Female adolescent contraceptive decision making and risk taking.

Johnson, Sharon A. ; Green, Vicki

Although research on the antecedents of adolescent contraceptive usage has focused on a wide variety of areas (e.g., personality and motivation, demography), the present research concerns the adolescent's ability to make decisions pertaining to contraceptive use and how that ability relates to contraceptive risk taking. For a brief review of other areas of study, see Green, Johnson, and Kaplan (1992); for an extensive review of this literature, see Morrison (1985).

Brooks-Gunn and Furstenberg (1989) identified social cognitive ability as a potential perspective for understanding adolescent sexual decision making. Gordon (1990) and Cobliner (1974) identified aspects of adolescent reasoning (formal operational thinking) as relevant to decision making about contraceptives. The literature documents cognitive capacity as an important factor in decision making related to contraceptive use (Rogel, Zuehlke, Peterson, Tobin-Richards, & Shelton, 1980; Smith, Nenney, Weinman, & Mumford, 1982). Two other potentially related factors are adolescent cognitive egocentrism (Cvetkovich, Grote, Bjorseth, & Sarkissian, 1975; Elkind, 1978; Peel, 1971) and familiarity with the content domain in which the decision is being made (Byrnes, 1988; Levine & Linn, 1977).

Green et al. (1992) examined these three factors, cognitive (adolescent reasoning) ability, adolescent cognitive egocentrism, and experience in contraceptive decision making (the decision to use contraceptives), as predictors of decision making in a contraceptive use paradigm. Fifty sexually active, unmarried females, ages 14-19, served as subjects. Cognitive ability and cognitive egocentrism variables were significantly predictive of several of the decision-making variables, while the decision to use contraceptives was not found to be a significant predictor. Forty-one percent of the variance was accounted for in predicting the canonical decision-making variable. Green et al. concluded that one possible reason for the nonsignificance of the experience factor was that a single dichotomous variable was used (using/seeking contraceptives versus not using contraceptives); a more elaborate variable or set of variables might have yielded different results. The literature on adolescent sexuality identifies specific measures of sexual and contraceptive behaviors and attitudes (see Zelnik & Kantner, 1979). These might be of value in predicting contraceptive decision making.

A logical extension of decision making related to contraceptive use would be to focus upon risk-taking behavior--engaging in sexual intercourse without using contraceptives. Although no study directly demonstrates the relationship of contraceptive use decision making to contraceptive use risk taking, Adler, Kegeles, and Irwin (1987) showed that adolescent behavioral intention is related to actual use of contraceptives. Possible predictors of such risk taking have been identified in the research literature. For example, the level of sexual activity is positively related to contraceptive usage (Geis & Gerrard, 1984). Also adolescents demonstrate erroneous beliefs/attitudes about nonuse of contraceptives (Delamater & MacCorquodale, 1979; Kantner & Zelnik, 1972). These factors may directly influence such risk taking. Additionally, certain demographic variables (i.e., age, gender, and ethnic status) are related to unprotected sexual activity (Zelnik & Shah, 1983).

The present research was a modified replication of the Green et al. (1992) study. Cognitive capacity, cognitive egocentrism, experience, and demographic factors were examined as predictors of decision making in a contraceptive use paradigm. Experience was assessed by two questionnaires, one measuring sexual/contraceptive behaviors, the other sexual/contraceptive attitudes. Additionally, the present study examined cognitive capacity, cognitive egocentrism, experience, and demographic variables as predictors of risk-taking behavior--nonsensible use of contraceptives.

Predictions were as follows: Based upon the Green et al. (1992) study, it was hypothesized that cognitive capacity and cognitive egocentrism would be significant predictors of contraceptive decision making. Based upon the literature pertaining to domain content (Byrnes, 1988; Levine & Linn, 1977), it was hypothesized that experience would also be a significant predictor. There was no basis in the literature for a prediction pertaining to age, grade, or ethnic status. Based upon the literature (Delamater & MacCorquodale, 1979; Kantner & Zelnik, 1972; Zelnik & Shah, 1983), it was hypothesized that sexual/contraceptive behaviors and attitudes, age, and ethnic status would be significant predictors of risk-taking behavior.

