Coital debut: the role of religiosity and sex attitudes in the add health survey.

Research on adolescent development has shown that an earlier coital debut is associated with increased health risks such as sexually transmitted diseases and early pregnancy and child-bearing. As a developmental milestone, however, 60.5% of high school seniors report that they have had sex (Brener et al., 2002), and fewer than 20% remain virgins until marriage (Sex Information and Education Council of the Unites States [SIECUS], 1999). While age is perhaps the most consistent predictor of coital debut, systematic variations in timing have been documented across gender, race, and social class. For instance, research indicates that 65.9% of African American males have had sex by the 9th grade as opposed to 46.7% and 35.4% of Hispanic and White adolescent males, respectively (Youth Risk Behavior Surveillance System [YRBSS, 2002]). While adolescent boys engage in more sexual activity than girls (DeGaston, Weed, & Jensen, 1996), gender differences are more pronounced for African American than for Caucasian adolescents until late adolescence, when girls seemingly "catch up" to boys (Bearman & Bruckner, 2001).

One important context that influences the timing of this transition is a youth's religious context. In a review of over 250 studies conducted between 1980 and 1999, Kirby (1999) identified 13 clusters of antecedents of sexual risk-taking (e.g., initiation of sex, number of partners, frequency, contraceptive use) that ranged from community-level to individual-level factors. One of the clusters--attachment to religious institutions--described the protective nature of religiosity. Whether measured at the community level, family level, or individual level, religiosity was associated with delaying the initiation of sexual intercourse and with reporting fewer sex partners.

In spite of the importance of religious context to adolescent sexual decision making, a theoretically and methodologically rigorous empirical literature has been seemingly piecemeal and slow to develop (see Wilcox et al., 2001, for a critical review of published research on adolescent religiosity and sexual behavior). However, current interest in faith-based initiatives and the role of faith-based organizations in addressing social problems related to adolescent sexual behavior (e.g., teen pregnancy and sexually transmitted disease) suggests a need for serious, immediate, and focused empirical attention to the role of religiosity in the sexual decision making of adolescents. Therefore, the purpose of this project was to test a model based on social control theory (DeLamater, 1981, 1989) and sexual socialization theory (Reiss, 1989) that hypothesizes both direct and indirect effects of religiosity on adolescent sexual decision making. Specifically, we tested a hypothesis that individual-level religiosity influences adolescent coital debut via its impact on inhibitory sexual attitudes.

In the following section, we first summarize the few longitudinal studies that have examined the role of religiosity in delaying coital debut. Then, we locate our hypothesized model within the more general empirical literature on the correlates of adolescent coital debut. Finally, we discuss social control theory and sexual socialization theory as applied to sexual decision making. In this way, we lay the theoretical groundwork for testing the hypothesis that negative sex attitudes may account for some (if not all) of previously observed associations between religiosity and adolescent coital debut.

OVERVIEW OF LITERATURE

To date, a meager number of longitudinal studies indicate that religiosity, variously defined as religious affiliation, church attendance, self-reports of the importance of religion, or a composite of these, delays coital debut (see Rostosky et al., in press, for a more extensive review of these). When examined by gender, race, and social class, however, findings indicate that the association between religiosity and coital debut is far less straightforward. In a majority of studies, White adolescent girls who score high on religiosity (however defined) delay coital debut relative to their nonreligious peers. The associations for boys and for African Americans, however, are often inconsistent across studies. For instance, in a sample of low-socioeconomic-status (SES) adolescents from the National Longitudinal Survey of Youth, White and African American adolescents of both genders who participated in religious services and who had friends who attended church were more likely to delay coital debut (Mott, Fondell, Hu, Kowaleski-Jones, & Menaghan, 1996). Miller et al.'s (1997) analysis of three waves of the National Survey of Children, however, failed to find a relationship between religious participation and delayed coital debut for either boys or girls. In a recent analysis of the Add Health data, Bearman and Bruckner (2001) reported that religiosity delayed coital debut for White, Asian, and Hispanic middle and late adolescents, but had no effect on Black adolescents.

In terms of social status differences, previous findings suggest a positive association between parental education and occupational status and age of coital debut (Miller, 1998) for boys in some studies (e.g., Ku, Sonenstein, & Pleck, 1993) but only for girls in other studies (e.g., Bearman & Bruckner, 2001). Brewster (1994) reported that adolescent females whose mothers attended college were more likely to delay coital debut. Yet, in another sample of 926 high school students, parent education was not associated with the virginity status of boys or girls (Feldman et al., 1997). Taken together, the studies reviewed above lead us to formulate a model of the differential influence of religiosity on sexual attitudes and coital debut, varying by gender, race, and parental education status. Because much of the research has been atheoretical, our intent was to situate our model within the framework of social control and sexual socialization theories.

SEXUAL SOCIALIZATION AND SOCIAL CONTROL THEORIES AND ADOLESCENT SEXUAL BEHAVIOR

Religion--along with peers, parents, and the media--is a primary socialization agent of children and adolescents (Wallace & Williams, 1997). Traditional religious institutions in America promote particular sexual ideologies (abstinent, procreational, celibate, or perhaps relational in some less conservative religious traditions) that are intended to assert social influence and control over sexual behavior by prescribing and proscribing (rewarding and punishing) the who, what, when, where, and how (cultural scenarios and behavioral standards) of sexual activity. In contrast to these generally prohibitive sexual ideologies, American popular culture and mass media often are understood to actively promote a sexual ideology characterized by pleasure (DeLamater, 1989).

Reiss (1989) and DeLamater (1989) note that sexual ideology and sexual standards vary across subcultures. For instance, males and females have traditionally been socialized into different sexual ideologies, with females often sanctioned more severely for adopting pleasurable or recreational sexual ideologies. Reiss (1989) maintains that this double standard of sexual socialization and accompanying conflicted and anxious attitudes often work to prevent a positive approach to sexuality, interfering with the preparation of youth for responsible sexual choice. In sum, gendered power relations have often thwarted the acceptance of a fully equal sexual role for women, which "makes us much more likely to behave in ways that produce unwanted outcomes such as disease or pregnancy" (Reiss, 1989, p. 26).

Moreover, the religious institutions in some subcultures (e.g., African American) appear to constitute an even more powerful socialization agent. Thus, religious groups in subcultures with fewer resources and less power than more privileged groups may socialize members differently with respect to sexual ideologies and standards. Brown (1985), for instance, concluded that Black religious institutions are an important influence on Black adolescent girls' sexual behavior. One recent national survey reported that 55% of African American youth say that religion is very important to them, compared to only 24% of White youth (Johnston, Bachman, & O'Malley, 1999). Yet, 30% of African American boys reported having sexual intercourse before they were 13 years old, compared to 12% of White boys and 23% of Latino boys (Centers for Disease Control and Prevention [CDC], 2000). The mechanisms that account for African American males being more religious yet more sexually active at earlier ages await closer empirical examination.

