective and'be more satisfied with it than nonvolunteers. Knowledge and Practice of'Birth'COntrol HypOthesis 5. Volun- teers will report obtaining Pap smears more frequently than nonvolunteers. The idea that public sources of care, especially several years ago at a time of greater conservatism among private physicians, were the most available nonjudgemental, and experienced sources of family planning care and informa- tion underlies the next hypotheses. One could suggest that those more interested in the topic of birth control, more knowledgeable about it, successfully using a modern medical method of contraception, etc. would be more likely to have had some contact with a public clinic. For lower income women, especially the working poor, public clinics could be seen as the only available source. It could be hypothesized therefore that: Knowledge and Practice of Birth Control Hypothesis 6. Volun- teers will report more knowledge of and be more likely to have used the Ingham County Family Planning Clinic, the main public source of care in this area. Decision‘making;inf1uences‘and‘process. The reports that volunteers have a greater need for social approval and III|III 37 are more readily influenced than nonvolunteers (Rosenthal and Rosnow, 1975) would support the idea of their decision making about family planning being more influenced by various outside sources. Volunteers' hypothesized greater knowledge about these matters and higher educational level would also lead in the direction of their receiving more information about these matters fram a greater number of sources than nonvolunteers and thus possibly being more influenced by them. The final hypotheses to be made are therefore: DecisiOn Making InfluenceS'and'Process HypOthesis 1. Volunteers will be more likely than nonvolunteers to have had discussions with their husbands regarding birth control and to report more satisfaction with those discussions. Decision Making Influences and PrOCesS HypOthesis 2. Volunteers as compared to nonvolunteers will report that, between them- selves and their husbands, they have less say over the choice of birth control methods. Decision Making Influences and Process Hypothesis 3. Volunteers as compared to nonvolunteers will report a greater degree of influence from all the possible sources of communication on their decisions regarding use of birth control and choice of a desired family size. CHAPTER II METHODS AND PROCEDURES Subjects The subjects for this study were 199 female Ingham County, Michigan residents who obtained marriage licenses in the County between June 1970 and November 1971 and who were 25 years of age or younger at the beginning of the study, November, 1971. This 199 represents a subsample of 331 women who were attempted to be contacted. Almost twenty per- cent of the original sample could not be located and another 20% refused to participate in this survey study. This study was a part of a larger research project, the Comparative Field Experiment in Voluntary Birth Planning (Taler, 1975), a study of the efficacy of various modes and messages of family planning education. This group of subjects was chosen as representative of the young recently married women who are a prime target audience for such educational programs. Starting with the most recent months and moving back- ward, the names of all women under 25 years of age obtaining marriage licenses in the County were obtained giving a total sample of these women with the following constraints. Only women who were presently living in the County and who had listed telephone numbers were included for ease in subject 38 39 contact procedures and to allow them to participate in the later educational presentations of the Family Life Service ProjeCt, as the educational program of the Comparative Field Experiment was labeled. Design The study is a comparative analysis of volunteer, partial volunteer, and nonvolunteer groups of young recently married women for a family planning educational program. The design is presented pictorially in Figure l. The groups were determined and variables measured by means of an interviewer FIGURE 1. Design Used in Study Volunteer Partial Non- Group Volunteer Volunteer n = 110 Group Group n = 39 n = 50 N = 199 administered scale. At the end of the scale, the Family Life Service Project was briefly described. Since different con- tents and channels of messages were to be evaluated, no speci- fic time, place, format or topic was mentioned