HUM.BEV_Discussion: Values Consistent With Social Work Practice

profilexmamprosf
adolescent_sexual_debut_a_multisystem_perspective.pdf

Journal of Human Behavior in the Social Environment, 19:330–358, 2009

Copyright © Taylor & Francis Group, LLC

ISSN: 1091-1359 print/1540-3556 online

DOI: 10.1080/10911350902787437

Adolescent Sexual Debut: A Multi-System Perspective of Ethnic

and Racial Differences

RICHARD K. CAPUTO Wurzweiler School of Social Work, Yeshiva University, New York, New York

Data from the 1997 cohort of the National Longitudinal Survey

of Youth were used to assess the association between family, self,

proximate extra-familial, and distal extra-familial system factors

and abstinence/virginity and age of first sexual intercourse by

ethnicity and race (N D 1,854). Findings show how and in what

circumstances measures associated with these four factors vary for

Black, White, and Hispanic youth. Findings point to the robustness

of class and gender for each ethnic-racial group on timing of

sexual initiation and of delinquency and negative peer relations

on abstinence among Black and White youth and of religious

affiliation among Hispanic youth.

KEYWORDS Abstinence, adolescent sexuality, ethnicity, families,

race, religiosity, peers, SES

This study focused on four sets of theoretically related factors found in the literature to be associated with abstinence and/or timing of first intercourse (e.g., see Kirby, 2001; Miller, Forehand, & Kotchick, 2000): family (e.g., parent-teen relationship, family cohesion); self (e.g., intellectual achieve- ment); proximate extra-familial (e.g., peer influences, school environment); and distal extra-familial (e.g., unemployment rate, urban residence). These four partially overlapping perspectives or spheres of influence were deemed consistent with the person-in-environment or ecological approach to social work and other helping professions’ practices (Perkins, Luster, Villarruel, & Small, 1998; Smith, 1997). To assist practitioners’ efforts to identify appropri- ate interventions, the relative weights of specified factors within and between these four spheres of influence of adolescent sexual behavior were at issue

Address correspondence to Richard K. Caputo, PhD, Wurzweiler School of Social Work,

2495 Amsterdam Avenue, New York, NY 10033-3299. E-mail: [email protected]

330

Adolescent Sexual Debut 331

for this study. Specifically, the study addressed the following questions of White, Black, and Hispanic adolescents:

1. Which familial, self, proximate extra-familial, and distal extra-familial at- tributes are correlated with abstinence and/or timing of first intercourse among young adolescents?

2. Which attributes are more robustly correlated with abstinence and/or timing of first intercourse among young adolescents across perspectives? a. Which if any family-related attributes are more robust predictors of

abstinence and/or timing of first intercourse than proximate extra- familial attributes?

b. Which self or personal attributes are more robust predictors of absti- nence and/or timing of first intercourse than family-related attributes?

c. Which proximate extra-familial characteristics are more robust predic- tors of abstinence and/or timing of first intercourse than family-related or self-system attributes?

d. Which distal extra-familial factors are more robust predictors of ab- stinence and/or timing of first intercourse than either family-related, self-system, or proximate extra-familial attributes?

LITERATURE REVIEW

In their literature review of 410 studies reported between 1990 and 2004 inclusive, Kirby, Lapore, & Ryan (2005) identified more than 400 factors related to adolescent pregnancy and sexually transmitted disease. Their di- vision into ‘‘risk’’ and ‘‘protective’’ factors, however, though helpful in many ways (e.g., by focusing on factors amendable to change from those that are not or are unlikely to be) was nonetheless deemed too limiting as it masked circumstances under which any given factor served as a risk or a protection. Further, their four themes of (a) individual biological fac- tors; (b) disadvantage, disorganization, and dysfunction in multiple domains; (c) sexual values, attitudes, and modeled behavior in multiple domains; and (d) connection to adults and organizations that discourage sex, unprotected sex, or early childbearing were too broad conceptually. Permissiveness was a good example. Kirby et al. (2005) reported permissiveness of parents as a risk factor, but this could have benefited from a larger theoretically related context to be substantively meaningful (e.g., by contrasting it with other parenting styles such as authoritarian, authoritative, or uninvolved). Further, just as adolescents as individuals are embedded in families both of which are amenable to therapeutic types of interventions, so too are families embedded in larger structural contexts about which things can be done, such as unemployment levels in a community or the socioeconomic status of the families, albeit at policy levels of interventions.

332 R. K. Caputo

Beyond sociodemographic characteristics of age, ethnicity-race and gen- der, my reading of the literature suggested that four sets of theoretically related factors were associated with abstinence and/or timing of first inter- course. These factors, identified as familial, self, proximate extra-familial, and distal extra-familial, were more streamlined or parsimonious than the similar but more amorphous framework Kirby et al. (2005) used to organize their literature review. These four factors or spheres of influence were also consistent with the more theoretically coherent ecological approach to social work and other helping professions’ practices (Perkins et al., 1998; Smith, 1997). Family-related factors were seen as influencing an adolescent’s de- velopment, as were proximate-extra distal factors, and each of these three sets of influences was concomitantly subject to influences found in the larger environment of distal extra-familial factors.

The Family System

The family system comprised three interlocking or overlapping spheres of influence on adolescents: structure (e.g., single-parent), relationships (e.g., close, warm) or interactions (e.g., supportive, or permissive parenting style, doing things together) among members and, following Miller (2002), re- ligion. In regard to structure, adolescents in single-parent families were shown to become sexually active earlier than those in two-parent families, even when controlling for age, ethnicity-race, religiosity, and socioeconomic status (Whitbeck, Yoder, Hoyt, & Conger, 1999). Davis and Friel (2001) reported, however, that only girls living in single-parent households had earlier sexual debut than girls in other households (intact two-parent, step- family, cohabitating family, or lesbian family) or boys in any family structure type.

In regard to family relationships, delayed onset of sexual debut for both boys and girls was found among those whose mothers reported in- creasingly satisfactory relations with their children (Davis & Friel, 2001). Communicative, supportive, and warm parenting styles were also found to be associated with delayed sexual experience among offspring (e.g., Blum, 2002; Miller, 2002). Donenberg, Wilson, Emerson, and Bryant (2002) found that greater degree of perceived parental permissiveness was related to riskier sexual behavior among girls. Longmore, Manning, and Giordano (2001) found that parental monitoring was inversely related to timing of first inter- course, though no effect was found for degree of support or coerciveness.

In regard to religion, links between religiosity (regardless of how it was measured) and adolescent sexual behavior were found to be well established in the literature (Benson & King, 2005; The National Campaign to Prevent Teen Pregnancy, 2005). Relying on Round 6 NLSY97 data, for example, the National Campaign to Prevent Teen Pregnancy (2005) found linkages be- tween religiosity-related measures and whether sexually inexperienced teens

Adolescent Sexual Debut 333

ages 12 to 14 in 1997 had sex (heterosexual vaginal intercourse) before age 18. Strong parental religious beliefs, frequent attendance at religious services, shared family religious activities, and overall religiosity were all associated with males’ and females’ abstinence prior to age 18. When controlling for parental religiosity, however, having strong relations with their mothers de- creased the likelihood of teens’ having sex before age 18. Although this study controlled for sociodemographic factors of age, race, and gender and for family structure, it did not account for peer, community, or other influences beyond that of adolescents’ immediate families. Rostosky, Wilcox, Wright, and Randall (2004) examined 10 longitudinal studies published between 1980 and 2001 for evidence that the religiosity of adolescents was causally related to their sexual behaviors. They found that religiosity delayed the sexual debut of female adolescents, with similar effects for Whites and Blacks. And Rostosky, Regnerus, and Wright (2003; also see Brewster, Cooksey, Guilkey, & Rindfuss, 1998) reported that Black males who had either signed a virginity pledge or were more religious were significantly more likely to debut than both White non-Hispanic or Black males who were less religious and/or who had not signed a pledge.

Religious affiliation was also found to be associated with adolescent sexual activities. Adolescents, especially girls, from more conservative re- ligious denominations (e.g., conservative Protestants and fundamentalists) were reported to begin having sexual intercourse at later ages than those from other religious denominations (e.g., Bearman & Bruckner, 2001; Ros- tosky et al., 2004). Other research (National Campaign to Prevent Teen Pregnancy, 2005) suggested that adolescents of any religious affiliation were less likely to have sexual intercourse than those unaffiliated with a religious group. Earlier studies suggested, however, that religious attendance was a more important antecedent of adolescent sexual behavior than religious affiliation, with greater levels of attendance at religious services associated with later sexual debut (e.g., Thornton & Camburn, 1989) reported that Black males who had either signed a virginity pledge or were more religious were significantly more likely to debut them both White non-Hispanic and Black males who were less religious and/or who had not signed a plege.