Consistent with social control and sexual socialization theories, the reinforcement of specific sex attitudes or beliefs may be one potential mechanism that at least partially accounts for associations between religiosity and coital debut. In samples of adolescents, religiosity has been associated with more conservative attitudes and beliefs about premarital sex (Ku et al., 1993; Miller & Olson, 1988; Werner-Wilson, 1998). Likewise, other studies have demonstrated a positive correlation between negative attitudes toward premarital sex and the timing of coital debut (e.g., Miller, Christensen, & Olson, 1987; Thornton & Camburn, 1989). Relying on correlations, however, together with the cross-sectional design and selective samples in many of these studies, has precluded drawing conclusions appropriate only to rigorous empirical tests of larger longitudinal samples. This study is meant to comprise such a test. Toward this end, we drew on two waves of nationally representative data to examine the extent to which adolescents' positive and negative (or inhibitive) attitudes and beliefs about sexual intercourse account for the effects of religiosity on coital debut.

Additionally, attention has been increasingly paid to the "virginity pledge" movement that has become popular recently among certain segments (primarily evangelical Protestants) of adolescents. Signing a virginity pledge implies endorsement of a prohibitory attitude toward engaging in sexual intercourse outside of marriage. Others, however, have noted that in some settings, normative pressure may be the motivating factor for pledge signing rather than internalized attitudes and values. Yet regardless of motivation, previous analyses have indicated that White, Asian, and Hispanic pledgers of both genders were less likely than nonpledgers to lose their virginity by the following data collection approximately one year later (Bearman & Bruckner, 2001). Here we examined the extent to which the effect (on coital debut and virginity maintenance) of signing a pledge is accounted for by adolescents' beliefs or attitudes about engaging in sexual intercourse.

Finally, we note that previous studies have often failed to take into consideration the dyadic nature of sexual intercourse. That is, to choose to engage in sexual intercourse, one must have opportunity to do so. Theoretically, coital debut requires a willing partner. While researchers have documented an association between involvement in adolescent romantic relationships and coital debut (Feldman et al., 1997; Herold & Goodwin, 1981; Miller, McCoy, & Olson, 1986; Miller et al., 1997), strikingly few longitudinal studies have accounted for the availability of potential partners in predicting the timing of coital debut (see Bearman & Bruckner, 2001, for an exception). By including in our model the number of reported romantic partners, we examined the influence of a typical source of sexual opportunity on adolescent coital debut.

HYPOTHESES

On the basis of social control/sexual socialization theory and previous empirical studies reviewed above, we posed several hypotheses concerning coital debut in adolescents. To begin with, four "beliefs about sexual consequences" collectively measured adolescents' sexual standards or ideology. We assumed that particularly for devoutly religious adolescents, sexual ideology primarily concerns values of abstinence and procreation rather than pleasure and recreation. Therefore, we hypothesized that religiosity indirectly delays coital debut through beliefs about the negative consequences (e.g., guilt, embarrassment of pregnancy, perceived risk of disease) of adolescent sexual behavior. Conversely, less religious adolescents may be more influenced by a sexual ideology that is recreational and/or relational so that beliefs about the positive outcomes of engaging in sexual intercourse (pleasure and status enhancement) would predict earlier coital debut. We also hypothesized that these beliefs or attitudes would fully account for the "pledge effect" that has been demonstrated in prior analyses. Based on studies reviewed above that document differential sexual socialization norms in White, Latino, Asian, Native American, and African American populations and among males and females, we expected that the effects of religiosity and sex attitudes would vary by gender and race-ethnicity. Specifically, we hypothesized that beyond the effects of religiosity, negative beliefs about engaging in sexual intercourse would decrease the likelihood of coital debut among adolescent females of all races and ethnicities, and that positive beliefs about sexual intercourse would increase the likelihood of coital debut, particularly for non-White males.

METHOD

Participants and Procedures

We used data from Waves 1 and 2 of the National Longitudinal Survey of Adolescent Health (Add Health) in-home interviews to test the proposed model. Using the Quality Education database, which lists all high schools in the United States, the researchers selected a stratified sample of 80 high schools with probability proportionate to size. These schools were stratified by region, urban location, school type (public, private, or parochial), ethnic diversity, and enrollment size so the sample of schools ranged in size from fewer than 100 students to more than 3,000 students from urban, suburban, and rural locations. A feeder school (typically a middle school) for each high school was also identified and selected, which yielded a pair of schools for each of the 80 communities. Nearly 80% of the schools contacted agreed to participate, and those that refused were replaced with another pair of schools. Thus, the schools and the students in them are representative samples.

In this analysis, we used the in-home interview data collected from adolescents and their parents between April and December 1995 (Wave 1) and again from only adolescents between April and August 1996 (Wave 2). Further details about the sample and methods have been documented elsewhere (e.g., Bearman, Jones, & Udry, 1997). The two in-home interviews included measures of sensitive health-risk behaviors, including sexual behaviors. For all sensitive health-status questions, audio computer-assisted self-interviewing (ACASI) technology was used to reduce response bias. That is, adolescents listened to questions through earphones and directly entered their responses on a laptop computer. This procedure eliminated the need to interact with the interviewer. Only those adolescents who indicated that they had not engaged in sexual intercourse by Wave 1 were selected to comprise the sample for these analyses predicting coital debut. Furthermore, only adolescents ages 15 and older were asked the sex attitude questions. Therefore, only adolescents for whom were available complete information, sample weights, and parent data on all model variables at both waves were retained for analysis. We deleted cases with missing data on the variables of interest using listwise deletion, leaving a total sample of 3,691 available for testing the hypothesized model.

Measures

Coital debut. The dependent variable for this study was coital debut status reported at Wave 2. The adolescents responded to the following question at both waves of data collection: "Have you ever had sexual intercourse? When we say sexual intercourse, we mean when a male inserts his penis into a woman's vagina." Positive responses were coded 1 and negative responses were coded 0.

Demographic variables. Demographic variables included in the analyses were Wave 1 self-reports of age, which in the subsample for these analyses ranged from 15 to 21 years. Ethnicity and race were also based on self-reports and were dummy coded such that Caucasian adolescents were contrasted with the categories African American/Black, Latino/Latina, Asian American, and Native American. In addition, we ascertained mother's education level from the parent respondent (0 = no college degree and 1 = college degree) and included it in light of previous findings reviewed above regarding the positive association between parent education and later age of coital debut.

Romantic relationships. Adolescents answered questions about their romantic relationships with up to four specific partners in the previous 18 months. Thus, the range of reported number of romantic partners was coded 0 to 4. We considered this variable an indicator of opportunity to have sex based on associations between romantic relationships and sexual behavior in the literature cited above.

Religiosity. Adolescents' religiosity scores were based on their Wave 1 responses to a 3-item religiosity scale. The items comprising the scale were the frequency of attendance at religious services, frequency of attendance at religious youth activities, and self-rated importance of religion. For all three items, participants responded on a 4-point Likert-type scale ranging from 1(never) to 4 (once a week or more) for attendance items and 1 (not at all important) to 4 (very important) for the importance of religion item. The sum of responses to these items constituted each adolescent's religiosity score; scores ranged from 3 to 12. Thus, higher scores indicated higher religiosity. The religiosity scale demonstrated adequate internal consistency for the entire subsample as well as for girls and boys separately ([alpha] = .69, .70, and .69, respectively).