or discussed. The subjects were then asked whether they would like to participate. They were allowed to give answers of "yes", "maybe", or "no." Three groups were identified: the volunteers 40 or those who answered "yes", partial volunteers or those who answered "maybe", and nonvolunteers, who answered "no". The survey data was then analyzed for differences among these three groups of subjects. Procedures The Ingham County Clerk's public records of marriages were conveniently obtained by ordering back copies for the previous year and a half of the weekly publication of a local credit bureau. This publication lists the names, addresses, and ages of each couple applying for a license. Those women over 25 years of age were eliminated from the lists. Because newlyweds are likely to change their places of residence after their marriages, current addresses and phone numbers were needed. Michigan Bell was contacted and rather than call— ing the Information Operator for each name, the lists were sent to the Michigan Bell offices and their clerical staff looked up the names and recorded the current information when available. Each woman for whom an address was located was then sent a letter (Appendix A). This letter introduced them to the Family Life Service Project and informed them of the in- tended home visit and survey. A total of twelve interviewers were hired. The major- ity of these women had had previous interviewing experience, many with an urban survey research unit of the University. Many of them had done previous survey work on family planning. 41 Extensive training procedures were carried out using materials from the University of Michigan's Survey Research Center and from Research Methods in Social RelatiOns (Selltiz, Jahoda, & Cook, 1959). The interviewers were assigned all potential subjects in a given geographic area to minimize interviewing costs. It had been found in a pretest of the instrument and subject con- tacting procedures with women meeting the same age and length of marriage criteria but living in a neighboring country, that the most efficient procedure after the introductory letter was a phone call rather than a personal visit because of the high percentage of women not at home. Therefore the interviewers attempted to contact each potential subject by phone and during that call set up a subsequent time for the interviewer to visit the woman at home and complete the survey instrument. Our first finding was that this young married popula- tion was highly mobile. The interviewers did extensive checks with neighbors, rechecked with the phone company, etc. to locate as many women as possible and many women were traced through these procedures. As tabulated from the first half of the subject contacting procedures, however, almost 20% of the women had moved out of the County or could not be located. An equal percent were contacted and refused to be interviewed. The majority of these women cited lack of time as the reason for their refusal. Several of these women had agreed to the interview, and then their husbands, overheard by the interviewers 42 in the background, told them to refuse. A few others set appointments and were then not at home when the interviewer came to their house at the agreed upon time. If a woman ‘missed three appointments in this manner, she was counted as a refusal. It was found that having the interviewers call the women right before they left for the appointment helped save unnecessary trips. The interview took an average of 28 minutes (S.D. = 11.4) to complete. After its completion, the name and address of the respondent were removed from the form which was identi- fied by an identification number only for the rest of the research procedure. The confidentiality procedures governing the project were described to the subject in the initial letter and phone call and in more detail at the beginning of the interview. Measures The variables of interest in this study were measured by an 81-item interviewer-administered questionnaire. After final scoring, 95 variables were generated for analysis in this study. The questionnaire, containing the coding scheme for the majority of the items, is presented in Appendix B. This questionnaire contains items from many different studies in family planning. Some were developed in research prelimin- ary to the present study (Tornatzky, Taylor, Harris, Beck, McKenzie and Nemon, 1970). Other items were similar to those used in Ronald Freedman's (1969) work in Taiwan. 