The Self System

The self system comprised personal attributes or characteristics, including propensity for socially unacceptable behaviors and cognitive or intellectual capacity. Summarizing results of three nationally representative studies, Al- bert, Brown, and Flanigan (2003) reported that nearly 20% of sexually active youth drank regularly compared to 3% of virgins and 90% of sexually active youth participated in at least 1 of 15 types of delinquent behavior (such as shoplifting and fighting) in the previous year as compared to 69% of virgins. The relationship between substance use and sexual behavior in this and

334 R. K. Caputo

other studies remained unknown owing to unmeasured personal and social behaviors (Rashad & Kaestner, 2004).

Relying on data from the National Longitudinal Study of Adolescent Health (Add Health), Harris, Duncan, and Boisjoly (2002) found that greater cognitive ability, measured by the Peabody Picture Vocabulary Test Revised, was associated with lower risks of first sexual intercourse for boys but not for girls, while controlling for a variety of individual, family, and school en- vironment factors. Family factors, however, were limited to structure. Other measures were dubious. Welfare receipt, used to measure individual level poverty status, invariably excluded many adolescents living in non-welfare working poor families (Caputo, 2007b). There was no direct measure of school-level socioeconomic status. The average of mother’s level of educa- tion in the school served as a proxy measure of school-level socioeconomic status (SES).

Proximate Extra-Family System

The proximate extra-familial system captured influences on adolescent be- havior that extended beyond the family per se but which nonetheless were immanent to an adolescent’s lifestyle and had influence on it. For the most part, this meant peers, whose influences were also found to correlate in both positive and negative ways with early sexual activity. French and Dish- ion (2003) found that deviant-peer involvement (self- and teacher-reported adolescent interactions with misbehaving students, with smokers and friends’ use of substances) was a robust predictor of early sexual involvement among a high-risk sample of adolescents between the ages of 11 and 14. French and Dishion did not account for ethnicity/race. In their study of how neigh- borhood characteristics mediated parental practices on adolescent sexual initiation, Roche et al. (2005) included a single-item measure of deviant peer affiliation, namely how many of an adolescent’s three best friends drink alcohol, smoke cigarettes, and use marijuana. This measure was found to be a robust predictor of first intercourse. Miller et al. (1997) reported a positive relationship between number of friends having had sex by age 16 and likelihood of an adolescent’s having had sexual intercourse for both boys and girls. The National Campaign to Prevent Teen Pregnancy (2005) reported that higher percentages of peers who regularly attended church services were also associated with delayed first sexual intercourse. In their study of abstinence-only programs, Trenholm et al. (2007) reported that friends’ support for abstinence was a significant predictor of future sexual abstinence: the greater the support, the more delayed the onset of sexual activity. Their evidence suggested that promoting support for abstinence among peer networks should be an important feature of future abstinence programs, even though the efficacy of this support eroded as youth moved through adolescence.

Adolescent Sexual Debut 335

Distal Extra Family

The distal extra family system comprised the largest constellation of measures in which family, self, and proximate-family systems measures were embed- ded and which were found to moderate other systems’ effects on adolescent sexuality. Roche et al. (2005), for example, reported that greater parental involvement was related to a lower likelihood of sex initiation only when youths lived in socioeconomically advantaged neighborhoods. In addition, parental decision making, which centered on adolescents’ activities within (e.g., watching television) and outside (e.g., hanging out with peers) of the home was associated with a lower likelihood of sexual initiation for adolescents in disadvantaged neighborhoods but to a greater likelihood of sexual initiation for youth in advantaged neighborhoods. In their study of adolescents 11 to 16 years old in Chicago, Browning, Leventhal, and Brooks- Gunn (2004) reported that neighborhood-level concentrated poverty was a robust predictor of earlier-age first sexual intercourse and accounted for much of the differences found between African-American youth and those of European heritage.

Relying on data about women 15 to 19 years old (N D 1,280) ob- tained from Cycle V of the National Survey of Family Growth (NSFG) and several secondary aggregate data sources, Averett, Rees, and Argys (2002) reported that personal (e.g., years since menarche) and family characteristics (e.g., household structure, religious affiliation) were robust correlates of the decision to have sexual intercourse when also accounting for a variety of neighborhood and other contextual characteristics. Of neighborhood mea- sures, median income and percentage Black of census tracts were found correlated with whether adolescent girls were sexually active in this manner. Median income was negatively related to decisions to have sexual inter- course, whereas living in census tracks with a higher proportion of Blacks lowered the probability of being sexually active.

SOCIODEMOGRAPHIC CHARACTERISTICS

Summarizing results of three nationally representative studies, Albert et al. (2003) reported a positive relationship between age and timing of first inter- course, with 18% to 19% of youth having had sexual intercourse at age 14 or younger. Boys were more likely than girls to have sex at an early age, a finding that has been consistent for quite some time (e.g., Furstenberg, Morgan, Moore, & Peterson, 1987).

Many studies used socioeconomic status as a control, and overall find- ings indicated an inverse relationship between SES and likelihood and timing of first intercourse (e.g., see Hofferth, 1987). In their study based on the 1992 Youth Risk Behavior Survey/Supplement to the National Health Interview Survey, Santelli, Lowry, Brener, and Robin (2000), however, reported that household income was not linearly related to any sexual behavior and that

336 R. K. Caputo

adjustment for SES and family structure had a limited effect on the association between ethnicity/race and sexual behaviors. On the whole, although some evidence of a socioeconomic dimension was found for likelihood of and age at time of first intercourse, findings remained inconclusive (Pedersen, Samuelsen, & Wichstrom, 2003). Although Black males were found consis- tently to have sexual intercourse at earlier ages than other adolescents (e.g., Ramirez-Valles, Zimmerman, & Juarez, 2002), relying on multivariate analy- ses, however, Roche et al. (2005) found no relationship between ethnicity- race and sexual initiation when controlling for sex.

My study investigated the role of aforementioned theoretically linked systems and sociodemographic characteristics on adolescent abstinence from sexual intercourse and age of having sexual intercourse for the first time for White, Black, and Hispanic adolescents. Following Miller et al. (2000) and others (e.g., Perkins et al., 1998), such a multi-system approach was viewed as affording human service professionals, health agencies, parents, and policy makers the best opportunity to help adolescents avoid the neg- ative consequences associated with sexual risk taking. This multi-system approach accords with Bronfrenbrenner’s ecological systems model (1986) theoretically posited as providing a close approximation of the ‘‘real world’’ of adolescent sexuality. As noted earlier, factors from each system were found to influence adolescent sexual activity. The relative influence of specific factors within and across these theoretically linked systems or spheres of influence when controlling for sociodemographic characteristics remained uncertain. In light of findings from previous studies, particular attention was given to family- and proximate extra-familial or peer-related influences when taking into account sociodemographic factors.

This study differed from others by pooling data across waves of a longitudinal data file representing a national sample of youth. Many other studies relied either on one-shot cross-sectional data alone (e.g., Miller et al., 2000; Regan, Durvasula, Howell, Ureno, & Rea, 2004) or used one wave of longitudinal surveys (e.g., Davis & Friel, 2001). Only a few studies used longitudinal data and examined sexual activity across multiple waves of the same subjects (e.g., Rostosky et al., 2003). Only one study (Caputo, 2007a) adopted the multi-system perspective formulated in this study, but that study did not examine factors differentiating sexual debut among White, Black, and Hispanic adolescents, respectively.

METHOD

Data and Sample

This study used data from the 1997 cohort of the National Longitudinal Survey of Youth (NLSY97), a nationally representative sample of U.S. youth (N D

Adolescent Sexual Debut 337

8,984) born between 1980 and 1984 and 12 to 16 years old as of December 31, 1996 (Center for Human Resource Research, 2003). Information about sexual activity, including age of first sexual intercourse, was asked of youth 14 years old and older in survey years 1997 and 1998. In 1999, all youth were asked questions related to sexual activity. This study used rounds 1 to 4 data, through survey year 2000, when Child Trends, Inc. created summary measures of self-reported adolescent sexual activity for 8,022 youth.