Attitudes and beliefs about sexual intercourse. Sex attitudes at Wave I were assessed using four scales created via factor analysis from a series of questions about respondents' perceptions of the likely outcomes of engaging in sexual intercourse. The survey questions, factors, and factor loadings are detailed in Table 1. For all sex attitude items, adolescents responded on a 5-point scale ranging from 1 (strongly agree) to 5 (strongly disagree). We computed factor scores by multiplying the individual factor weight (for a given measure) by the response value on that measure from each individual. These weighted responses were then summed to create the factor score.

To determine the factor structure, we analyzed the 12 items relevant to sex attitudes with principal components factor analysis with a varimax rotation to create orthogonal factors. Using the criterion of eigenvalue greater than 1.0, four factors emerged that accounted for 67% of the variance. All items loaded on one factor with a factor loading of 0.56 or greater. Items loaded on factors in a manner that made sense conceptually. We used results of the factor analysis to create different sex attitude scales in accord with the item makeup of each factor. The four scales included perceptions that sexual intercourse would result in negative emotions (Factor 1, [alpha] = 0.68), perceptions that sexual intercourse would result in positive emotions (Factor 2, [alpha] = 0.79), perceptions that pregnancy would lead to negative outcomes (Factor 3, [alpha] = 0.85) and perceived chances of incurring negative health consequences (Factor 4, [alpha] = 0.79). Higher scores for each factor were indicative of stronger beliefs that sexual intercourse would incur specific positive and/or negative consequences.

Pledge status. Adolescents responded to the question, "Have you taken a public or written pledge to remain a virgin until marriage?" We coded positive responses 1 and negative responses 0.

RESULTS

We performed correlational analyses to assess associations between independent and dependent variables for males and females separately. A Bonferroni correction factor was employed to account for the large number of cases and correlations. Next, we fit a series of hierarchical logistic regression models predicting coital debut at Wave 2. All models were run separately for each gender based on previous research documenting gender differences in timing of coital debut. In a baseline model, we tested the contribution of age, ethnicity, mother's education level, and number of romantic partners. Then, to test our hypothesis that religiosity would add uniquely to the model fit, we added this variable to the second model. To test the hypothesis that sexual attitudes would contribute to the model beyond the effects of religiosity, we next added the four sex attitude factors to Model 3. Then, to test the hypothesis that these sex attitudes accounted for the associations between pledging status and coital debut, we entered the pledging status variable in Model 4. Finally, we tested interactions between ethnicity and religiosity, and ethnicity and pledging status, based on theoretical and empirical scholarship on differential sexual socialization norms among ethnic communities. To account for survey design effects, we ran the models using STATA (Version 7.00), a statistical software package that can account for all three of Add Health's design effects (unequal probabilities of selection from region, school, and classroom). Thus, the resulting estimates are representative of adolescents ages 15 and up who had not experienced coital debut at the time of the first interview (1995).

Table 2 presents descriptive data on the variables of interest in this study. Of those adolescents who reported never having had sexual intercourse at Wave 1, 23% of the males and 25% of the females reported that they had engaged in sexual intercourse by Wave 2. At Wave 1, 52% of the males and 41% of the females reported having no romantic partner in the previous 18 months. Thirty-three percent of the males and 40% of the females reported having one romantic partner. Only 1% of the adolescents reported four or more romantic partners.

Significant gender differences were found in religiosity and sex attitude scores (Table 2). Specifically, adolescent females reported being more religious and anticipated more negative emotional outcomes as a result of engaging in sexual intercourse. At Wave 1, a significantly greater proportion of adolescent females (22%) than males (13%) had reported that they had signed pledges to remain virgins until marriage. Male adolescents, on the other hand, anticipated more positive emotional outcomes and yet perceived more negative health outcomes from engaging in sexual intercourse.

Intercorrelational analyses for females and males are displayed in Table 3. For both males and females, coital debut at Wave 2 was moderately associated with the number of romantic partners reported at Wave 1. Also, we found a moderate negative association between sex attitude Factor 1 (anticipation of negative emotions) and coital debut. For both males and females, significant positive associations of low magnitude were found between coital debut and being less religious, anticipating more positive emotional outcomes after sexual intercourse, and anticipating few negative outcomes of pregnancy. For females only, coital debut at Wave 2 was significantly but weakly associated with mother having less than a college-level education. For males only, coital debut was weakly associated with Wave 1 higher levels of anticipated negative health consequences of having sexual intercourse.

Religiosity was also moderately correlated with more negative (r = 0.28 males, r = 0.24 females) and weakly correlated with less positive (r = -0.11 for males and females) beliefs and attitudes about sexual intercourse. Furthermore, religiosity was weakly associated with anticipating more negative pregnancy outcomes (r = 0.11 males, r = 0.12 females) and with anticipating fewer negative health outcomes (r = -0.15 males, r = -0.10 females). Religiosity and pledging were moderately correlated (r = 0.27 males, r = 0.25 females). Finally, pledging at Wave 1 was negatively but weakly associated with coital debut for both males and females. Moderate correlations, however, were found between pledging and some of the sex attitude factors. Specifically, pledging was significantly related to anticipating more negative outcomes (r = 0.25 males, r = 0.19 females). Pledging was also weakly associated with anticipating less positive outcomes (r = -0.15 males, r = -0.13 females) of sexual intercourse.

Tables 4 and 5 present the weighted results of logistic regression analyses predicting coital debut at Wave 2 for adolescent males and females, respectively. The baseline model (Model 1) for males (Table 4) presents odds ratios for the demographic variables and the number of romantic partners. In this model, each year of age increased the odds of coital debut by 21%. Furthermore, the odds of coital debut were more than doubled if the adolescent male was African American compared to Caucasian, whereas the odds were reduced by 55% if the adolescent male was Asian American compared to Caucasian. For adolescent boys whose mothers had at least a college education, the odds of coital debut were reduced by 36%. This effect, however, was no longer significant once religiosity entered the model. For each additional romantic partner reported at Wave l, the odds of coital debut at Wave 2 increased by 89%.

Models 2, 3, and 4 added religiosity, sex beliefs, and then pledging status at Wave 1. Among these adolescent males, each unit increase in religiosity reduced the odds of coital debut by 12%. After accounting for the effect of religiosity, anticipation of negative emotional outcomes of sexual intercourse further reduced the odds of coital debut by 30%. After accounting for the effects of this sex belief, having pledged to remain a virgin until marriage had no significant effect. We can surmise from this table, then, that religiosity has both direct and indirect effects on coital debut among males. That is, religiosity operates at least partially through anticipated negative emotional consequences of coitus, since its influence is slightly reduced when this sex belief factor is entered into the model. In turn, pledging has no effect beyond its association with being religious and anticipating negative emotional outcomes of engaging in sexual intercourse. In the final models, we tested the effects of interactions between race (African American) and both religiosity and pledging on coital debut. In this sample of male virgins (with the proviso that a considerable majority of African Americans had been left-censored or dropped from the analysis because they experienced coitus prior to Wave I), both greater religiosity and pledging appeared to be modestly positively correlated with sexual debut for African American males when compared with White non-Hispanic males and less religious or nonpledging African American males.