METHOD

Subjects

Subjects were 60 unmarried female patients, ages 14-18 years, recruited from two adolescent medicine outpatient clinics of a large, urban, city and county general hospital. The clinics served a lower socioeconomic status population. All subjects tested were considered "indigent"--that is, they were required to pay little or nothing for medical care. All subjects were sexually active. Since it was unclear whether prior pregnancy would influence decision making regarding contraceptives, those who had previously been pregnant were omitted from consideration. Thirty subjects were nonusers of contraceptives and thirty were contraceptive users or seeking contraceptives. One third of the subjects were white, one third were black, and one third were Hispanic. The groups were matched on the demographic variables of race and age. Mean age of subjects was 16.2 years.

Materials

In order to assess the level of cognitive functioning, two cognitive tasks, Displaced Volume and PUN, were employed. Both are considered measures of formal operational thinking. However, in a study by Jones and Green (1991), the former discriminated between sixth graders and those in higher grades, the latter discriminated between eighth graders and those in higher grades. Thus, Displaced Volume would appear to measure a lower level of cognitive functioning within formal operational thinking than would PUN.

Displaced Volume (DV). This questionnaire included eight problems with multiple-choice answers. Each required the subject to identify the relevant variables and then mentally manipulate them in order to predict their effect upon the level of water in a pictured container. (Refer to Linn & Pulos, 1983, for further description of the questionnaire and reliability and validity data.) A total score was calculated; a higher number of correct responses indicated higher cognitive functioning.

Puns (PUN). This task required the subject to explain the two meanings implied in each of three puns. One pun was: Elevator companies have their ups and downs. An understanding of this "play on words" type of humor required thought reversal. Responses were coded into one of four categories: (a) response containing two quality meanings of the pun, (b) a response containing two meanings, one or both of marginal quality, (c) a response containing only one meaning, that one being a quality or marginal response, and (d) an inappropriate response or no response. A higher score indicated higher cognitive functioning. Jones and Green (1991) have categorized the PUN responses of 448 children (Grades 5-12). Consensus by two raters, who were experts in the administration of psychological tests, was required to place a response into a category. (See Jones & Green, 1991, for further description of scoring the Puns instrument.)

Personal Fable Questionnaire (Fable). This questionnaire contained 43 items. Subjects were asked to respond to each using a Likert-type scale, ranging from strongly agree (4) to strongly disagree (0). In a study by Green, Morton, Starr, Jones, and Jaynes (1991), using a fifth-through twelfth-grade normative sample, psychometric properties and factorial structure were assessed. Five factors were identified: Rejection of Rules/Impulsivity (RU/IMP), Egocentrism (EGO), Uniqueness (UNIQ), Magical Thinking (MAG), and Independence (IND). Twenty-five items had salient loadings on one of these five factors. (Refer to Green et. al., 1991, for scoring criteria, reliability and validity data, and data related to grade and gender differences.) For four of the five factors (not IND), high scores reflected a high level of cognitive egocentrism. For IND, a high score reflected thoughts regarding independent action. Five scale scores were obtained, each representing the average response to all items loading on that factor. Since the factors were found to be independent, no combined score was used.

Decision-Making Problem. The decision-making problem, as experienced by another female, focused on contraceptive usage in a relationship in which sexual intimacy was anticipated. The procedures used to operationalize the decision-making process were designed according to a model defined by Linn (1980). (Refer to Green et al., 1992, for a description of the specific problem presented.) Factors listed by all subjects were rated on a scale from very immature (1) to very mature (5) by twelve health care professionals from an adolescent medicine clinic. Those with mean ratings of 3 or above were categorized as mature. Seven variables were generated from the decision-making problem: (a) the number of variables the subject listed (NUMV), (b) the proportion of "mature" variables listed (PRMAT), (c) the number of matches between the variables listed by the experimenter and the subject (MATCH), (d) the number of variables the subject used in the solution to the problem (USEDV), (e) whether the subject used her own variables in the solution (OWNV), (f) the number of experimenter's variables the subject used in the solution (EXPV), and (g) the proportion of "mature" variables used in the solution (PMUSV).

In a pilot study, two instruments were developed that measured sexual/contraceptive behaviors and attitudes. Questions previously used in research (Zelnik & Kantner, 1979) were used as items. Refer to Tables 1 and 2 for lists of these items. In a factor analysis study, 42 unmarried females, between the ages of 14 and 19 years, served as subjects. They were recruited through an adolescent medicine outpatient clinic at a southwestern, university-affiliated health sciences center and through introductory psychology classes at a university in the same state. Twenty-four were sexually active and 18 were not. Of the 24 sexually active subjects, 15 used contraception (this included women engaged in intercourse whose partners used condoms). Results from the factor analysis are presented separately for the two questionnaires.