The study sample (n D 1,854) was confined to youth ages 13 or 14 as of December 31, 1996 who remained in the survey through round 4 and for whom all baseline and subsequent information was available. In survey year 2000, there were 3,506 youth who met the age-eligibility criterion. No differences were found among these youth between the 146 lost to attrition and those remaining in the population sample in regard to SES, gender, or ethnicity-race. An additional 1,651 eligible youth were lost owing to missing data on one or more measures. In 1997, those in the study sample had slightly higher income-to-poverty ratios (2.9 vs. 2.7) and lower weighted percentages of Blacks (13.9% vs. 17.8%) and urban residents (67.4% vs. 71.9%) than those lost to missing data, but no differences were found by gender. Income, racial, and urban differences were seen as substantively small, and imputing for missing data was not done.

The NLSY97 was particularly suited for this study because it captured through self-reports sexual activity about abstinence and timing of first in- tercourse for all adolescents ages 14 or older in any given survey year. The longitudinal nature of the NLSY97 enabled data to be pooled as younger adolescents turned 14 were added to those who were 14 years or older in 1997. In addition, because the same subjects were interviewed across multi- ple waves, errors due to recall were minimized for virginal adolescents who were 14 years old in the survey year they were asked about the experience of first intercourse. These features of the NLSY97 distinguished the present study from Rostosky et al. (2003) who examined adolescents 15 years old or older and who confined their study of coital debut to that between Waves 1 and 2 of the National Longitudinal Survey of Adolescent Health (Add Health).

Early adolescent sexual activity was deemed particularly important in light of the reported paucity of related studies for this age group and the sobering problems associated with the increased likelihood of bearing chil- dren with greater health problems and the lower levels of parenting skills than those of older adolescents who are sexually active and bear children (Al- bert et al., 2003). In addition, developmental theory suggested that younger adolescents were more closely linked to parental influences or values and more likely to participate in religious activities, given that individuation and cognitive processes are further developed in older adolescents (Steinberg, 2004; Whitbeck et al., 1999).

It should be noted that there has been long-standing criticism of reliance on self-reported retrospective event history data in general and about sexual

338 R. K. Caputo

initiation in particular (Kahn, Kalsbeek, & Hofferth, 1988; Roche et al., 2005). Accuracy of the data on the onset of sexual activity has been found to decline with duration of recall, whereas short durations of recall suggested conceal- ing sexual activity from interviewers to the extent such activity might have been viewed as a sensitive topic (Wu, Martin, & Long, 2001). Nonetheless, as Kahn et al. reported, consistent estimates of background correlates of adolescent sexual activity were found across three earlier national surveys especially for older teens but also for younger teens as well.

Measures

There were two dependent measures: sexual behavior status as of survey year 2000, when Child Trends, Inc. created the summary measure, and, for those youth who reported ever having had sexual intercourse, age at sexual debut. Both measures were obtained and pooled from self-reports of youth in survey years 1997, 1998, 1999, and 2000, by which time all youth were age 14 or older. Sexual behavior status was a nominal level measure with the following mutually exclusive categories: virgin, having had sexual intercourse by age 14, and having had sexual intercourse at age 15 or older.

FAMILY SYSTEM

The family system comprised three subcategories: structure, relations, and religion. Family structure included measures of whether adolescents at age 14 lived with both of their parents and whether adolescents lived in the same household. Each was coded such that 1 D yes and 0 D no.

Family relations included measures for family routines, parental knowl- edge or ‘‘monitoring,’’ parenting styles, and parent-teen relationship. Child Trends, Inc. (Moore, McGroder, Hair, and Gunnoe, 1999) created the Index of Family Routines, which comprised four items (e.g., number of days per typical week youth ate dinner with the family). Scores could range from 0 to 28, with higher scores signifying more days spent in routine activities with the family. As this measure is an index and not a scale, internal consistency- reliability is not applicable. That is, it is not assumed that the frequency of one family routine should necessarily be correlated with the frequency of another family routine (see Moore et al., 1999, p. 43).

The measure parenting styles was also created by Child Trends, Inc. (Moore et al., 1999). It combined measures of supportiveness and permis- siveness in such a way as to produce the following four types: Uninvolved (permissive and not very or somewhat supportive), Authoritarian (strict and not very or somewhat supportive), Permissive (permissive and very support- ive), and Authoritative (strict and very supportive). Parenting style is that of the residential mother when she was present in the household (which was much more often the case) and such information was available; otherwise,

Adolescent Sexual Debut 339

the style of the residential father was used if he were present or if he and the residential mother were absent, that of the non-residential biological mother was used, or if no other parental figure was available, that of the non-residential biological father was used. Response items and categories for residential parents were (a) when you think about how s/he acts toward you, in general, would you say that s/he is very supportive, somewhat supportive, or not very supportive? and (b) in general, would you say that s/he is permissive or strict about making sure you did what you were supposed to do? Items for non-residential parents were phrased roughly comparable.

The parent-teen relationship scale was created from two separate eight- item Likert-scale measures that assessed the youths’ perception of their re- lationships with residential mothers and fathers respectively (e.g., thinking highly of, wanting to be like the parent). Individual items in each scale had values ranging from 0 to 4 (strongly disagree to strongly agree) and were summed and divided by eight. Both measures (alpha D .75 for youths’ perceptions of residential mothers and .82 for fathers) were used to create the parent-teen relationship scale (range D 0–4). If data were available for both residential parents (alpha D .61), scores were summed and divided by 2; otherwise, scores on residential parents for whom data were available, in most instances mothers, were used. Higher scores signified more positive teen perceptions of parent-teen relationships.

Family religion comprised measures of parental religiosity and religious affiliation. Parental religiosity was used as one of the familial system measures in light of findings suggesting that parental behavior and attitudes affect religious practices and other behavior in early adolescence (Potvin & Lee, 1982). Parental religiosity comprised a yes-no six-item scale (e.g., ‘‘I don’t need religion to have good values’’ (reverse coded), ‘‘I often ask God to help me make decisions’’) created by Child Trends (Moore et al., 1999; range D 0–6; alpha D .617). Higher scores signified a greater degree of parental religiosity. Following the National Campaign to Prevent Teen Pregnancy (2005), religious affiliation comprised the following categories: Baptist; Catholic; Conservative Protestant (United Church of Christ, Holi- ness, Pentecostal); Mainstream Protestant (Anglican/Episcopalian, Lutheran, Methodist, Presbyterian, Disciples of Christ, Reform, Nondenominational Christian, Other Protestant); Other Religions (Buddhist/Hindus, Eastern Or- thodox, Jews, Mormons, Muslims, Native American Tribal Religion, Other, Other Eastern, Unitarian, Wicca/Witchcraft/Magic/Pagan); and No Affiliation (including atheists and agnostics).

SELF SYSTEM

Self-system predictors of abstinence from and timing of first sexual inter- course included measures of delinquency and cognitive or intellectual ability.

340 R. K. Caputo

The Delinquency Index was created by Child Trends, Inc. in baseline year 1997 (Moore et al., 1999). Its score was obtained by summing responses from a list of delinquent/criminal acts (e.g., ever run away from, carried a hand gun, belonged to a gang) youths identified as having ever done for a possible total of 10. Three dummy measures were created from the Delinquency Index, signifying adolescents who reported no delinquency acts (46.5% of the study sample and the referent category), those who reported one delinquent act (21.5% of the study sample), and those who reported two or more such acts (32.0%). Each dummy measure was coded such that 1 D yes and 0 D no.

The Peabody Individual Achievement Test—Revised (PIAT-R) version’s percentile ranking on the math component was used as the measure of cognitive competency. It is one of the most widely used brief assessments of academic achievement of children ages five and older, with demonstrably high test-retest reliability and concurrent validity (Markwardt, 1998). The PIAT-R was administered only once to age-eligible respondents, and scores were obtained in each of the study survey years between 1997 and 2000. Scores signified the percent of youths who had lower scores on the same test; hence, higher scores indicated greater cognitive capacity compared to other, lower-scoring test-takers.

PROXIMATE EXTRA-FAMILIAL SYSTEM

There were two proximate extra-familial system measures: positive and neg- ative peer influences that were obtained from 10 items asked of adolescents about percentages of peers who engaged in a variety of activities. Adoles- cents’ perceptions of peers have been found to overestimate peer influence vis-à-vis that of parents on adolescent behavior owing in part to ignoring parental contributions to peer effects (Kandel, 1996). My study partially addressed this limitation by using hierarchical modeling, detailed further. Positive and negative peer influences were identified in light of Giordano (2003), who was critical of focal actors’ tendencies to characterize friends as monolithic (either good or bad influences) and of portended peer-family antitheses. A factor analysis was performed on the 10 items, and 2 clusters of items were identified, having eigenvalues above 1.0 and accounting for 51.6% of the cumulative variance. The first cluster signified percentage of peers involved in negative activities (e.g., % in gangs, % who use illegal substances; eigenvalue D 1.6; alpha D .55) and the second signified the percentage in positive activities (e.g., % who attend religious services regularly, % who do volunteer work; eigenvalue D 3.5; alpha D .83). These two measures were inversely but weak-to-moderately correlated (r D �.255; p < .01). Retaining both factors was deemed a relevant corrective in light of studies that accounted for only or primarily deviant (negative) peer influences (e.g., French & Dishion, 2003; Miller et al., 1997; Roche et al., 2005).