Results of the baseline model for females (Table 5) showed that the odds of coital debut increased by 26% for each year of age reported at Wave 1. African American girls in contrast to Caucasian girls were 67% more likely to experience coital debut between waves. Having a mother with at least a college education reduced the odds of coital debut by 57%. Finally, the odds of coital debut increased by 65% for every romantic partner reported at Wave 1.

In addition to these demographic variables and number of romantic partners, a unit increase in religiosity reduced the odds of girls' coital debut by 16%. Adding the sex attitude factors reduced the effect of religiosity only slightly, indicating that religiosity still has a significant direct effect on coital debut. However, for adolescent girls, anticipating more negative emotional consequences to sexual intercourse reduced the odds of coital debut by 26%, even after accounting for religiosity. Furthermore, after accounting for religiosity, the odds of coital debut increased by 8% among adolescent girls who anticipated more positive emotional consequences to engaging in sexual intercourse. Pledging, beyond the religiosity and sex attitude variables, did not significantly contribute to explaining the odds of coital debut, nor did interactions between race (African American) and both religiosity and pledging.

DISCUSSION

Delaying adolescent coital debut is a goal of many parents, educators, and government officials, the latter having recently invested millions of dollars in abstinence-only sex education programs that meet strict federal guidelines but whose effectiveness remains undocumented (see reviews in Christopher, 1995; Franklin, Grant, Corcoran, Miller, & Bultman, 1997; Kirby, 2002). The current study used social control and sexual socialization theory to test a model in which religiosity delays coital debut directly and indirectly through its association with prohibitory sexual attitudes. While previous studies have similarly examined associations among these variables (e.g., Thornton & Camburn, 1989), this study extends previous research that has been primarily atheoretical, correlational in analysis, and limited to single-item measures of religiosity and sex attitudes (Wilcox et al., 2001).

Our findings lend support to the hypothesis that religiosity has direct and indirect effects on coital debut. That is, for males and females, religiosity reduces the likelihood of coital debut, even when controlling for demographic characteristics such as age, race, and parent education, and for the availability of romantic partners. These findings are consistent with other longitudinal studies and with recent polling data. For instance, Moore et al. (1998) found that 44% percent of adolescent females who report that they have not engaged in sexual intercourse cited religious or moral beliefs as their motivation for abstinence.

Religiosity appears to indirectly affect coital debut through a sexual ideology or belief system founded on anticipated negative consequences of engaging in sexual intercourse. Specifically, the effects of religiosity on the odds of coital debut are slightly reduced after accounting for adolescents' beliefs that engaging in sexual intercourse will lead to negative emotional consequences such as guilt, loss of respect from one's partner, and/or the anticipation that having sex will emotionally upset one's mother. While the addition of religiosity and then sex attitude factors resulted in significant model improvement in this study, the further addition of the pledge variable did not. In other words, beyond the direct and indirect effects of religiosity and sex attitudes, taking an abstinence pledge appears to have no additional significant effect on coital debut among a sample of virgins (who are at least 15 years old).

These findings are consistent with Rohrbaugh and Jessor's (1975) hypothesized mechanisms of social control. Specifically, they hypothesized that religion helps "control" adolescent behavior by embedding the individual in "an organizing sanctioning network," which sensitizes adolescents to and then reinforces conservative behavior through the creation of an "obedience orientation" to religious authority (p. 317). These findings are also consistent with Sack, Keller, and Hinkle's (1984) study of the association between sexual guilt (measured by the 39-item Mosher Sex Guilt Inventory), religiosity, and sexual behavior. Sack et al. (1984) found that in a sample of college students, sexual guilt was directly and positively correlated with religiosity, and both variables were associated with not having engaged in sexual intercourse. In addition to anticipated negative emotional consequences, anticipated positive emotional consequences (i.e., pleasure-producing) increased the likelihood of coital debut for adolescent females, but not males. Together, these findings lend support to the theoretical notion that religiosity impacts sexual behavior at least partially by producing and reinforcing a sexual ideology that is prohibitive and/or negative toward adolescent sexual intercourse.

Understanding the impact of specific sexual ideologies or beliefs may have practical implications as well as theoretical value. For instance, surprisingly, fears of the negative consequences of pregnancy or sexually transmitted diseases did not significantly reduce the likelihood of coital debut for either boys or girls in this sample. One implication of this is that pregnancy and HIV prevention programs that focus on the "fear factor" may be less effective. Yet, many abstinence-only programs both within and outside the religious context rely on fear of the practical consequences of intercourse as a motivator (Wilcox & Wyatt, 1997). Conversely, since perceiving sex as pleasurable and relationship-enhancing increases the likelihood of coital debut for girls, those abstinence programs (e.g., True Love Waits) that focus on romantic love as a key sexual motivator may be less effective among those girls who are motivated by sexual ideologies that portray coitus as relationship-enhancing.

Many abstinence-only programs include a written or public pledge as part of their intervention to delay sexual intercourse among adolescents. Our findings suggest that for both males and females, the effect of pledging was not significant after accounting for specific negative beliefs concerning sex (which of course are correlated with pledging). This would suggest the considerable influence of sex ideology and beliefs (especially guilt or disrespect-producing concerns) on behavior that is consistent with social control theory. In other words, the results of this study suggest that the influence of the pledge is largely indirect, accounted for by its association with beliefs about the negative emotional consequences of engaging in sexual intercourse.

Ultimately, sexual decisions are personal value choices. Yet, they are molded and shaped by the many social contexts in which adolescents develop. Effective programs directed at reducing sexual risk taking among adolescents must address multiple social contexts that often promote contradictory sexual ideologies. For example, popular teen magazines may extol the virtues of virginity, yet teach girls how to seduce boys, all within the confines of a single issue (Carpenter, 1998; Durham, 1998). While we focused in this study on an oft-neglected context at the individual level (religiosity), we recognize that this context is influenced by and interacts with other important social contexts. Available empirical evidence supports the effectiveness (for some outcomes and for some adolescents) of prevention programs that are multidimensional and multileveled, targeting parents, schools, and communities (including religious communities) as well as the self-efficacy and social skills of individual adolescents (Christopher, 1995; Kelly, 1995; Kirby, 2002).

Other important social contexts that we examined in this study and that affected coital debut were race, mothers' education level (for girls), and the available number of romantic partners. In this sample, including religiosity and sex attitudes in the model did not appreciably diminish the effects of race. African Americans were significantly

more likely than Caucasians to debut between Wave 1 and Wave 2. Interestingly, African American males who were more religious or who had signed an abstinence pledge were significantly more likely to debut. Although we do not pursue an extensive explanation for this finding here, it is an aggravating rather than a protective factor for the coital debut of African American boys--or at least those who have not debuted prior to age 15. This finding is in sharp contrast to studies documenting the protective effect of conservative Protestantism for White girls (e.g., Brewster, Cooksey, Guilkey, & Rindfuss, 1998).