43 The pretesting of this instrument was first done on a sample of twenty-two women who fit the age and length of marriage requirement but who lived in an adjacent county. This procedure was also carried out to test the subject con- tacting procedures and to give the interviewers field experi- ence with the instrument. The reliability of the instrument was studied using a sample of twenty additional subjects who fit the age, length of marriage and residency requirements for the experimental population, but who had not obtained their marriage licenses in Ingham County and were thus not in the sample. The test-retest method was used with twenty-four hours elapsing between the first and second administration of the instrument. The reliability of the items was computed by a correlation procedure for the attitudinal items or a percent agreement procedure for nominal scale items. There was a 100% agreement for these latter items while the overall correlation was .90. Item analysis showed correlations ranging from .33 to 1.00. Test-retest correlation cutoffs of .80 were estab- lished as the criteria for the retention of the item in the same form.in the final instrument. The items were ordered so that the more acceptable questions came first and the items that concerned more personal tOpics were asked in later stages of the interview process. The items can be most clearly presented as dealing with six general content areas. These are: 1. interview process 2. demographic 3. reproductive history 4. family planning 44 attitudes and expectancies 5. knowledge and practice of birth control and 6. decision making influences and process. On the items where a very closed-ended response was desired, and/or where it was felt the subject needed a visual guide, a card was given to her by the interviewer listing the possible res- ponses. Interview process. These items were included as measures of the potential effect of the process of the data collection itself on volunteering behavior. These items are the time spent in the interview; the interviewers themselves as identified by a number; and a question asked of the res- pondent at the end of the interview which concerned whether she disliked or resented any of the questions asked. Demographic items. The demographic items included as being possibly relevant to family planning and volunteering behavior were the following. The race/ethnic group of the respondent was observed and recorded by the interviewer; education for the respondent and her husband was recorded as highest grade of formal school completed; and the year of birth for each partner was obtained. The religion of the husband and wife were asked and scored as "1. Jewish 2. None 3. Protestant 4. Catholic and 5. other" in line with the traditional findings in the literature (Whelpton and Kiser, 1966, Pohlman, 1970) which report fertility ranging from the lowest to the highest level for religious groups in roughly that order (except for the "other" category which wasxnot used 45 in some of the later data analyses). This ordering also follows Rosenthal and Rosnow's (1975) conclusion reported in Chapter I. relating religion and volunteering. Their occupationswere recorded and later, with the educational level, those respondents and husbands who were employed were scored using the Hollingshead Two-Factor Index of Social Position (Hollingshead, 1957). Three other variables were generated to fully utilize this information: the wife's employment status which was coded as 1. not working 2. working part-time 3. working full-time," and the wife's and husband's student statuses which were each coded as "1. not a student 2. part-time student 3. full-time student." To obtain the income level with the least amount of subject reactance, the subject was handed a card giving various dollar amounts lettered from a to o. The subject then just had to say the letter corresponding to her family's gross total monthly income. The dollar amount, which were the medians of standard income ranges, was recorded for use in the data analysis. The last items in this section of background informa- tion are the number of organizations belonged to by the husband and by the wife. Reproductive history. The reproductive history, or vital data as it is sometimes called, was obtained by asking the number of pregnancies; number of stillbirths, abortions or miscarriages; number of children born live; number of living children; and to obtain the number