Adolescent Sexual Debut 341

DISTAL EXTRA-FAMILIAL SYSTEM

There were three distal extra-familial measures. Urbanity was contrasted with non-urban. Region of residence included Northeast, North Central, South, and West. This measure was included because teen pregnancy has been found to vary by regions, with many of the Southern states having the highest rates as compared to states in other parts of the country (National Campaign to Prevent Teen Pregnancy, 2006). It was deemed import to know whether younger adolescent sexual activity was more likely to occur in the South when controlling for other factors to assess whether that region of the country could benefit from targeted interventions. Unemployment rate referred to the youth’s area of residence. It was originally coded on a six-point scale such that 1 D less than 3.0% unemployment; 2 D 3% to 5.9%; 3 D 6% to 8.9%; 4 D 9% to 11.9%; 5 D 12.0% to 41.9%; and 6 D 15.0% plus.

SOCIODEMOGRAPHIC OR CONTROL MEASURES

Sociodemographic measures of age, class (SES), and gender were used as controls. Age was that reported at the time of interview in survey year 2000. Class was a function of household income, household size, and federal poverty thresholds. Staff at the Center for Human Resources Research (2003) created the income-to-poverty ratio (IPR) for each youth’s household in every survey year. All sources of annual household income were summed and the number of household members determined. Household incomes were then divided by federal poverty thresholds for respective household sizes in that survey year, thereby making it an appropriate measure of class. An IPR equal to 1 signified that a youth resided in a household whose annual income was at the appropriate poverty level for that survey year. For purposes of this study, youths residing in households with an IPR at or below 1 were classified as poor or lower class (22.1% of the entire study sample), above 1 through 4 as middle class (55.4%), and above 4 as upper class (22.5%). Gender was coded such that 1 D male, 0 D female. Ethnicity- race comprised three mutually exclusive groups of adolescents: White, Black, and Hispanic.

Procedures

Separate analyses for each ethnic-racial group incorporated two multivariate statistical procedures: multiple regression and multinomial logistic regres- sion. Because age at first intercourse was skewed slightly toward the left (Skewness D �1.64; standard error [SE] D .073) for the entire study sample, (Log10) age at time of first intercourse was used as the dependent measure in the multiple regression procedure for each racial-ethnic group. Correlates were entered hierarchically or sequentially from the following blocks: family system (measures of structure first, relations second, then religion); self

342 R. K. Caputo

system; proximate extra-family system; distal extra-family system; and finally sociodemographic measures. Tolerance and variance inflation factor colin- earity statistics were examined to ensure that relationships among the cor- relates were orthogonal (i.e., statistically independent). Standardized betas were used to show the relative predictive capacity of statistically significant correlates only. Multinomial logistic regression was used to determine the likelihood of having intercourse for the first time as a young adolescent (i.e., at age 14 or younger as reported in each survey year through 2000) vis-à- vis not having had sexual intercourse at all (i.e., being abstinent) through survey year 2000 or having intercourse as an older adolescent (15 years of age or older). The referent category for Sexual Behavioral Status in 2000 was Age at First Sex � 14 at Time of First Intercourse. Only measures found statistically significant in either the Virgin or the Age � 15 at Time of First Intercourse full distal extra-familial model were presented. In both regression procedures lower income or poor and the South were referent categories.

RESULTS

Descriptive Statistics

Of the 1,060 White youths in the study sample, 579 (44.6%) had lost their virginity by survey year 2000; of 443 Black youths, 335 (75.5%) had lost their virginity; and of 351 Hispanic youth, 218 (62.1%) had done so. Among White youths who had sexual intercourse, 174 (30.1%) had their first sexual inter- course by age 14; among Black youths, 183 (54.6%), and among Hispanic youth, 90 (41.3%). Analysis of variance statistics showed that among youths who had lost their virginity by age 14, the age of White youths at the time of first intercourse (13.1 years old) was older to a statistically significant degree than that of Hispanic youth (12.8), who in turn were older than Black youth (12.5) (F D 18.81; p < .001; post hoc p < .05; eta2 D .02). Likewise, among youths who had lost their virginity between the ages of 15 and 19, the age of White youths at the time of first intercourse (16.3 years old) was older to a statistically significant degree than that of Hispanic youth (16.2), who in turn were older than Black youth (16.0) (F D 16.84; p < .001; post hoc p < .05; eta2 D .01).

Black males (66.8%) and Hispanic males (66.9%) were twice as likely as their female counterparts (33.2% and 33.1%, respectively) to have lost their virginity at or before the age of 14, which was much greater than difference between White males (53.7%) and White females (46.3%; �2 D 30.02; p < .001). Black (43.6%) and Hispanic (45.8%) adolescents from low- income families more than twice as likely as their White (16.1%) counterparts to have lost their virginity at or before the age of 14 (43.6% and 45.8% vs.

Adolescent Sexual Debut 343

16.1%), whereas White adolescents from upper-income families were more than three times more likely than were Black and Hispanic youth (20.5% vs. 6.2% each; �2 D 147.70; p < .001).

Among White youth, 59.5% of adolescents who reported no delinquent acts were virgins in survey year 2000, compared to 40.7% of those reporting one delinquent act, and 27.9% reported two or more delinquent acts, whereas 06.7% of those who reported no delinquent acts had lost their virginity at or before the age of 14, compared to 11.9% of those who reported one delinquent act, and 33.9% of those who reported 2 or more delinquent acts (�2 D 146.10; p < .001). Among Black youth, 34.5% of adolescents who reported no delinquent acts were virgins in survey year 2000, compared to 28.0% of those reporting one delinquent act, and 09.2% reported two or more delinquent acts, whereas 31.0% of those who reported no delinquent acts had lost their virginity at or before the age of 14, compared to 36.6% of those who reported one delinquent act, and 57.5% of those who reported 2 or more delinquent acts (�2 D 38.77; p < .001). Among Hispanic youth, 52.3% of adolescents who reported no delinquent acts were virgins in survey year 2000, compared to 30.0% of those reporting one delinquent act, and 19.6% reported two or more delinquent acts, whereas 14.4% of those who reported no delinquent acts had lost their virginity at or before the age of 14, compared to 22.9% of those who reported one delinquent act, and 45.8% of those who reported 2 or more delinquent acts (�2 D 48.65; p < .001). No relationship was found between delinquency and SES by ethnicity/race.

Multiple Regression: Correlates of Age at Time of

First Intercourse

FAMILY SYSTEM MODELS

As can be seen in Tables 1 to 3, Models A to C, family-related factors accounted for only modest amounts of variation in (Log10) age at the time of first sexual intercourse by ethnicity/race, 1% or less each for set of structural, relationship, and religious factors, for a total of 5.2% when taken as a whole for White adolescents, 4.1% for Black adolescents, and 3.7% for Hispanic adolescents. The addition of family relations measures improved the overall explanatory power of family structural factors (Model A) only for White ado- lescents (see Table 1, Model B: F� D 3.82; p < .01); the addition of religious measures (Model C), however, failed to improve the overall explanatory power of Model B for any ethnic/racial group. For White adolescents, living with both parents was associated with delayed age of first intercourse (see Table 1, Model A: Beta D .099; p < .05), but this measure lost statistical significance when family relations measures were taken into account (see Table 1, Model B). No statistically significant family structure measures were found for Black or Hispanic youth.