While race and ethnicity effects were more complex for boys, mothers' education level remained significant across models only for adolescent girls. This effect was substantial, reducing the likelihood of adolescent female coital debut by half, an effect that held steady even when religiosity and sex beliefs were introduced into the model. Perhaps mothers who have college degrees provide important role models for their daughters' educational aspirations that in turn serve as deterrents to early coital debut. Other research findings have also documented associations between educational aspirations of both parents and children and later coital debut (Costa, Jessor, Donovan, & Fortenberry, 1995; Hanson, Myers, & Ginsburg, 1987; Lauritsen, 1994).

For both boys and girls, the number of romantic partners significantly increased the likelihood of coital debut. This effect did not diminish with the inclusion of religiosity and sex belief measures. In another study, 10th-grade boys but not girls reported that they perceived themselves as unable to refuse sexual intercourse with a romantic partner (Zimmerman, Sprecher, Langer, & Holloway, 1995). This finding was consistent across race and level of sexual experience. Sexual socialization may promote girls' and prohibit boys' sexual refusal self-efficacy; nevertheless, our findings support other empirical studies that have documented a strong relationship between romantic relationships and coital debut in adolescence (Feldman et al., 1997; Miller et al., 1986; Miller et al., 1997).

Obviously, other social contexts not examined in this study, including peers, parents, schools, and neighborhoods, are also important to understanding adolescent coital debut. While we focused on religiosity at the level of the individual, future studies should test multilevel contextual effects and their interactions in relation to sexual behavior (see Ellison & Levin, 1998). Religion may influence adolescent coital debut, for example, via macro-level sociocultural support for and intolerance of specific ideologies.

Together, these findings lend additional support to the notion that promoting sexual health in adolescence cannot be achieved through "cookie cutter" programming (Christopher, 1995, p. 390) that makes "uniformity assumptions" (Kelly, 1995, p. 348) about the utility of any single intervention. Certainly, effective interventions must extend to multiple social contexts and must cultivate assets and resources at multiple levels. In addition to prevention efforts, basic research on healthy adolescent sexual development and the contexts that facilitate it must proceed despite the many challenges (Irwin & Duncan, 2002)

We also note a few other limitations to this study. First, we did not examine the contribution of other risk behaviors (e.g., alcohol and drug use) to coital debut. These associations have been extensively documented by others (e.g., Costa et al., 1995; Valois, Kammermann, & Drane, 1997). We also did not examine contraceptive behavior in this project, although past research has reported that while religious adolescent girls (both White and African American) delay coital debut, they are less likely to use contraception upon debut (Cooksey, Rindfuss, & Guilkey, 1996; Brewster et al., 1998). Future qualitative and quantitative studies should focus on this phenomenon.

This study focused on coital debut, yet adolescents may abstain from sexual intercourse while engaging in a variety of other sexual behaviors such as anal and/or oral sex that, unprotected, pose health risks (SIECUS, 1999). For instance, Feldman et al. (1997) found that high school students who considered themselves virgins were nonetheless sexually active in ways that put them at risk for sexually transmitted disease. Apparently, among some adolescents, performing oral sex is a strategy used to maintain "technical" virginity and avoid pregnancy. The extent to which highly religious adolescents use this strategy to avoid the perceived negative emotional consequences of violating their religious values is worth examining in future studies.

Finally, in future studies, these questions should be examined developmentally using growth-curve analyses of at least three developmental time points. Wave 3 of the National Longitudinal Survey of Adolescent Health will provide an opportunity for this type of empirical examination. Scholarship on this subject would also benefit from more "thick" description and in-depth qualitative analysis of adolescents' commitments to the sometimes contradictory sexual ideologies promoted by the various sources of socialization in our culture. If we are to fully understand the process of adolescent sexual decision making, we still have much to learn regarding important social contexts, including the religious context, for these decisions.

Table 1. Factor Loadings for Sex Attitude Items Using
Principal Components Analysis With Varimax Rotation

Items                                               F1        F2

"If you had sexual intercourse, afterward
  you would feel guilty."                          0.76     -0.20
"If you had sexual intercourse, your
  partner would lose respect for you."             0.81      0.03
"If you had sexual intercourse, it would
  upset your mother."                              0.56     -0.13
"If you had sexual intercourse, it would
  make you more attractive (to women/men)."       -0.00      0.78
"If you had sexual intercourse, you would
  feel less lonely."                              -0.00      0.77
"If you had sexual intercourse, it would
  relax you."                                     -0.26      0.77
"If you had sexual intercourse, it would give     -0.25      0.69
  you a great deal of physical pleasure."
"If you had sexual intercourse, your friends
  would respect you more."                         0.05      0.62
"If you got (someone) pregnant, it would be
  embarrassing for your family."                   0.17      0.02
"If you got (someone) pregnant, it would be
  embarrassing for you."                           0.17     -0.07
Perceived chances of getting AIDS                  0.06     -0.07
Perceived chances of getting other STDs            0.08     -0.10
% of variance                                     14.6      22.9
Cronbach's [alpha]                                 0.65      0.79

Items                                               F3        F4

"If you had sexual intercourse, afterward
  you would feel guilty."                          0.20      0.10
"If you had sexual intercourse, your
  partner would lose respect for you."             0.02      0.00
"If you had sexual intercourse, it would
  upset your mother."                              0.36      0.08
"If you had sexual intercourse, it would
  make you more attractive (to women/men)."       -0.08     -0.02
"If you had sexual intercourse, you would
  feel less lonely."                              -0.08     -0.05
"If you had sexual intercourse, it would
  relax you."                                      0.07     -0.03
"If you had sexual intercourse, it would give      0.19     -0.05
  you a great deal of physical pleasure."
"If you had sexual intercourse, your friends
  would respect you more."                        -0.14     -0.11
"If you got (someone) pregnant, it would be
  embarrassing for your family."                   0.90      0.02
"If you got (someone) pregnant, it would be
  embarrassing for you."                           0.89      0.04
Perceived chances of getting AIDS                  0.02      0.91
Perceived chances of getting other STDs            0.04      0.90
% of variance                                     15.3      13.9
Cronbach's [alpha]                                 0.85      0.79

Note. Factor labels: F1 (negative emotional outcomes),
F2 (positive emotional outcomes), F3 (negative pregnancy outcomes),
F4 (negative health outcomes). Total variance explained = 67%.