Sexual/Contraceptive Behaviors Questionnaire. An iterated principal axes factor analysis scree plot of eigenvalues indicated that the potential factor solutions ranged from 1 to 3. Principal axes factor analysis with varimax rotation for two- and three-dimensional solutions was applied to the numerically coded data. The most desirable solution appeared to be the two-factor solution. The eigenvalues of the reduced correlation for the two factors were 2.11 and 1.70, respectively. Salience of item loadings on a factor was based upon two criteria: (a) the loading value was .30 or greater, and (b) the difference between the salient loading and the largest loading on the other factor was at least .20. Four of the six items were found to have salient loadings on one of the two factors. The two variables used represent the total score for each of the two factors. These were labeled: (a) previous birth control usage (SBEH1) and (b) sexual activity (SBEH2).

Sexual/Contraceptive Attitudes Questionnaire. An iterated principal axes factor analysis scree plot of eigenvalues indicated that the potential factor solutions ranged from 1 to 3. Principal axes factor analysis with varimax rotation for two- and three-dimensional solutions was TABULAR DATA OMITTED applied to the data. The most desirable solution appeared to be the three-factor solution. The eigenvalues of the reduced correlation for the three factors were 2.48, 1.13, and 1.00, respectively. Salience of the item loadings on a factor was based upon identical criteria to that specified for the Sexual/Contraceptive Behaviors Questionnaire. Eight of the nine items were found to have salient loadings on one of the three factors. The three variables used represent the total score for each of the three factors. These were labeled: (a) partner's attitudes toward contraceptive use (SATT1), (b) myths regarding pregnancy (SATT2), and (c) subject does not want to, or cannot, use contraception (SATT3).

TABLE 2
SEXUAL/CONTRACEPTIVE ATTITUDES QUESTIONNAIRE ITEMS
Item number Scoring Item
 1 True/False When I have sex, usually have not
 planned to.
 2 True/False I want to use birth control but
 usually I cannot.
 3 True/False I want to use birth control but
 usually I do not.
 4 True/False I do not use birth control it
 would be too costly.
 5 True/False My partner does not want to use
 birth control.
 6 True/False My partner does not want me to
 use birth control.
 7 True/False I do not have intercourse
 frequently enough to become
 pregnant.
 8 True/False I am too young to become
 pregnant.
 9 True/False I could not become pregnant.

Procedure

Individuals waiting for medical appointments were asked to participate; all individuals approached agreed. Subjects were tested individually in a private room (approximately 40 minutes). The order of administration of the measures was PUN, Displaced Volume, Fable, decision-making problem, and the behaviors and attitudes questionnaires. The experimenter read the instructions for all instruments and actively worked with each subject to insure clarification if explanations were needed. Subjects were instructed to take as much time as necessary.

TABLE 3
SEXUAL/CONTRACEPTIVE BEHAVIORS QUESTIONNAIRE: FACTOR LOADINGS
FOLLOWING VARIMAX ROTATION FOR FOUR SALIENT ITEMS
FACTOR FACTOR LOADINGS
Item(s) 1 2
Factor 1 Previous Birth Control Usage
 7 -.34 .04
 8 .49 .07
 9 .67 .07
Factor 2 Sexual Activity
 2 .32 .76
TABLE 4
SEXUAL/CONTRACEPTIVE ATTITUDES QUESTIONNAIRE: FACTOR LOADINGS
FOLLOWING VARIMAX ROTATION FOR EIGHT SALIENT ITEMS
FACTOR FACTOR LOADINGS
Item(s) 1 2 3
Factor 1 Partner's Attitudes Toward Contraceptive Use
 4 .53 .26 .18
 5 .96 -.03 -.14
 6 .62 -.00 .16
Factor 2 Myths Regarding Pregnancy
 7 .18 .66 .22
 8 -.19 .40 -.05
 9 .44 .81 .22
Factor 3 Cannot/Do Not Use Contraceptives
 2 .20 -.12 .86
 3 .13 .15 .61

RESULTS

Multiple regression techniques were used to determine whether decision making and risk taking could be predicted by specific variables. Two sets of stepwise multiple regression analyses were performed using the SAS PROC stepwise technique. In the first set, the cognitive egocentrism, cognitive capacity, demographic, sexual/contraceptive behaviors variables, and sexual/contraceptive attitudes variables were used to predict the seven decision-making variables from the contraceptive problem. One predictor variable was found to be significant in predicting two outcome variables, NUMV (number of variables the subject listed) and PRMAT (the proportion of mature variables listed). Two predictor variables were found to be significant in predicting three other outcome variables, MATCH (number of matches between experimenter and subject), EXPV (number of experimenter's variables used), and PMUSV (proportion of mature variables used). Three predictor variables were found to be significant in predicting a sixth, USEDV (number of variables used in the solution). No significant predictors were found for one of the outcome variables, OWNV (subject's use of own variables). See Table 5 for a summary of these regression analyses.