of children presently cared for, number of children living in the home. The respondent was 46 then asked if she was currently pregnant. Family planninggattitudeS‘and expeCtancies. The res- pondent's and her husband's attitudes toward birth control, which was delicately phrased at this point in the study as doing "something to keep from.getting pregnant" was tapped by 5-point Likert-type scales. The number of children desired and number of children expected were obtained. The interviewer then asked the number of boys, number of girls and number of either sex children wanted. Then the respondents were asked if they were planning to adopt any children. The next group of items was the sub- ject's opinion on the idggl: time interval between marriage and the birth of the first child, interval between children, and number of children for the average U.S. couple. If she had no children and was not currently pregnant and had indicated she desired and expected to have children, she was asked how long she planned to wait for the birth of the first child. The 'respondent's opinions about how many children were "too many" and "too few" were then obtained. Three other items were asked in this category. The first was the respondent's attitude toward abortion which could range from "1. no abortion no matter what reason" to "5. abortion for any reason on request of the mother." Next a 5- point Likert-type scale measured the importance of having a male heir. The last question in this section asked for the respondent's educational aspirations for her children which was scored from "1. junior high" to "5.post-graduate." 47 ‘Knowledge'and'praCtice'of'birth'control. The first item in this section is used to assess the "reproductive sophistication" of a group. This item asks the respondent to name the most fertile days of the menStrual cycle. For the purposes of this data analysis, the number choosing the correct answer was compared with the number choosing the other answers. In this study a more lenient right answer, 8-21 days after the first day of the menstrual cycle, was used instead of the standard 15-17 days. The measurement of the knowledge and use of the various methods of birth control was measured by the following long item. For a list of 13 methods of birth control read to them by the interviewer, the subject could reply "1. never heard of it" "2. heard of it but never used" or "3. used it." If the sub- ject had used some method, she was asked if she was using a method of birth control at present, how often she used it, and on 5-point Likert-type scales how effective she felt it was and how satisfied she was with it. The method of birth control she was using was recorded and scored from."l. douche" to "9. pill" and "10. sterilization" in order of the general weight of evi- dence as to their efficacy (Emory University Family Planning Program, 1973). Other related reported behavior items are the frequency the respondent obtained Pap Smears which was scored from "0. never" to "1. every three years or longer" to "5. more than once a year." Her contact with the main public source of 48 family planning care in this area was obtained by asking if she had heard of the Ingham County Family Planning Clinic and if so, if she had received services from them. Decision making influenCes and process. In assessing the decision-making process within the family, in an early section of the interview the respondent was asked if she and her husband had discussed "doing something to keep from getting pregnant." If they had had these talks, her satisfaction with them was measured on a 5-point Likert-type scale. After the item concerning her knowledge and practice of specific birth control methods, the subjects who had used a method were asked an item.assessing, between their husbands and themselves, how ‘much say‘phgy had in the choice of birth control methods. This item was scored from "1. no say at all" to "5. decision totally mine." Two long items were designed to asseSs the influence of various sources of communication on the subjects' decisions regarding first, using or not using birth control and secondly, the number of children wanted. The possible sources of influ- ence which were read to the respondent can be organized into three groups: interpersonal which includes husband, close relatives, friends, druggist and physician; group eduCational or what could be seen as institutions including high school, hospital education programs, family