T A

B L E

1 R e g re

ss io

n : C o rr

e la

te s

o f

(L o g 1 0 )

A g e

a t T im

e o f

F ir st

S e x u a l In

te rc

o u rs

e —

W h it e

S y st

e m

m o d e ls

F a m

il ia

l S e lf

P ro

x im

a te

e x tr

a -f a m

il ia

l D

is ta

l e x tr

a -f a m

il ia

l S o c io

d e m

o g ra

p h ic

S tr

u c tu

re M

o d e l A

R e la

ti o n s

M o d e l B

R e li g io

n M

o d e l C

M o d e l D

M o d e l E

M o d e l F

M o d e l G

S y st

e m

m e a su

re s

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

F a m

il ia

l st

ru c tu

re L iv

e d

w it h

b o th

p a re

n ts

.0 9 9

* .0

0 6

.0 7 8

n s

.0 0 6

.0 7 8

n s

.0 0 6

.0 5 6

n s

.0 0 6

.0 4 8

n s

.0 0 6

.0 5 5

n s

.0 0 6

.0 3 6

n s

.0 0 6

R e la

ti o n s

In d e x

o f

fa m

il y

ro u ti n e s

.0 9 9

* .0

0 1

.1 0 6

* .0

0 1

.0 6 9

n s

.0 0 1

.0 6 6

n s

.0 0 1

.0 6 1

n s

.0 0 1

.0 6 5

n s

.0 0 1

S e lf D

e li n q u e n c y

T w

o o r

m o re

� .2

2 9

** *

.0 0 6

� .2

0 2

** *

.0 0 6

� .2

0 9

** *

.0 0 6

� .1

8 0

** *

.0 0 6

P IA

T sc

o re

.1 8 4

** *

.0 0 0

.1 6 7

** *

.0 0 0

.1 7 1

** *

.0 0 0

.1 4 6

** *

.0 0 0

P ro

x im

a te

e x tr

a -f a m

il ia

l N

e g a ti v e

p e e r

� .1

0 2

* .0

0 1

� .1

1 2

* .0

0 1

� .1

6 3

** .0

0 1

D is

ta l e x tr

a -f a m

il ia

l U

n e m

p lo

y m

e n t ra

te �

.0 8 0

* .0

0 3

� .0

7 9

* .0

0 3

S o c io

d e m

o g ra

p h ic

A g e

in 2 0 0 0

.1 5 8

** *

.0 0 4

S E S U

p p e r

c la

ss .1

7 1

* .0

0 9

S e x

(1 D

m a le

) �

.0 8 9

* .0

0 5

R 2 /�

R 2

.0 1 0

.0 4 2 /.

0 3 2

.0 5 2 /.

0 1 0

.1 3 7 /.

0 8 5

.1 4 5 /.

0 0 9

.1 6 0 /.

0 1 5

.2 0 0 /.

0 4 0

F -C

h a n g e

3 .8

2 ,

p <

.0 1

1 .1

6 , p

> .0

5 1 8 .5

0 , p

< .0

0 1

5 .6

5 , p

< .0

5 1 .9

3 , p

> .0

5 6 .9

8 ,

p <

.0 0 1

N o te

. O

n ly

m e a su

re s

fo u n d

st a ti st

ic a ll y

si g n if ic

a n t

in a n y

o f

th e

sy st

e m

m o d e ls

a re

p re

se n te

d .

F o r

D e li n q u e n c y ,

N o

d e li n q u e n t

a c t

w a s

th e

re fe

re n t

c a te

g o ry

; fo

r S E S ,

lo w

e r

in c o m

e o r

p o o r

w a s

th e

re fe

re n t c a te

g o ry

. *p

< .0

5 , **

p <

.0 1 , **

*p <

.0 0 1 .

344

Adolescent Sexual Debut 345

Different family relations measures were found significant by ethnicity- race. For White adolescents, participating in family routines was positively related to delayed onset of sexual initiation (Beta D .099; p < .05; see Table 1, Model B). For Black adolescents, having more permissive vs. uninvolved parents was associated with younger age at sexual initiation (Beta D �.178; p < .05), whereas more positive parent-teen relations were associated with delayed sexual initiation (Beta D .155; p < .05; Table 2, Model B). Religious affiliation was not found significant in regard to age at first sexual intercourse for any ethnicity/race group.

SELF-SYSTEM MODEL

Self-system factors nearly tripled the explanatory power of the family system models for Whites, accounting for an increase of 8.5% to 13.7% of the variation in (Log10) age at the time of first sexual intercourse (see Table 1, Model D: F� D 18.50; p < .001). The increased explanatory power of self- system factors was more modest and statistically insignificant for Black and Hispanic adolescents, accounting for and additional increase of .7% to a total of 4.8% of the variation in (Log10) age at the time of first sexual intercourse for Black youth (see Table 2, Model D) and for an increase of 1.6% to 5.4% for Hispanic youth (see Table 3, Model D). Among White adolescents, those who reported two or more delinquent acts vs. none were younger at the time of first intercourse (Beta D �.229; p < .001), whereas greater levels of cognitive capacity were associated with later onset of sexual activity (Beta D .184; p < .001; see Table 1, Model D). Neither delinquency nor cognitive capacity was found to be associated with age of sexual intercourse for Black or Hispanic youth.

PROXIMATE EXTRA-FAMILIAL MODEL

The proximate extra-familial factor of negative peer activities increased the explanatory power of the family system models for Whites by .09% to 14.5% of the variation in (Log10) age at the time of first sexual intercourse (Beta D �.102; p < .05; see Table 1 Model E: F� D 5.65; p <. 05). Greater percentage of peers involved in socially unacceptable behaviors was associated with younger onset of sexual intercourse for White adolescents. No relationship between negative peer activities and (Log10) age at time of first intercourse was found for Black or Hispanic adolescents.

DISTAL EXTRA-FAMILIAL MODEL

The distal extra-familial factors accounted for statistically insignificant in- creases of 1.5% to 16.0% of the variation in (Log10) age at the time of first sexual intercourse for White youth (see Table 1, Model F) and of 2.6%

T A

B L E

2 R e g re

ss io

n : C o rr

e la

te s

o f

(L o g

1 0 )

A g e

a t T im

e o f F ir st

S e x u a l In

te rc

o u rs

e —

B la

c k

S y st

e m

m o d e ls

F a m

il ia

l S e lf

P ro

x im

a te

e x tr a -f a m

il ia

l D

is ta

l e x tr a -f a m

il ia

l S o c io

d e m

o g ra

p h ic

S tr u c tu

re M

o d e l A

R e la

ti o n s

M o d e l B

R e li g io

n M

o d e l C

M o d e l D

M o d e l E

M o d e l F

M o d e l G

S y st

e m

m e a su

re s

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

F a m

il ia

l R e la

ti o n s

P a re

n ti n g

st y le

P e rm

is si

v e

� .1

7 6

* .0

1 9

� .1

7 8

* .0

1 9

� .1

8 5

* .0

1 9

� .1

8 5

* .0

1 9

� .2

0 5

* .0

1 9

� .1

6 3

n s

.0 1 9

P a re

n t- T e e n

.1 5 5

* .0

1 0

.1 5 5

* .0

1 0

.1 4 4

* .0

1 0

.1 4 3

* .0

1 0

.1 3 1

* .0

1 0

.1 1 8

n s

.0 1 0

S o c io

d e m

o g ra

p h ic

S E S M

id d le

c la

ss .1

4 4

* .0

1 1

S e x

(1 D

M a le

) �

.2 6 9

** *

.0 1 0

R 2 /�

R 2

.0 0 7

.0 3 4 /.

0 2 7

.0 4 1 /.

0 0 7

.0 4 8 /.

0 0 7

.0 4 8 /.

0 0 0

.0 7 5 /.

0 2 6

.1 5 8 /.

0 8 3

F -C

h a n g e

1 .8

6 0 ,

p >

.0 5

0 .5

7 ,

p >

.0 5

0 .8

2 ,

p >

.0 5

0 .0

2 ,

p >

.0 5

1 .7

9 ,

p >

.0 5

7 .6

4 ,

p <

.0 0 1

N o te

. O

n ly

m e a su

re s

fo u n d

st a ti st

ic a ll y

si g n if ic

a n t in

a n y

o f

th e

sy st

e m

m o d e ls

a re

p re

se n te

d . F o r

P a re

n ti n g

S ty

le , U

n in

v o lv

e d

w a s

th e

re fe

re n t

c a te

g o ry

; fo

r S E S ,

lo w

e r in

c o m

e o r

p o o r

w a s

th e

re fe

re n t c a te

g o ry

.

*p <

.0 5 , **

*p <

.0 0 1 .

346

Adolescent Sexual Debut 347

to 7.5% for Black youth (see Table 2, Model F). For White adolescents, however, living in areas with higher unemployment was associated with younger age of first sexual intercourse (Beta D �.080; p < .05). Distal extra- familial factors accounted for a statistically significant increase of 8.1% to 13.8% of the variation in (Log10) age at the time of first sexual intercourse for Hispanic youth (see Table 3, Model F: F� D 3.71; p < .01). Hispanic youth residing in the North Central (Beta D .152; p < .05) and West (Beta D .226; p < .01) regions of the United States vs. those in the South were older at the time of first sexual intercourse, as were those residing in urban vs. non-urban areas (Beta D .193; p < .01).