The predictor variables, the outcome variable, and the percentage of variance accounted for were as follows: (a) PUN accounted for 12% of the variance of the outcome variable--number of variables listed, (b) PUN accounted for 8% of the variance of the outcome variable--proportion of mature variables listed, (c) PUN accounted for 20% of the variance and AGE 9% of the variance of the outcome variable--number of matches between variables listed by subject and experimenter, (d) GRADE accounted for 15%, DV 8%, and HISP (Hispanic/non-Hispanic) 6% of the variance of the outcome variable--number of variables the subject used in the solution, (e) PUN accounted for 21% and GRADE 6% of the variance of the outcome variable--number of experimenter's variables used in the solution, (f) UNIQ accounted for 12% and IND 10% of the variance of the outcome variable--proportion of mature variables used in the solution.

In the second stepwise multiple regression analysis, the cognitive egocentrism, cognitive capacity, demographic, sexual/contraceptive TABULAR DATA OMITTED TABULAR DATA OMITTED behaviors variables, and sexual/contraceptive attitudes variables were used to predict SENSIB--"sensible" use of contraceptives (contraceptives used at least 75% of the time when engaging in sexual activity). Two predictor variables were found to be significant in predicting SENSIB. See Table 6 for a summary of this regression analysis. The predictor variables, the outcome variable, and the percentage of variance accounted for were as follows: PUN accounted for 30% of the variance and SBEH1 (previous birth control usage) accounted for 10% of the variance of the outcome variable--SENSIB.

DISCUSSION

The hypothesis that cognitive capacity, cognitive egocentrism, and experience would be significant predictors of contraceptive decision making was partially supported. Cognitive capacity and cognitive egocentrism variables were found to be significant predictors of contraceptive decision making; however, experience, as measured by factors from the sexual/contraceptive behaviors and attitudes questionnaires, was not. Three demographic variables, age, grade, and ethnic status, also were found to be significant predictors. The amount of variance accounted for in the analyses predicting the six outcome variables ranged from a total of 8% to 29%. Generally, the largest amount of variance accounted for was by PUN and/or AGE/GRADE. Additionally, DV, HISP, UNIQ, AND IND were significant predictors. The decision-making variables with the largest amount of variance accounted for were MATCH, USEDV, and EXPV.

Generally, these findings were similar to those in the Green et al. (1992) study, in which cognitive capacity and cognitive egocentrism were found to be predictive of contraceptive decision making. In the Green et al. study, PUN was the only significant cognitive predictor of decision-making variables. In the present study, PUN was the most common predictor. The cognitive egocentrism variables found to be significant predictors in the present study were different from those found in the Green et al. study. However, in addition to the Personal Fable Questionnaire, Green et al. used Elkind's Imaginary Audience Scale (Elkind & Bowen, 1979). Thus, it is impossible to compare the cognitive egocentrism results from these two studies.

Results in the present study were consistent with the literature that subscribes to a developmental model which assumes that, as adolescents become older, they are likely to function at a higher cognitive level. AGE and GRADE accounted for a large proportion of the variance in predicting the decision-making variables. However, additional variance was accounted for by cognitive capacity. Growing older over the adolescent period does not guarantee increased cognitive capacity. In fact, it is a common assumption of the Piagetian model that not all people attain formal operational thinking in the adolescent years, and some people never achieve it (Tomlinson-Keasey, 1972).

An adolescent reasoning variable (PUN or DV) was a significant predictor in every analysis but one. Of these two variables, PUN was the more frequent significant predictor. This finding is consistent with the literature which assumes that certain aspects of abstract reasoning, such as the ability to take more than one perspective, are required for more mature decision making and problem solving (Inhelder & Piaget, 1958; Keating, 1980; Peel, 1971).