planning clinic, adult education program (which was designed to cover the proposed Family Life Service Project) and a mixed item: church, 49 synagogue or clergyman; and lastly media, which included T.V. and newspapers, magazines and books. Each was rated on a 5- point scale from "1. none" to "5. nearly total influence." Volunteering measurement. One of the last items in the questionnaire was the item on which the volunteer/nonvolunteer measurement was made. "Michigan State University is going to offer a Family Life Education Program for recently married women in the near future. There will be one presentation last- ing about 90 minutes. WOuld you like to attend the Family Life Education Program?" The respondent could answer "yes", "maybe" or "no" which were coded as 3, 2, and 1 respectively. CHAPTER III RESULTS Of the 199 women interviewed, 110 or 55% stated that they would like to attend the Family Life Education Program. They will be called the volunteer group. Fifty women, 25% of the total, stated that they would not like to attend; they are the nonvolunteers. The‘partial‘volunteer group is the 39 women, 20% of the total, who answered "maybe" to the volunteer- ing question. Different statistical procedures were performed to analyze differences among these three groups on the variables of interest. Variables scored on nominal scales and some vari- ables on poSsible ordinal scales were analyzed by a chi-square procedure. The mean scores on variables which could be con- sidered as ordinal or interval scales were analyzed in a one-way analysis of variance with unequal N's. After the analyses of variance were run, the variance generated by the computer program for that procedure were used to test for the homogeneity of the variances. The nine vari- ables for which significant findings, were obtained and the seven variables for which trends in the data approaching sig- nificance were obtained were analyzed by the procedure developed by Hartley (Winer, 1962). These variables were selected for 50 51 this analysis as those of greatest interest for the interpre- tation of the results of the study. The‘Fpmax values ranged from 1.3311 to 7.054, with a mean of 3.0730. The largest of the group N's, 110, was used for the degrees of freedom to enter the table. "This procedure leads to a slight positive bias in the test, i.e., rejecting H1 more frequently than should be the case (Winer, 1962, p. 94). The critical value, a = .05, for 60 d: is 1.85 and the value for w, the next level appearing in the table, is 1.00. Using the 1.00 critical value, the H1 was rejected in all cases. Using the 1.85 level, the H1 was rejected in 10 of the 16 cases. The analysis of vari- ance procedure is reported to be robust with respect to departures from homogeneity of variance (Winer, 1962; Kirk, 1968). However, these departures should be kept in mind in evaluating the significance of the findings. The second procedure to be carried out was to evaluate the number of significant findings in the chi-square and analyses of variance procedures against the total number of tests run. The concern here is that a certain number of tests in a large series of tests are significant by chance. The figures developed by Sakoda et a1. (Sakoda, Cohen, and Beall, 1954) were used for this procedure. For the chi-square, the finding of 5 significant tests out of 28 tests was significant at the p <.05 level. For the analysis of variance, the find- ing of 10 significant tests out of 76 tests run was also signi- ficant at the p <.05 level. This small number of significant 52 results is,however, at best a marginal deviation from chance for the large number of tests run. A correlation matrix and a cluster analysis procedure (Tryon and Bailey, 1970) were performed on the data as a whole, for the variables on at least ordinal scales. A second cluster analysis with the variable of interest, the volunteering score, preset in a cluster structure was then run. The presentation of the comparative results, the chi- square and analysis of variance tests, will be presented for the variables as organized into content areas in the preceding chapter: interview process, demographics, reproductive history, family planning attitudes and expectancies, knowledge and practice of birth control, and decision making influences and process. The associative results, the cluster analyses, will then be presented. ComparatiVe Results Interview process. The first possible differences be- tween these three groups are those concerning the data collec- tion process itself. The first variable in this section is the interviewer. It was felt that different interviewers might obtain different rates of volunteering from the subjects they interviewed. The twelve interviewers hired worked for differ— ent amounts of time, were able to locate different numbers of their assigned potential subjects, and therefore interviewed from 1 to 45 total subjects.