Table 2. Means, Standard Errors, and Percentages for Number of Romantic
Partners, Pledge Status, Religiosity, Sex Attitudes, and Coital Debut
Between Wave 1 and Wave 2 fo  Males and Females (N = 3,291)

Variable                                     Males          95% CI

Coital debut                                  23%         (0.20-0.26)
Romantic partners = 0 (a)                     52%         (0.48-0.55)
Romantic partners = 1 (a)                     33%         (0.30-0.36)
Romantic partners = 2                         10%         (0.08-0.12)
Romantic partners = 3                          4%         (0.03-0.05)
Romantic partners = 4+                         1%         (0.01-0.02)
Pledger (a)                                   13%         (0.11-0.16)
Religiosity (a)                           7.87 (0.12)     (7.63-8.11)
Sex attitude factors
  F1: Negative Emotional Outcomes (a)     3.27 (0.04)     (3.19-3.35)
  F2: Positive Emotional Outcomes (a)    10.53 (0.10)    (10.33-10.72)
  F3: Negative Pregnancy Outcomes         7.14 (0.08)     (6.99-7.30)
  F4: Negative Health Outcomes (a)        3.22 (0.59)     (3.10-3.34)

Variable                                    Females         95% CI

Coital debut                                  25%         (0.22-0.28)
Romantic partners = 0 (a)                     41%         (0.38-0.45)
Romantic partners = 1 (a)                     40%         (0.37-0.43)
Romantic partners = 2                         13%         (0.10-0.15)
Romantic partners = 3                          5%         (0.04-0.06)
Romantic partners = 4+                         1%         (0.01-0.02)
Pledger (a)                                   22%         (0.18-0.26)
Religiosity (a)                           8.39 (0.14)     (8.12-8.67)
Sex attitude factors
  F1: Negative Emotional Outcomes (a)     3.82 (0.04)     (3.74-3.90)
  F2: Positive Emotional Outcomes (a)     8.33 (0.10)     (8.13-8.53)
  F3: Negative Pregnancy Outcomes         7.31 (0.07)     (7.17-7.45)
  F4: Negative Health Outcomes (a)        2.74 (0.46)     (2.65-2.83)

(a) Means for males and females are significantly different from
each other at the .05 level.

Table 3. Correlations for Demographics, Coital Debut, Romantic
Partners, Religiosity, and Sex Attitudes for Females and Males

                           1           2            3          4

 1. Coital Debut         1.00         0.09 **     -0.01      -0.11 *
 2. Age                  0.11 *       1.00        -0.07      -0.02
 3. White               -0.07        -0.06         1.00       0.06
 4. Mom education       -0.06        -0.01         0.05       1.00
 5. Romantic partners    0.25 *       0.04         0.06       0.05
 6. Religiosity         -0.11 *       0.00        -0.06       0.07
 7. Pledge              -0.09 **      0.01        -0.03       0.03
 8. Sex attitude F1     -0.21 *      -0.02         0.02       0.07
 9. Sex attitude F2      0.10 *       0.03        -0.04       0.01
10. Sex attitude F3     -0.15 *      -0.08 **      0.18  *    0.14 *
11. Sex attitude F4      0.08 (b)     0.02        -0.03       0.02

                           5          6            7            8

 1. Coital Debut         0.21 *     -0.13 *     -0.08 ***    -0.15 *
 2. Age                  0.01        0.01        0.02         0.01
 3. White                0.13 *     -0.09 **    -0.06         0.01
 4. Mom education        0.06        0.07       -0.03         0.03
 5. Romantic partners    1.00       -0.04       -0.00        -0.04
 6. Religiosity          0.01        1.00        0.25 *       0.24 *
 7. Pledge               0.00        0.27 *      1.00         0.19 *
 8. Sex attitude F1     -0.09 **     0.28 *      0.25 *       1.00
 9. Sex attitude F2      0.07       -0.11 *     -0.15 *      -0.24 *
10. Sex attitude F3      0.02        0.11 *      0.09 **      0.28 *
11. Sex attitude F4      0.03       -0.15 *     -0.11 *      -0.15 *

                            9         10        11

 1. Coital Debut          0.11 *    -0.12 *     0.07
 2. Age                   0.00      -0.05      -0.02
 3. White                -0.02       0.13 *     0.00
 4. Mom education         0.03       0.15 *     0.00
 5. Romantic partners     0.01       0.00       0.03
 6. Religiosity          -0.11 *     0.12 *    -0.10 *
 7. Pledge               -0.13 *     0.04      -0.09 **
 8. Sex attitude F1      -0.19 *     0.29 *    -0.16 *
 9. Sex attitude F2       1.00      -0.07       0.11
10. Sex attitude F3      -0.03       1.00      -0.08 **
11. Sex attitude F4       0.13 *    -0.12 *     1.00

Note. Correlations above the diagonal are for females; below the
diagonal are for males. Sex attitude factor labels: F1 (Negative
Emotional Outcomes), F2 (Positive Emotional Outcomes), F3
(Negative Pregnancy Outcomes), F4 (Negative Health Outcomes).

* p<0.001. ** p<.01. *** p<.05.

Table 4. Likelihood of Coital Debut at Wave 2 for Males, Odds Ratios
From Logistic Regressions, Weighted (N = 1,799)

Effect                 Range                1            2

                             Demographics

Age                    (15-21)           1.21 *        1.20 *
Black                  (0,1)             2.41 ***      2.75 ***
Latino                 (0,1)             1.38          1.48
Asian American         (0,1)             0.45 *        0.45
Native American        (0,1)             0.84          0.76
Mother's education     (0,1)             0.64 *        0.66
Romantic partners      (0-4)             1.89 ***      1.95 ***
Religiosity            (3-12)             --           0.88 ***
Sex attitude F1        (2.1-10.5)         --            --
Sex attitude F2        (3.5-17.7)         --            --
Sex attitude F3        (2.3-11.7)         --            --
Sex attitude F4        (1.8-9.1)          --            --
Pledge                 (0,1)              --            --
Black x Religiosity                       --            --
Black x Pledge                            --            --

                          Model fit statistics

-2 log likelihood                     1951.6        1922.7
Pseudo [R.sup.2]                         0.09          0.10

Effect                 Range               3             4

                            Demographics

Age                    (15-21)          1.19 *         1.19 *
Black                  (0,1)            2.46 ***       2.47 ***
Latino                 (0,1)            1.21           1.22
Asian American         (0,1)            0.54           0.55
Native American        (0,1)            0.85           0.85
Mother's education     (0,1)            0.67           0.67
Romantic partners      (0-4)            1.87 ***       1.87 ***
Religiosity            (3-12)           0.92 *         0.92 *
Sex attitude F1        (2.1-10.5)       0.70 ***       0.70 ***
Sex attitude F2        (3.5-17.7)       1.04           1.03
Sex attitude F3        (2.3-11.7)       0.95           0.95
Sex attitude F4        (1.8-9.1)        1.03           1.03
Pledge                 (0,1)             --            0.86
Black x Religiosity                      --             --
Black x Pledge                           --             --

                      Model fit statistics

-2 log likelihood                    1863.7         1861.9
Pseudo [R.sup.2]                        0.13           0.13

Effect                 Range               5

                   Demographics

Age                    (15-21)          1.20 *
Black                  (0,1)            0.50
Latino                 (0,1)            1.22
Asian American         (0,1)            0.55
Native American        (0,1)            0.85
Mother's education     (0,1)            0.66
Romantic partners      (0-4)            1.88 ***
Religiosity            (3-12)           0.91 *
Sex attitude F1        (2.1-10.5)       0.70 ***
Sex attitude F2        (3.5-17.7)       1.04
Sex attitude F3        (2.3-11.7)       0.94
Sex attitude F4        (1.8-9.1)        1.03
Pledge                 (0,1)            0.85
Black x Religiosity                     1.21 (a) *
Black x Pledge                          3.90 (a) *

                     Model fit
                     statistics

-2 log likelihood                    1858.2
Pseudo [R.sup.2]                        0.14

Note. White is the contrast category (0). Sex attitude factor
labels: F1 (Negative Emotional Outcomes), F2 (Positive Emotional
Outcomes), F3 (Negative Pregnancy Outcomes), F4 (Negative Health
Outcomes).