Cognitive egocentrism variables, UNIQ and IND, accounted for additional variance in predicting one of the decision-making variables. UNIQ represents thinking such as, "I'm the only one in the world who feels the way I do." IND represents thoughts regarding independent actions, such as "I know what is best for me." The correlation between UNIQ and PMUSV was negative; thus, greater cognitive egocentrism was related to having used a smaller proportion of mature variables in the solution to the problem. The correlation between IND and PMUSV was also negative; thus, more independent thinking was related to having used a smaller proportion of mature variables in problem solving. The finding for uniqueness is logically consistent with the model used; the finding for independence is not. A possible explanation might be that subjects with more independent thought would be less likely to care what the experimenter thought about their performance and less likely to try to match their performance with that of the experimenter. In fact, although not significant, the correlations between IND and the variables of EXPV, USEDV, MATCH, and PRMAT were negative. Notably, the only two positive correlations were those between IND and the variables of NUMV (subject's initial listing of variables) and OWNV (subject's use of own variables in the solution). Further research is needed to ascertain the differential influence of cognitive egocentrism and independence of thought upon such decision making.

One ethnic status variable, HISP, was also found to be a significant predictor. The correlation between HISP and USEDV was positive; Hispanic subjects were more likely to use a greater number of variables in the problem solution than were non-Hispanic subjects (blacks and whites). There is no basis in the literature for interpreting this finding. Further research is needed regarding the influence of ethnic status upon such decision making. However, it is important to note that all subjects in the present study were from low socioeconomic status (SES) groups. The ethnic status finding may be limited to this SES group.

In the present study, MATCH, USEDV, and EXPV were the outcome variables that had the largest amounts of variance predicted by cognitive egocentrism, cognitive capacity, and demographic variables. High scores on these three variables would require a higher level of cognitive functioning. High scores on MATCH and EXPV would require the subject to step back from her own perspective, recognize the experimenter's (or expert's) perspective, and then use the variables generated by the experimenter. High scores on USEDV would require the subject to use a comparatively larger number of variables in solving the problem. These findings are consistent with the prevalent view that adolescent reasoning includes use of multiple variables (Keating, 1980) and with Linn's (1980) view that the ability to generate variables for problem solving as well as the ability to use the experimenter's variables represents a higher level of functioning and a more mature processing of information. The only outcome variable that was not significant in the regression analysis was OWNV. Notably, OWNV did not require a high level of cognitive ability, but rather the mere use of the subject's own listed variables.

The hypothesis that sexual/contraceptive behaviors and attitudes, age, and ethnic status would be significant predictors of risk-taking behavior was only minimally supported. Only one sexual/contraceptive behaviors variable was found to be a significant predictor. Furthermore, cognitive capacity was found to be a significant predictor of contraceptive risk taking.

Sensible use of contraceptives (SENSIB) was predicted by previous birth control usage. Perusal of answers to the nonscaled items on the Sexual/Contraceptive Behaviors Questionnaire yielded information of possible relevance to this finding. In general, contraceptive behavior was found to be consistent over time. When the subject reported using an "effective" method, she was very likely to report using the method consistently. Similarly, when a subject reported using an "ineffective" method, she was very likely to report using it inconsistently. Additionally, nonusers of contraceptives were very consistent in their behavior. The finding that earlier behavior (previous contraceptive usage) predicts later behavior (SENSIB) is consistent with results reported by Green (1985), in which earlier sexual behavior was positively correlated with sexual behavior in late adolescence, while other factors (e.g., parental influences) were not found to be correlated. Such a finding emphasizes the importance of early education regarding contraceptive use.

Pun ability was found to be the best predictor of SENSIB. It is notable that the variable found to be important in decision making was also of importance in predicting risk taking in contraceptive use. The assumption that higher cognitive functioning is related to "better decision making," which in turn is related to less risk taking, is one easily made based on face validity. Yet, with the exception of the Adler et al. (1987) study demonstrating that adolescent behavioral intention is related to actual use of contraceptives, there is no precedent in the adolescent contraceptive literature for this finding. Future research might continue to explore the influence of cognitive capacity, as well as other potentially influential variables, upon adolescent contraceptive risk taking.

The authors gratefully acknowledge the assistance of the medical and nursing staff of the Eastside and Westside Health Clinics of the city and county of Denver, Colorado.

Sharon A. Johnson, Ph.D., Texas Christian University, Harris College of Nursing, Fort Worth, Texas.

Reprint requests to Vicki Green, Ph.D., Department of Psychology, Oklahoma State University, Stillwater, Oklahoma 74078.