‘ In the first analysis of this variable, the data were collapsed so that the six interviewers 53 TABLE 1. Comparison of Volunteer Status Groups for High Active . , . ..and.Low:Active Interviewers ............... . .. _. . Volunteer Type of Interv1ewer Status ,Hish Active , Low Active Volunteer N_ 96 14 % 54 61 Partial N 35 4 Volunteer %' . 20 10 Non— N 45 _ 5 Volunteer %' 26 10 2 _ . x - 34 (2 df) Note: N = Number of individuals interviewed by that type of " interviewer in that volunteer status group. % = Percent of individuals interviewed by that type of interviewer in that volunteer status group. 54 'who completed the most interviews, the high active interviewers, were compared with the six interviewers who completed the few- est interviews, the low active interviewers, to determine if these different types of interviewer obtained different rates of volunteering. The chi-square on these data was not signi- ficant and is presented in Table 1. It was also felt that there could be individual differ- ences between the six high active interviewers. The chi-square analysis of these data was significant and is presented in Table 2. We can state therefore, that, looking just at the high active interviewers, interviewers do obtain significantly different rates of volunteering, with from 38% to 69% of their subjects agreeing to participate in the educational program. Table 3 presents means and analysis of variance results comparing the volunteer status groups on the number of minutes spent in the interview. The results do not approach acceptable levels of significance. For the third and last variable presented in this section, the number of questions on the form disliked or re- sented by the respondents, only three of the 199 subjects replied that they had disliked or resented any of the items. These three women all mentioned the item.on income as the one they objected to. One of the women did volunteer to attend the program, and the two others were partial volunteers. Demographic. The chi-squares in Table 4 present the results for the variables of race, religion, the wife's employ- ment status, and the wife's and husband's student statuses. 55 Table 2 . Comparison of Volunteer Status Groups for Most Active Interviews Volunteer Interviewer Status A E G I J L Volunteer E 27 5 25 12 12 15 7. 60 38 69 43 63 43 Partial N 5 5 3 5 3 14 Volunteer % 11 38 8 18 16 40 Non N 13 3 8 ll 4 6 Volunteer Z 29 23 22 39 21 17 x2 = 21.48* (10 £19 *p <.05 56 H>wrm.w-..3mmbm.moa.>bmpwmwm.om.BmHHnmo HnQHmo o o o o o o meow N N H u N b hmnHo >BmHHomn b b o o o o 20 umnm N N o o N b meHmHoa I SHmm umfiHmr H H o o o o m.NHw m Pm zoom Na Nb m NH Hm um wnonmmnmon mo bu Nb mN Ho wo omnSOHHo Nu Nm a Hm o Hm Onrmn m u H w b m meHmHou I momomba umszr N N o o o o N.H¢m Ho pm. zoom Nw NH 0 Nu Hu ma muonmmnmon mu am NH m» No no nmnonHo Nb NN m NH 9 Hm onsmn N a H w a m SonHom mnmncm I szm Soannm I mcHH HHBm mu mN Na mu No um Ho.HNN a am. zoansm I wmnn HHBm HN Hm N u N a 20" zoanbm um uw HH Nm Hm um 20 umnm o o o o H N 58 amch a. orHImnomnmm II umaomnmern bmHNmmm om bw.mN >0.No anrHo Gnocum How w.mo~ N.0H .Huu mamomoa.m «was om monsoon nnocom N Nm.HNH anno pu.ub bu.au Ro.mo aHnrHo onocvm Hem HN.wwu N.Nm .Hou SHmm.m mHmrmmn wmnsmmo Quocpm N >.omm nnmmm ooBonnmm Hu.mH Hu.mo Hb.05 anan mnocom Hem ».Nuo H.Hu .wHu mcm6m59.m mHmrmmn wmntmmo onocpm N Hm.m>H mamas noBvanma H¢.Ho Hw.NN H¢.m> anrHo anoovm Hem m.me u.wu .owH» SHmm.m Hanna om unnammo onocwm N op.ooN monHmH womHnHoo N¢.NN NN.oN No.»u anrHa onocum HHm bo.okw N.uo .Hom momcmom.m Hanna wmnflmmo muocwm N Ham.oaH om monHmH womHnHoo Nm.¢u NN.Hm Nm.Ho anan Onocpm Hum mu.wwb N.¢N .owo seam.m wmpameos N.oo ~.os ~.mo amnamms ouocem N .muu H.ou .uso anrHo mnocvm Ham .mHo mcmemsa.m wowemaos ~.os ”.mu ~.m~ assess: onocum N .Nmm .sou .mpw anrHo Onocvm Hm» .uNN Hsooam uHm.ou HoH~.o~ mms.