SOCIODEMOGRAPHIC MODEL

Sociodemographic factors accounted for statistically significant increases of 4.0% to 20.0% of the variation in (Log10) age at the time of first sexual intercourse for White youth (see Table 1, Model G: F� D 6.98; p < .001), of 8.3% to 15.8% for Black youth (see Table 2, Model G: F� D 7.64; p < .001), and of 4.5% to 18.4% for Hispanic youth (see Table 3, Model G: F� D 2.67; p < .05). Age in survey year 2000 was positively related to age at time of first sexual intercourse for White adolescents (Beta D .158; p < .001), whereas upper-class White youth were older than were lower-class White youth (Beta D .171; p < .05), and White males were younger than White females (Beta D �.089; p < .05). Middle-class Black youth were older than lower-class Black youth at time of first intercourse (Beta D .144; p < .05), whereas Black males were younger than Black females (Beta D �.269; p < .001). Hispanic males were younger than Hispanic females at the time of first sexual intercourse (Beta D �.215; p < .01).

The sociodemographic model comprised all study measures and the robustness of several measures varied by ethnicity/race. For White adoles- cents (see Table 1, Model F), in addition to the previously reported so- ciodemographic measures of age and class, the self-system measures of delinquency (Beta D �.180; p < .001) and intellectual capacity (Beta D .146; p < .001) remained statistically significant, as were the proximate extra-familial measure of negative peer activities (Beta D �.163; p < .01) and the distal extra-familial measure of unemployment rate in the area of residence (Beta D �.079; p < .05). For Black adolescents, family relation measures of parenting style (Permissive vs. Uninvolved) and of parent-teen relations had remained robust through the addition of distal extra-familial measures (see Table 2, Models B–G), but lost statistical significance when sociodemographic factors were taken into account (Model F). Finally, for Hispanic adolescents (see Table 3, Model F), in addition to the previously reported sociodemographic measure of gender, distal extra-familial region measures of West versus South (Beta D .193; p < .05) and of urban versus non-urban residence (Beta D .183; p < .05) were found to be significant.

T A

B L E

3 R e g re

ss io

n : C o rr

e la

te s

o f

(L o g

1 0 )

A g e

a t T im

e o f F ir st

S e x u a l In

te rc

o u rs

e —

H is

p a n ic

S y st

e m

m o d e ls

F a m

il ia

l S e lf

P ro

x im

a te

e x tr a -f a m

il ia

l D

is ta

l e x tr a -f a m

il ia

l S o c io

d e m

o g ra

p h ic

S tr u c tu

re M

o d e l A

R e la

ti o n s

M o d e l B

R e li g io

n M

o d e l C

M o d e l D

M o d e l E

M o d e l F

M o d e l G

S y st

e m

m e a su

re s

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

B e ta

p S E

F a m

il ia

l D

is ta

l e x tr a -f a m

il ia

l R e g io

n o f

re si

d e n c e

N o rt h

C e n tr a l

.1 5 2

* .0

2 5

.1 0 8

n s

.0 2 6

W e st

.2 2 6

** .0

4 2

.1 9 3

* .0

1 5

U rb

a n

(1 D

y e s)

.1 9 3

** .0

2 1

.1 8 3

** .0

5 4

S o c io

d e m

o g ra

p h ic

S e x

(1 D

m a le

) �

.2 1 5

** �

.0 4 0

R 2 /�

R 2

.0 0 7

.0 3 1 /.

0 2 4

.0 3 7 /.

0 0 6

.0 5 4 /.

0 1 6

.0 5 7 /.

0 0 3

.1 3 8 /.

0 8 1

.1 8 4 /.

0 4 5

F -C

h a n g e

1 .0

6 , p

> .0

5 0 .3

3 ,

p >

.0 5

1 .1

8 ,

p >

.0 5

0 .7

4 ,

p >

.0 5

3 .7

1 , p

< .0

1 2 .6

7 ,

p <

.0 5

N o te

. O

n ly

m e a su

re s

fo u n d

st a ti st

ic a ll y

si g n if ic

a n t in

a n y

o f th

e sy

st e m

m o d e ls

a re

p re

se n te

d . F o r

R e g io

n o f R e si

d e n c e , S o u th

w a s

th e

re fe

re n t c a te

g o ry

.

*p <

.0 5 , **

p <

.0 1 .

348

Adolescent Sexual Debut 349

Multinomial Regression: Correlates of Virginity and

Older Vs. Younger Adolescent First Sex

WHITE ADOLESCENTS

As can be seen in Table 4, White youths who lived with both parents were 1.7 times as likely to be virgins than to have had sexual intercourse before the age of 15 (p < .05). Each additional family routine increased the odds of being a virgin by 10.4% (p < .001) and of having first sexual intercourse after age 14 by 6.0% (p < .01) vis-à-vis having first sexual intercourse before age 15. Delinquency decreased the odds of being a virgin versus having first intercourse before age 15 by 54% for White adolescents who reported one

TABLE 4 Multinomial Regression: Correlates of Sexual Behavioral Status in Survey Year 2000 (Full Model)—White

Sexual behavioral status in survey year 2000

Virgin Age at first sex � 15

System measures B p SE Exp(B) B p SE Exp(B)

Familial Structure

Same household .557 * .218 1.745 Relations

Index of family routines .099 *** .024 1.104 .058 ** .022 1.060 Self

Delinquency One delinquent act �.774 * .307 .461 Two or more �1.743 *** .274 .175 �1.186 *** .267 .305

PIAT score .016 *** .004 1.016 .011 * .004 1.011 Proximate

Negative peer �.181 *** .032 .835 �.110 *** .029 .896 Distal

Region of residence West 1.278 *** .351 3.590 .856 * .334 2.355

Sociodemograhpic Age in 2000 .654 *** .178 1.923 SES

Middle class .685 * .339 1.984 Upper class 1.061 ** .397 2.890 1.011 ** .369 2.747

�2 Log likelihood 1801.634 Chi-square 366.653, df D 50, p < .001 Nagelkerke R2 .336

Note. The referent category for Sexual Behavioral Status in 2000 is Age at First Sex � 14 at Time of First

Intercourse. Only measures found statistically significant in either the Virgin or the Age � 15 at Time of

First Intercourse full distal extra-familial model are presented. For Delinquency, No Delinquent Acts was

the referent category; for Region of Residence, South was the referent category; for SES, lower income

or poor was the referent category.

*p < .05, **p < .01, ***p < .001.

350 R. K. Caputo

delinquent act vis-à-vis those who reported none and by 83% for those who reported two or more delinquent acts. Delinquency also decreased the odds of having first sexual intercourse after age 14 versus having first intercourse before age 15 by 70%. Each additional increment of PIAT scores increased the odds of being a virgin by 1.6% and of having had first sexual intercourse after age 14 by 1.1% vis-à-vis having first intercourse before age 15. Each additional negative activity peers engaged in decreased the odds of being a virgin by 16.5% and of having had first sexual intercourse after age 14 by 10.4% vis-à-vis having first intercourse before age 15. Among distal extra- familial factors, those who resided in the West vis-à-vis those in the South were 3.5 times more likely to be virgins and 2.4 times more likely to have had their first sexual intercourse after age 15.

Among sociodemographic factors, in survey year 2000 each year older increased the odds of having had first sexual intercourse after age 14 by 92.3% (p < .001). Middle-class adolescents were nearly twice as likely as lower-income adolescents to be virgins versus having first sexual intercourse before age 15 (p < .05), whereas upper-class youth were nearly three times as likely (p < .05). Upper-class youth were also nearly three times as likely to have first intercourse after age 14 than before age 15 (p < .01).

BLACK ADOLESCENTS

As can be seen in Table 5 for Black adolescents, delinquency decreased the odds of being a virgin versus having first intercourse before age 15 by 79.7% for those who reported two or more delinquent acts. Each additional negative activity peers engaged in decreased the odds of being a virgin by 11.1%. Among distal extra-familial factors, residing in the Northeast vis-à- vis the South decreased the odds of first sexual intercourse after age 15 by 64.8%, whereas those residing in the West vis-à-vis those in the South were 3.1 times more likely have had their first sexual intercourse after age 15. No family structure measures were found to be associated with abstinence or older sexual activity for Black youth

Among sociodemographic factors, in survey year 2000 each year older decreased the odds of being a virgin versus having had first sexual inter- course before age 15 by 55.4% (p < .001), whereas being male decreased the odds by 73.4%. Being male increased the odds of first sexual intercourse after age 14 versus before age 15 by 72.1%.