(a) Interaction terms were added in separate steps but are
consolidated in the table to avoid redundancy.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

Table 5. Likelihood of Coital Debut at Wave 2 for Females, Odds Ratios
From Logistic Regressions, Weighted (N = 1,892)

Effect                 Range                1            2

                            Demographics

Age                    (15-21)           1.26 ***      1.27 ***
Black                  (0,1)             1.67 *        2.08 **
Latino                 (0,1)             0.66          0.71
Asian American         (0,1)             1.06          1.27
Native American        (0,1)             1.87          1.63
Mom's education        (0,1)             0.43 ***      0.44 ***
Romantic partners      (0-4)             1.65 ***      1.69 ***
Religiosity            (3-12)             --           0.84 ***
Sex attitude F1        (2.1-10.5)         --            --
Sex attitude F2        (3.5-17.7)         --            --
Sex attitude F3        (2.3-11.7)         --            --
Sex attitude F4        (1.8-9.1)          --            --
Pledge                 (0,1)              --            --
Black x Religiosity                       --            --
Black x Pledge                            --            --

                        Model fit statistics

-2 log likelihood                     2099.2        2065.2
Pseudo [R.sup.2]                         0.06          0.08

Effect                 Range               3             4

                                             Demographics

Age                    (15-21)          1.26 **        1.27 **
Black                  (0,1)            1.83 *         1.85 *
Latino                 (0,1)            0.67           0.67
Asian American         (0,1)            1.31           1.31
Native American        (0,1)            1.80           1.77
Mom's education        (0,1)            0.43 ***       0.43 ***
Romantic partners      (0-4)            1.67 ***       1.67 ***
Religiosity            (3-12)           0.87 ***       0.88 ***
Sex attitude F1        (2.1-10.5)       0.74 ***       0.75 ***
Sex attitude F2        (3.5-17.7)       1.08 *         1.08 *
Sex attitude F3        (2.3-11.7)       0.94           0.94
Sex attitude F4        (1.8-9.1)        1.01           1.01
Pledge                 (0,1)             --            0.89
Black x Religiosity                      --             --
Black x Pledge                           --             --

                       Model fit statistics

-2 log likelihood                    2015.5         2012.3
Pseudo [R.sup.2]                        0.10           0.10

Effect                 Range               5

                           Demographics

Age                    (15-21)          1.26 **
Black                  (0,1)            1.80 *
Latino                 (0,1)            0.67 *
Asian American         (0,1)            1.33
Native American        (0,1)            1.75
Mom's education        (0,1)            0.43 ***
Romantic partners      (0-4)            1.67 ***
Religiosity            (3-12)           0.87 ***
Sex attitude F1        (2.1-10.5)       0.75 ***
Sex attitude F2        (3.5-17.7)       1.08 *
Sex attitude F3        (2.3-11.7)       0.93
Sex attitude F4        (1.8-9.1)        1.01
Pledge                 (0,1)            0.90
Black x Religiosity                     1.07
Black x Pledge                          1.12 (a)

                                      Model fit
                                     statistics

-2 log likelihood                    2008.5
Pseudo [R.sup.2]                        0.10

Note. White is the contrast category (0). Sex attitude factor
labels: F1 (Negative Emotional Outcomes), F2 (Positive Emotional
Outcomes), F3 (Negative Pregnancy Outcomes), F4 (Negative Health
Outcomes).

(a) Interaction terms were added in separate steps but are
consolidated in the table to avoid redundancy.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

REFERENCES

Bearman, P. S., & Bruckner, H. (2001). Promising the future: Virginity pledges and first intercourse. The American Journal of Sociology, 106, 859-912.

Bearman, P. S., Jones, J., & Udry, R. J. (1997). The National Longitudinal Study of Adolescent Health: Research design. University of North Carolina at Chapel Hill, Carolina Population Center.

Brener, N., Lowry, R., Kann, L., Kolbe, L., Lehnherr, J., Janssen, R., et al. (2002). Trends in sexual risk behaviors among high school students-United States, 1991-2001. Morbidity and Mortality Weekly Report, 51, 856-859.

Brewster, K. L. (1994). Race differences in sexual activity among adolescent women: The role of neighborhood characteristics. American Sociological Review, 59, 408-424.

Brewster, K. L., Cooksey, E. C., Guilkey, D. K., & Rindfuss, R. R. (1998). The changing impact of religion on the sexual and contraceptive behavior of adolescent women in the United States. Journal of Marriage and the Family, 60, 493-504.

Brown, S. V. (1985). Premarital sexual permissiveness among black adolescent females. Social Psychology Quarterly, 48, 381-387.

Carpenter, L. M. (i998). From girls to women: Scripts for sexuality and romance in Seventeen magazine, 1974-1994. The Journal of Sex Research, 25, 158-168.

Centers for Disease Control and Prevention (CDC). (2000). Youth risk behavior surveillance--United States, 1999. Mortality and Morbidity Weekly Report Surveillance Summaries, 49, 1-96.

Centers for Disease Control and Prevention (CDC), Youth Risk Behavior Surveillance System (YRBS). (2002). YRBS 2001, Data and documentation files [Data file]. Retrieved January 3, 2003, from http://www.cdc.gov/nccdphp/dash/yrbs/data/index.htm Christopher, E S. (1995). Adolescent pregnancy prevention. Family Relations, 44, 384-391.

Cooksey, E. C., Rindfuss, R. R., & Guilkey, D. K. (1996). The initiation of adolescent sexual and contraceptive behavior during changing times. Journal of Health and Social Behavior, 37, 59-74.

Costa, F. M., Jessor, R., Donovan, J. E., & Fortenberry, J. D. (1995). Early initiation of sexual intercourse: The influence of psychosocial unconventionality. Journal of Research on Adolescence, 5, 93-121.

DeGaston, J. F., Weed, S., & Jensen, L. (1996). Understanding gender differences in adolescent sexuality. Adolescence, 31,217-231.

DeLamater, J. (1981). The social control of sexuality. Annual Review of Sociology, 7, 263-290.

DeLamater, J. (1989). The social control of human sexuality. In K. McKinney & S. Sprecher (Eds.), Human sexuality: The societal and interpersonal context (pp. 30-62). Norwood, New Jersey: Ablex Publishing Corporation.

Durham, M. G. (1998). Dilemmas of desire: Representations of adolescent sexuality in two teen magazines. Youth & Society, 29, 369-389.

Ellison, C. G., & Levin, J. S. (1998). The religion-health connection: Evidence, theory, and future directions. Health Education and Behavior, 25, 700-720.