~o amnsmms anocem N stusmm.uom s.mws .oww» anrHo Gnocum me womONu.N>m 61 amva m. amuse mom >omHwrm m. Amenm mua>bmemHm om oonnHoom on .NN .om .o¢ muocvm N .mbm ZHmomHHHmmmm , SHann H.wNN .Nmm nuocwm Hem .bmu, .... I . OUHHQHOU monsoon moan .wb .wm .NN muocwm N .OHN fiHanwmmm om nancmm Hosmnm anns noonnoH .uu b.wm ».mo anUHo H.0uw .wmm ,lmMocvm How .mou amnflmmn mcmvmom.m mnnuncam nosmna macs—Cm N H. om; GHHHU noonHoH b.b¢ b.No b.bm ZHHUHn H.mmN .NH» mnocum Hwo .Nom 66 The first part of family planning attitudes and expectancies hypothesis 1. was therefore not supported. The expectancy items of number of children wanted and number expected were also not significantly different for the three volunteer status groups. These analyses are presented in Table 9. The number of boys wanted, number of girls wanted and number of either sex children wanted were also not signi- ficantly different for the groups and are presented in the same table. Family planning attitude and expectancies hypothesis 2 was therefore not supported. Further analyses of variance were performed on some variables measuring attitudes regarding "ideal" family planning behavior. These variables are the ideal number of children for the average couple in the U.S., the number of children seen as too many, number seen as too few, ideal delay between the marriage and the birth of the firSt child, and ideal interval between children, both measured in years. There were no signi- ficant tests for these variables, presented in Table 10. An analysis of variance on an expectancy item, the number of years respondents not already having children planned to wait before the birth of their first child, counting from their marriage date, was also run and as presented in Table 10 showed no significant differences. Therefore, no support was found for family planning attitude and expectancies hypotheses 3, 4, 5, and 6. Three more analyses of variance were run on the atti- tudinal variables in this section and as shown in Table 11, were not significant. These variables were attitude toward abortion, 67 amva o. Zomba mom >bMmemm om bmemmm om nanoamm ammmHaHom HammH mmaHHw meobHom wmrdeon bmemmm Om nancmmm ammmanom HammH mmBHHN meobHom wmsm .oou .mmu 7O Hmch LL. Kmmom mom >omHNmmm om nancmHomH Human <0H¢5nmmn mwmncm mocnom menHmH zooI om Kama nanznm Hosmnm 2.338 PM: “.8 New Ross... N 88 .2; .me ZHHUHS onoapm Hob .owu wanton: Hafionnmnnm om >2»? mm: PS ....ON was 983 N PNS HEN L3 SHnrHo onocwm Hoa H.¢p> amazon: macomnHoan onocvm N .ouo >mvHumnHoom Obmm mum mow OSHHmHmo w.ww u.Nm w.w> anrHD . . muocvm Hmo .mao 71 the importance of a male heir and the respondent's educational aspirations for her children. (These items were scored so that the more permissive or liberal attitude in the case of abortion, the least importance of a male, and highest education- al aspirations were on the higher end of 5- point scales.) Family planning attitudes and expectancies hypotheses 1. (2nd part), and 8. were therefore not supported. The last variable in this section is the respondent's plans regarding adoption. The chi-square analysis for this item, presented in Table 12, was the only analysis with signi- ficant differences in this section, p <.01. This table indi- cates that the volunteers are the most likely to respond that they are planning on adopting a child. The nonvolunteers are most likely to respond that they are not planning to adopt. Family planning attitudes and expectancies hypothesis 7. is therefore supported. V Knowledge and praCtiCe of birth control. The first item to be considered in the assessment of the subject's know- ledge and practice of birth control is their knowledge of the time of ovulation, the most fertile days during the menstrual cycle. The chi-square analysis comparing the number of correct and incorrect responses for the three groups was not signifi- cant and is presented in Table 13. Overall, 37% of the sub- jects did not know this important fact for effective contracep- tion, even when scored on a more lenient basis than is standard as was explained in Chapter II. Knowledge and practice of birth control hypothesis 1. is therefore not supported. 