HISPANIC ADOLESCENTS

As can be seen in Table 6 for Hispanic adolescents, each additional unit on the parent-teen relation scale increased the odds of being a virgin and of having had first sexual intercourse after age 14 vis-à-vis before age 15 by more than 150% (p < .05). Compared to non-religiously affiliated ado-

Adolescent Sexual Debut 351

TABLE 5 Multinomial Regression: Correlates of Sexual Behavioral Status in Survey Year 2000 (Full Model)—Black

Sexual behavioral status in survey year 2000

Virgin Age at first sex � 15

System measures B p SE Exp(B) B p SE Exp(B)

Self Delinquency

Two or more �1.597 *** .374 .203 PIAT score

Proximate Negative peer �.118 ** .038 .889

Distal Region of residence

Northeast �1.045 * .422 .352 West 1.122 * .546 3.071

Sociodemograhpic Age in 2000 �.605 *** .257 .546 Sex (1 D male) �1.122 *** .291 .326 �1.144 *** .251 .319

�2 Log likelihood 818.343 Chi-square 135.288, df D 46, p < .001 Nagelkerke R2 .298

Note. The referent category for Sexual Behavioral Status in 2000 is Age at First Sex � 14 at Time of First

Intercourse. Only measures found statistically significant in either the Virgin or the Age � 15 at Time of

First Intercourse full distal extra-familial model are presented. For Delinquency, No Delinquent Acts was

the referent category; for Region of Residence, South was the referent category.

*p < .05, **p < .01, ***p < .001.

lescents, Catholics were nearly six times as likely to be virgins (p < .05), whereas Mainstream Protestants and those with other religious affiliations were more than seven times as likely to be virgins (p < .05). Each additional negative activity peers engaged in decreased the odds of being a virgin by 14.4%. Delinquency decreased the odds of being a virgin versus having first intercourse before age 15 by 76.8% (p < .01). Among distal extra-familial factors, residing in the West vis-à-vis those in the South were 3.1 times more likely to have been virgins (p < .01) and 1.5 times as likely to have had their first sexual intercourse after age 15 (p < .05). Each unit increase in the unemployment rate was associated with a 45.6% increase in the odds of being a virgin versus having sexual intercourse before age 15. No family structure measures were found to be associated with abstinence or older sexual activity for Hispanic youth.

Among sociodemographic factors, in survey year 2000 upper-class youth were nearly seven times as likely to be a virgin versus having had first sexual intercourse before age 15 (p < .05). Upper-class youth were also nearly five times as likely to have first intercourse after age 14 than before age 15 (p < .05). Being male decreased the odds of being a virgin versus having first

352 R. K. Caputo

TABLE 6 Multinomial Regression: Correlates of Sexual Behavioral Status in Survey Year 2000 (Full Model)—Hispanic

Sexual behavioral status in survey year 2000

Virgin Age at first sex � 15

System measures B p SE Exp(B) B p SE Exp(B)

Familial Relations

Parent teen .940 * .384 2.561 .931 * .360 2.536 Religion

Affiliation Catholic 1.783 * .750 5.947 Mainstream Protestant 1.967 * .868 7.148 Other 2.045 * .982 7.731

Self Delinquency

Two or more delinquent acts

�1.462 ** .428 .232

Proximate Negative peer �.156 * .054 .856 �.106 * .049 .899

Distal Region of residence

West 1.120 ** .424 3.066 .911 * .396 2.486 Unemployment rate .376 ** .138 1.456

Sociodemograhpic SES

Upper class 1.925 * .743 6.857 1.573 * .711 4.823 Sex (1 D male) �1.830 *** .392 .160 �1.821 *** .370 .162

�2 Log likelihood 605.832 Chi-square 155.522, df D 48, p < .001 Nagelkerke R2 .404

Note. The referent category for Sexual Behavioral Status in 2000 is Age at First Sex � 14 at Time of First

Intercourse. Only measures found statistically significant in either the Virgin or the Age � 15 at Time of

First Intercourse full distal extra-familial model are presented. For Religion, the category of Other included

Baptists and Conservative Protestants whose numbers were too small to allow for sufficient separation

and None was the referent category; for Delinquency, No Delinquent Acts was the referent category; for

Region of Residence, South was the referent category; for SES, lower income or poor was the referent

category.

*p < .05, **p < .01, ***p < .001.

sexual intercourse before age 15 by 84.0%, whereas being male decreased the odds of first sexual intercourse after age 14 versus before age 15 by 83.8%.

DISCUSSION

The multi-system perspective used in this study shows how family, self, proximate family, and distal family systems correlates of timing of first sexual

Adolescent Sexual Debut 353

initiation and of abstinence among young adolescents differed among White, Black, and Hispanic youth before and after accounting for sociodemographic factors. When sociodemographic factors are ignored among White youth, the family system structural factor of living with both parents and the family system relational factor of index of family routines, the self-system factors of delinquency and intellectual capacity, the proximate family system factor of negative peer relations, and the distal family system factor of unemployment rate predict timing of sexual initiation. Among Black youth, the family system relational factors of parenting style and parent-teen relation predict timing of sexual initiation. Among Hispanic youth, the distal family system factors of region of residence and of urban residence predict sexual initiation. Findings also point to the robustness of sociodemographic factors of class and gender for each ethnic-racial group on timing of sexual initiation and of delinquency (self-system) and negative peer relations (proximate familial system) among Black and White youth and of religious affiliation (familial system) among Hispanic youth on abstinence. These findings suggest that professionals who work with adolescents and their families can implement different intervention strategies aimed at delaying onset of sexual intercourse or, conversely, of prolonging abstinence contingent on the ethnicity-race of their clients.

Findings suggest, for example, that measures directed at improving the involvement in family routines are likely to be more effective among White adolescents than among Black or Hispanic adolescents, other factors being equal. Measures directed at increasing permissive parenting styles for those adolescents whose parents seem uninvolved are likely to be more effective for Black adolescents rather than for White or Hispanic youth. Measures enhancing parent-teen relations from the perspective of adolescents are likely to be more effective for Black and Hispanic youth than for White adolescents.

Findings of the study also point to the robustness of delinquency as a correlate of age at first intercourse among White adolescents and of sexual initiation at an early age (i.e., of 14 years old or younger among White and Hispanic adolescents). White adolescents who engage in two or more delinquent acts are at much higher risk to have sex younger than those who engage in either one act of delinquency or none, regardless of family structure, processes, or religion (whether religiosity or affiliation), whereas White adolescents who engage in any number of delinquent acts and His- panic youth who engage in two or more versus none are more likely to have sexual intercourse at an early age of 14 or younger. Delinquency is a robust correlate or risk-factor of early adolescent sexuality, independently not only of such family characteristics but of the sociodemographic characteristics gender and socioeconomic status among White and Hispanic youth but not among Black adolescents. Findings suggest that delinquency can serve as a useful flag to practitioners in regard to at-risk sexual behavior among White and Hispanic adolescents. In addition, policy makers who want to design programs that target delinquent youth can benefit from knowing that

354 R. K. Caputo

sexual activity at early ages is a robust component of White and Hispanic delinquents’ lives that warrants attention.

Findings also highlight ethnic-racial variation in the robustness of the proximate extra-familial system measure of negative peer influences on tim- ing of and age at first intercourse. Regardless of other sociodemographic factors, White adolescents whose peer groups engage in a range of socially objectionable activities have sexual intercourse at younger ages, and they are less likely to remain virgins by age 16 or to have first sex at age 15 or later than they are to have sex at age 14 or younger. Negative peer relations have no association with timing of first sexual intercourse among Black or Hispanic youth, but to the extent their respective peer groups engage in a range of socially objectionable activities, Black and Hispanic youth, like White youth, are less likely to remain virgins by age 16 or to have first sex at age 15 or later than they are to have sex at age 14 or younger. Findings suggest that interventions aimed at groups of youth known to engage in a variety of such behaviors might effectively increase the likelihood and timing of first intercourse among White adolescents and the likelihood of first intercourse for all adolescents.