REFERENCES

Adler, N. E., Kegeles, S. M., & Irwin, C. E. (1987). Understanding adolescent contraceptive choice: An empirical test. Paper presented at the meeting of the American Psychological Association, New York City.

Brooks-Gunn, J., & Furstenberg, F. F., Jr. (1989). Adolescent sexual behavior. American Psychologist, 44(2), 249-257.

Byrnes, J. (1988). Formal operations: A systematic reformulation. Developmental Review, 8, 66-87.

Cobliner, W. G. (1974). Pregnancy in the single adolescent girl: The role of cognitive functions. Journal of Youth and Adolescence, 3, 17-29.

Cvetkovich, G., Grote, B., Bjorseth, A., & Sarkissian, J. (1975). On the psychology of adolescents' use of contraceptives. The Journal of Sex Research, 11, 256-270.

Delamater, J., & MacCorquodale, P. (1979). Premarital sexuality: Attitudes, relationships, behavior. Madison: University of Wisconsin Press.

Elkind, D. (1978). Understanding the young adolescent. Adolescence, 8, 127-134.

Elkind, D., & Bowen, R. (1979). Imaginary audience behavior in children and adolescents. Developmental Psychology, 15, 38-44.

Geis, B. D., & Gerrard, M. (1984). Predicting male and female contraceptive behavior: A discriminant analysis of groups high, moderate and low in contraceptive effectiveness. Journal of Personality and Social Psychology, 46, 669-680.

Gordon, D. E. (1990). Formal operational thinking: The role of cognitive-developmental processes in adolescent decision-making about pregnancy and contraception. American Journal of Orthopsychiatry, 60(3), 346-356.

Green, V. A. (1985). Experiential factors in childhood and adolescent sexual behavior: Family interaction and previous sexual experiences. The Journal of Sex Research, 21, 157-182.

Green, V., Johnson, S., & Kaplan, D. (1992). Predictors of adolescent female decision making regarding contraceptive usage. Adolescence, 27(107), 613-632.

Green, V., Morton, K., Starr, B., Jones, F., & Jaynes, W. E. (1991). The development of an instrument to measure the personal fable aspect of adolescent cognitive egocentrism. Manuscript submitted for publication.

Inhelder, B., & Piaget, J. (1958). The growth of logical thinking. New York: Basic Books.

Jones, F. A., & Green, V. (1991). Measures of cognitive development in adolescence. Manuscript submitted for publication.

Kantner, J. F., & Zelnik, M. (1972). Sexual experiences of young unmarried women in the U.S. Family Planning Perspectives, 4, 9-18.

Keating, D. P. (1980). Thinking processes in adolescence. In J. Adelson (Ed.), Handbook of adolescent psychology (pp. 211-246). New York: Wiley.

Levine, D. I., & Linn, M. C. (1977). Scientific reasoning ability in adolescence: Theoretical viewpoints and educational implications. Journal of Research in Science Teaching, 14, 371-384.

Linn, M. C. (1980). When do adolescents reason? European Journal of Scientific Education, 2, 429-440.

Linn, M. C., & Pulos, S. M. (1983). Male-female differences in predicting displaced volume: Strategy usage, aptitude relationships, and experience influences. Journal of Educational Psychology, 75, 86-96.

Morrison, D. M. (1985). Adolescent contraceptive behavior: A review. Psychological Bulletin, 98, 538-568.

Peel, E. A. (1971). The nature of adolescent judgement. New York: Wiley-Interscience.

Rogel, M. J., Zuehlke, M. E., Peterson, A. C., Tobin-Richards, M., & Shelton, M. (1980). Contraceptive behavior in adolescence: A decision-making perspective. Journal of Youth and Adolescence, 9, 491-506.

Smith, P. B., Nenney, S. W., Weinman, M. L., & Mumford, D. M. (1982). Factors affecting perception of pregnancy risk in the adolescent. Journal of Youth and Adolescence, 11, 207-215.

Tomlinson-Keasey, C. (1972). Formal operations in females from eleven to fifty-four years of age. Developmental Psychology, 6, 364.

Zelnik, M., & Kantner, J. F. (1979). Reasons for nonuse of contraception by sexually active women aged 15-19. Family Planning Perspectives, 11, 289-296.

Zelnik, M., & Shah, F. K. (1983). First intercourse among young Americans. Family Planning Perspectives, 15, 64-70.