Feldman, L., Holoway, P., Harvey, B., Rannie, K., Shortt, L., & Jamal, A. (1997). A comparison of the demographic, lifestyle, and sexual behavior characteristics of virgin and non-virgin adolescents. The Canadian Journal of Human Sexuality, 6, 197-209.

Franklin, C., Grant, D., Corcoran, J., Miller, P. O., & Bultman, L. (1997). Effectiveness of prevention programs for adolescent pregnancy: A meta-analysis. Journal of Marriage and the Family, 59, 551-567.

Hanson, S. L., Myers, D. E., & Ginsburg, A. L. (1987). The role of responsibility and knowledge in reducing teenage out-of-wedlock childbearing. Journal of Marriage and the Family, 49, 241-256.

Herold, E. S., & Goodwin, M. S. (1981). Adamant virgins, potential virgins and nonvirgins. The Journal of Sex Research, 17, 97-113.

Irwin, C., & Duncan, P. (2002). Health futures of Youth II: Pathways to adolescent health, executive summary and overview. Journal of Adolescent Health, 31S, 82-89.

Johnston, L. D., Bachman, J. G., & O'Malley, P. M. (1999). Monitoring the future: Questionnaire responses from the nation's high school seniors. Ann Arbor, MI: Institute for Social Research.

Kelly, J. A. (1995). Advances in HIV/AIDS education and prevention. Family Relations, 44, 345-352.

Kirby, D. (1999). Looking for reasons why: The antecedents of adolescent sexual risk-taking, pregnancy, and childbearing. Washington, DC: National Campaign to Prevent Teen Pregnancy.

Kirby, D. (2002). Effective approaches to reducing adolescent unprotected sex, pregnancy, and childbearing. The Journal of Sex Research, 39, 51-57.

Ku, L., Sonenstein, F. L., & Pleck, J. H. (1993). Factors influencing first intercourse for teenage men. Public Health Reports, 108, 660-684.

Lauritsen, J. L. (1994). Explaining race and gender differences in adolescent sexual behavior. Social Forces, 72, 859-884.

Miller, B. C. (1998). Families matter: A research synthesis of family influences on adolescent pregnancy. Washington, DC: National Campaign to Prevent Teen Pregnancy.

Miller, B. C., Christensen, R. B., & Olson, T. D. (1987). Adolescent self-esteem in relation to sexual attitudes and behavior. Youth and Society, 19,93 111.

Miller, B. C., McCoy, J. K., & Olson, T. D. (1986). Dating age and stage as correlates of adolescent sexual attitudes and behavior. Journal of Early Adolescence Research, 1, 361-371.

Miller, B. C., Norton, M. C., Curtis, T., Hill, E. J., Schvaneveldt, P., & Young, M. H. (1997). The timing of sexual intercourse among adolescents: Family, peer, and other antecedents. Youth and Society, 29, 54-83.

Miller, B. C., & Olson, T. D. (1988). Sexual attitudes and behavior of high school students in relation to background and contextual factors. The Journal of Sex Research, 24, 194-200.

Moore, K. A., Driscoll, A. K., & Lindberg, L. D. (1998). A statistical portrait of adolescent sex, contraception, and childbearing. Washington, DC: National Campaign To Prevent Teen Pregnancy.

Mott, F. L., Fondell, M. M., Hu, P. N., Kowaleski-Jones, L., & Menaghan, E. G. (1996). The determinants of first sex by age 14 in a high-risk adolescent population. Family Planning Perspectives, 28, 13-18.

Reiss, I. (1989). Society and sexuality: A sociological explanation. In K. McKinney & S. Sprecher (Eds.), Human sexuality: The societal and interpersonal context (pp. 3-29). Norwood, New Jersey: Ablex Publishing Corporation.

Rohrbaugh, J., & Jessor, R. (1975). Religiosity in youth: A personal control against deviant behavior. Journal of Personality, 43, 136-155.

Rostosky, S. S., Wilcox, B. L., Wright, M. L. C., & Randall, B. A. (in press). The impact of religiosity on adolescent sexual behavior: A review of the evidence. Journal of Adolescent Research.

Sack, A. R., Keller, J. F., & Hinkle, D. E. (1984). Premarital sexual intercourse: A test of the effects of peer group, religiosity and sexual guilt. The Journal o]'Sex Research, 20, 168-185.

Sex Information and Education Council of the United States (SIECUS). (1999). Adolescence and abstinence. Retrieved June 13, 2002, from http://www.siecus.org.pubs/fact/fact0001 .html

Thornton, A. D., & Camburn, D. (1989). Religious participation and adolescent sexual behavior and attitudes. Journal of Marriage and the Family, 51, 641-653.

Valois, R. E, Kammermann, S. K., & Drane, J. W. (1997). Number of sexual intercourse partners and associated risk behaviors among public high school adolescents. Journal of Sex Education and Therapy, 22, 13-22.

Wallace, J. M., & Williams, D. R. (1997). Religion and adolescent health-compromising behavior. In J. Schulenberg, J. L. Matts, & K. Hurrelmann (Eds.), Health risks and developmental transitions during adolescence (pp. 444-468)" UK: Cambridge University Press.

Werner-Wilson, R. J. (1998). Gender differences in adolescent sexual attitudes: The influence of individual and family factors. Adolescence, 33, 519-531.

Wilcox, B. L., Rostosky, S. S., Randall, B. A., & Wright, M. L. C. (2001). Reasons for hope: A review of research on adolescent religiosity and sexual behavior. In B. D. Whitehead, B.L. Wilcox, & S. S. Rostosky (Eds.), Keeping the faith: The role of religion and faith communities in preventing teen pregnancy (pp. 31-82). Washington, DC: National Campaign to Prevent Teen Pregnancy.

Wilcox, B. L., & Wyatt, J. (1997, November). Adolescent abstinence education programs: A meta-analysis. Invited address presented at the joint annual meeting of the Society for the Scientific Study of Sexuality and the American Association of Sex Educators, Counselors, and Therapists, Arlington, VA.

Zimmerman, R. S., Sprecher, S., Langer, L. M., & Holloway, C. (1995). Adolescents' perceived ability to say "no" to unwanted sex. Journal of Adolescent Research, 10, 383-399.

Manuscript accepted July 2, 2003

This research used data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Hams, and funded by Grant No. P01-HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle tot assistance in the original design. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC 27516-2524 (www.cpc.unc.edu/addhealth/contract.html).

Dr. Rostosky's work on this project was partially supported by a 2002 University of Kentucky Summer Faculty Fellowship. This research was also funded in part by a grant from the National Institute of Child Health and Human Development under Grant No. R01 HD40428-02 to the Population Research Center, University of Texas at Austin (Chandra Muller, PI). Opinions reflect those of the authors and do not necessarily reflect those of the granting agency.

Address correspondence to Sharon Scales Rostosky, Assistant Professor of Counseling Psychology, 245 Dickey Hall, University of Kentucky, Lexington, KY 40506; e-mail: [e-mail].

Sharon Scales Rostosky

University of Kentucky

Mark D. Regnerus

University of Texas at Austin

Margaret Laurie Comer Wright

University of Kentucky