72 Table 12. Adoption Plans Adoption Plans Volunteer Status Will Adopt Uncertain Won't Adopt N_ 25 53 32 Volunteer Z 23 43 29 N_ 5 15 19 Partial Volunteer Z 13 38 49 N 8 ll 31 Non-Volunteer % 16 22 62 XL 17.468** (4 g) **p (.01 73 Table 13. Knowledge of Time of Ovulation Knowledge of Time of Ovulation Volunteer Status Correct Incorrect N_ 69 41 Volunteer % 63 37 N 27 12 Partial Volunteer % 69 31 N_ 29 21 Non-Volunteer % 58 42 X2- 1.18 (2 __d_f_) 74 The knowledge and practice of the various methods of birth control was analyzed by both chi-square and analysis of variance procedures. The scoring for each item, "1. never heard of it" "2. heard of it but never used" and "3. used", can be thought of as forming ordinal scales of the degree of familiarity with the different methods. The chi-square analyses, presented in Table 14, show significant differences for two of the methods, the IUD and the diaphragm. The analyses of variance presented in Table 15, show only the IUD as having significantly different (p = .009) familiarity for the groups. The chi-square also suggests that nonvolunteers are most likely to have used, and the volunteers most likely to have not even heard of the device. For the diaphragm, the chi-square table would indicate that the partial volunteers were more likely to have used the device than the volunteers or nonvolunteers. Therefore, partial support was found for knowledge and practice of birth control hypothesis 2. A chi-square analysis of the respondents' current birth control status is reported in Table 16. There were no signifi- cant differences in likelihood to be currently using a method of contraception. A chi-square analysis run on the method used by those currently contracepting is presented in the same table. There was a significant difference, p <.05, in the method used for the three groups. This result seems to be largely due to the differences reported above in the groups' likelihood to be using the IUD. 75 Table 14. Chi-squares -— Familiarity/Use of Birth Control Methods Volunteer Status Partial Non— Chi- Variable Volunteer Volunteer Volunteer Square d£_ E N_ % N_ ' Z N; Z P1111 1 O O 0 O 0 0 2 17 92 7 32 4 46 3 16 84 18 82 8 92 2.209 2 n.s. Morning 1 36 33 16 41 14 28 After Pill 2 73 66 23 59 35 70 3 1 1 O 0 l 2 2.445 4 n.s. IUD 1 10 9 2 5 2 4 2 85 77 34 88 32 64 3 15 14 3 8 16 32 12.421 4 .05 Condoms l 7 6 3 8 0 0 2 57 52 20 51 28 56 3 46 42 16 41 22 44 3.646 4 n.s. Diaphragm l l l 2 5 1 2 2 106 96 32 82 49 98 3 3 3 5 13 0 0 13.274 4 .05 Foam, Cream 1 2 2 1 3 l 2 or Jelly 2 73 66 22 56 28 56 3 35 32 16 41 21' 42 2.205 4 n.s Withdrawal 1 8 7 5 13 5 10 2 66 60 26 67 28 56 3 34 31 8 21 17 34 4.558 6 n.s Rhythm 1 2 2 4 10 l 2 2 86 78 30 77 40 80 3 22 20 5 13 9 18 7.114 4 n.s Douche l 2 2 1 3 l 2 2 82 75 33 85 41 82 3 26 24 5 13 8 16 2.723 4 n.s Abstinence l 19 18 9 23 9 18 2 73 66 24 62 30 60 3 18 16 6 15 11 22 1.547 4 n.s Tubal 1 4 4 2 5 5 10 Ligation 2 105 95 37 95 45 95 3 1 1 O 0 0 0 3.454 4 n.s 76 Table 14. Chi-squares -— Familiarity/Use of Birth Control Methods (cont) Volunteer Status Partial Non- Chi- Variable Volunteer Volunteer Volunteer Square .QE ‘p N_ Z N_ Z N_ Z Hysterectomy 1 l 1 2 5 2 4 2 109 99 37 95 48 96 3 0 0 0 0 0 0 2.696 2 n.s. Vasectomy l 2 2 1 3 l 2 2 108 98 38 97 49 98 3 0 0 0 0 0 0 .081 2 n.s NOTE: 1 - never heard of it 2 -— heard about but never used 3 — used Hmva me Zomba mom >bmemmm om mnmn wHHH H.sm H.mo N.Ns onocem N .Nsm N.oue .Nuo 7 Surge?» Onocom Hem .Nuo meadows Hes N.os N.oN N.Nm muocem N N.om s.muas» .ooo SHnrHu anocwm Hem .NN wanton: cosmos N.Nu N.NN N.ss muocem N .NaH .SNm .oNo SHnrHo mnocvm Hwa .wuu wmnsmmo esmasnmms N.oH N.om H.cm onocem N .How H.NNE .Hmo anan muocvm Hem .omo wmnsmmp wowsa Gamma. mum onofiom N .NHa umHHw N.uo N.um N.ao SHnrHo .mHH .bpm Onocom Hem .Nmu 78 Hmch gm. Emmom mam >bmHNmmm om ¢mnHumanm. H.oo H.©N N.o» anocvm N .Hmo .boo .m¢> SHnsHm nuocvm Hma .wom wmnzmma acemp esmmnmou H.NN N.sm H.oo muoasm N .owp N.MNm .Nmo SHner onoavm Hmo .omm Hmva Am. Zomba mom mbmHNmmm om bmemmm om vaHomVHm Hm HN HH Nm Ho wN m.¢mc a s.m. 85 emon Em. UmOHmHoo zmemm muonmmm mam Hmchmmom mocnnmm SHner nrm meHHN N.Ho onocum Hem H.bum Hancmmom om wmnzmmo memnH0¢Hn WMMHMMD N N00 muonmnHom wnomnma . o: 0Hnn: nomnHoH anme UmnHmHom H.00 H.H0 H.H0 onocvm H00 .00N H.00H .N00 Hanammom om >0¢Hn WMMHmMm N 0N0 mmcomnHom wnomnma 0 . on meHHw 0Hum SHnrHs UmOHmHoo H.N0 H.N0 H.00 anocvm H00 .000 H.N0b .N00 HmchmsOm om 0H0: meummo N 0 HNH morooH on 0Hnnr 0 . nomnHoH UmOHmHos H.NN H.N0 H.00 SHner 0.000» .0NN Quorum H00 .00H Hmchmnom om mHmr memmmo N HNm morooH om meHH% 0 . 0H0 0N0 mHum UmnHmHoo H.N0 H.N0 H.N0 SHner . . muocvm H00 .000 Hmchmbom om 0mn€mmm momanmH macomnHoo onocvm N N.0 wnomnma on 0Hnnu SHnoHo nonnHoH UmnHmHoo H.00 H.H0 H.NN nnocvm H00 .0N0 N.000 .000 91 Hmem N0. macaw. uncomnHommH on HomananommH monsoon om Huchmonm AooonHommav 40Hcmnmmn mnmnCm menHmH zomI mocnom om Kmmo 0Hnn5 oomnHoH SHnde UmOHmHom H.0H H.00 H.0N onocvm H00 .0N0 .000 .000 Hmchmsom om unnsmmo HmHm