Findings also highlight the importance of the distal extra-familial sys- tem measure region of residence as a robust correlate on timing of first intercourse among Hispanic adolescents and on the likelihood of having had sexual intercourse at an early age among White, Black, and Hispanic youth. In particular, when compared to adolescents who had their first sexual experience at 14 years of age or younger, those residing in the West are more likely to be virgins and to have their first sexual intercourse at age 15 or older than youth residing in the South among White, Black, and Hispanic youth. These findings may in part account for the leadership role of Southern Baptists in the virginity pledge movement and million-man-marches. Both of these activities are aimed at reducing teen pregnancy, whose rates are higher in the South (National Campaign to Prevent Teen Pregnancy (2006), and at promoting more socially responsible sexual behavior (Bearman & Bruckner, 2001).

Finally, sociodemographic findings suggest that class, gender, and to a lesser extent ethnicity-race can serve as useful identifiers of adolescents at risk for first sexual intercourse at early ages and for whom appropriate public policies might best be targeted. Findings corroborate many previous studies showing that adolescents from low-income families are at greater risk for early sexual initiation, although there is variation by ethnicity/race (Kirby et al., 2005). Among sexually experienced White adolescents, only upper- income vis-à-vis low-income adolescents are likely to be older at time of first sexual intercourse, whereas among sexually experienced Black youth, only middle-class youth are. Middle- and upper-class White youth and upper- class Hispanic youth are nonetheless more likely to be virgins and to have sex after age 15 than they are to have sex at an early age, at age 14 or

Adolescent Sexual Debut 355

below. These findings suggest that income-related, means-tested policies such as the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 that target adolescent sexuality in general and teen pregnancy in particular invariably bypass sexually active middle-income White and Hispanic adolescents and upper-income Black and Hispanic adolescents. Further, sociodemographic findings suggest that the focus on teen pregnancy in such legislation places too much emphasis on adolescent women as among sexually adolescents, males are younger than females at the time of first intercourse across ethnic-racial groups and, among Black and Hispanic youth, they are less likely to be virgins and to have sex after age 15 than they are to have sex at an early age, at age 14 or below. Findings of this study suggest it a key strategy for reducing teen pregnancy would focus on the sexual behavior of males regardless of ethnicity-race.

In conclusion, findings of this study suggest that professionals and policy makers interested in adolescent sexuality can benefit from taking a multi- system perspective when assessing the merits of intervention strategies aimed at delaying first sexual intercourse. Attributes of family, self, proximate extra- familial, and distal extra-familial systems are associated with the likelihood and timing of first intercourse in nuanced ways, and appropriate interven- tions can be targeted accordingly. As with any study based on secondary data analysis, this one had several limitations that should be kept in mind regarding implications for practice and policy. First, the explanatory power of the models is modest at best, suggesting that other significant measures need to be identified and subject to future research. Second, the study design and reliance on survey data preclude analysis of causal factors and instead permit identification of correlations and associations of factors thought to influence adolescent sexual behavior. Statistical controls used in survey research can only approximate experimental designs and even though temporal ordering can be established on some measures, such designs lack the random as- signment in experimental and control groups characteristic of true or classic experimental designs. Third, the study is limited to one cohort of adolescents and hence is not representative of all adolescents in the United States over the study period. Despite these limitations, however, findings highlight how and in what circumstances fairly specific measures associated with the four theoretically relevant aspect of the social environment, namely family, self, proximate family, and distal family systems, are associated with sexual debut among young adolescents.

REFERENCES

Albert, B., Brown, S., & Flanigan, C. (Eds.). (2003). 14 and younger: The sexual behavior of young adolescents (summary). Washington, DC: National Campaign to Prevent Teen Pregnancy.

356 R. K. Caputo

Averett, S. L., Rees, D., & Argys, L. M. (2002). The impact of government policies and neighborhood characteristics on teenage sexual activity and contraceptive use. American Journal of Public Health, 92, 1773–1778.

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

Benson, P. L., & King, P. E. (2005). Adolescence. In H. R. Ebaugh (Ed.), Handbook of religion and social institutions (pp. 121–138). New York: Springer.

Blum, R. W. (2002). Mothers’ influence on teen sex: Connections that promote postponing sexual intercourse. Minneapolis: Center for Adolescent Health and Development, University of Minnesota.

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.

Bronfrenbrenner, U. (1986). Ecology of the family as a context for human develop- ment: Research perspectives. Developmental Psychology, 22, 723–742.

Browning, C. R., Leventhal, T., & Brooks-Gunn, J. (2004). Neighborhood context and racial differences in early adolescent sexual activity. Demography, 41, 697– 720.

Caputo, R. K. (2007a). Sex at an early age: A multi-system perspective. Race, Gender, & Class, 17(3–4), 206–227.

Caputo, R. K. (2007b). Working and poor: A panel study of maturing adults in the U.S. Families in Society, 88, 351–359.

Center for Human Resource Research (2003). NLSY97 user’s guide: A guide to the rounds 1-5 data national longitudinal survey of youth 1997. Columbus: Ohio State University.

Davis, E. C., & Friel, L. V. (2001). Adolescent sexuality: Disentangling the effects of family structure and family context. Journal of Marriage and Family, 63, 669– 681.

Donenberg, G. R., Wilson, W. H., Emerson, E., & Bryant, F. B. (2002). Holding the line with a watchful eye: The impact of perceived parental permissiveness and parental monitoring on risky sexual behavior among adolescents in psychiatric care. AIDS Education and Prevention, 14, 138–157.

French, D. C., & Dishion, T. J. (2003). Predictors of early initiation of sexual in- tercourse among high-risk adolescents. Journal of Early Adolescence, 23, 295– 315.

Furstenberg, F. F., Morgan, P., Moore, K. A., & Peterson, J. L. (1987). Race differences in the timing of adolescent intercourse. American Sociological Review, 52, 511– 518.

Giordano, P. C. (2003). Relationships in adolescence. Annual Review of Sociology, 29, 257–281.

Harris, K. M., Duncan, G. J., & Boisjoly, J. (2002). Evaluating the role of ‘‘nothing to lose’’ attitudes on risky behavior in adolescence. Social Forces, 80, 1005– 1039.

Hofferth, S. L. (1987). Factors affecting initiation of sexual intercourse. In S. L. Hof- ferth & C. D. Hayes (Eds.), Risking the future: Adolescent sexuality, pregnancy, and childbearing. Vol. 2 (pp. 7–35). Washington, DC: National Academy Press.

Adolescent Sexual Debut 357

Kahn, J. R., Kalsbeek, W. D., & Hofferth, S. L. (1988). National estimates of teenage sexual activity: Evaluating the comparability of three national surveys. Demog- raphy, 25, 189–204.

Kandel, D. B. (1996). The parental and peer contexts of adolescent deviance: An algebra of interpersonal influences. Journal of Drug Issues, 26, 289–315.

Kirby, D. (2001). Understanding what works and what doesn’t in reducing adolescent sexual risk-taking. Family Planning Perspectives, 33, 276–281.

Kirby, D., Lapore, G., & Ryan, J. (2005). Sexual risk and protective factors: Factors af- fecting teen sexual behavior, pregnancy, childbearing and sexually transmitted disease: Which are important? Which can you change? Retrieved September 3, 2007, from http://www.teenpregnancy.org/works/risk_protective_kirby/kirby_ riskandprotectivefactor_paper.pdf

Longmore, M. A., Manning, W. D., & Giordano, P. C. (2001). Preadolescent parenting strategies and teens’ dating and sexual initiation: A longitudinal analysis. Journal of Marriage and Family, 63, 322–335.

Markwardt, F. C. (1998). Peabody Individual Achievement Test—Revised. Circle Pines, MN: American Guidance Service.

Miller, B. C. (2002). Family influences on adolescent sexual and contraceptive be- havior. The Journal of Sex Research, 39, 22–26.

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 & Society, 29, 54–83.

Miller, K. S., Forehand, R., & Kotchick, B. A. (1999). Adolescent sexual behavior in two ethnic minority groups: The role of family variables. Journal of Marriage and the Family, 61, 85–98.

Miller, K. S., Forehand, R., & Kotchick, B. A. (2000). Adolescent sexual behavior in two ethnic minority groups: A multisystem perspective. Adolescence, 35, 313– 333.

Moore, K. A., McGroder, S. M., Hair, E. C., & Gunnoe, M. L. (1999). NLSY97 codebook supplement main file round 1. Appendix 9: Family processes and adolescent

outcomes measures. Washington, DC: Bureau of Labor Statistics, Department of Labor.

National Campaign to Prevent Teen Pregnancy (2006, October). Teen pregnancy and birth rates in the United States. Washington, DC: Author.

National Campaign to Prevent Teen Pregnancy (2005). The association between parent, family, and peer religiosity and teenagers’ sexual experience and contra- ceptive use. Science Says, 20. Washington, DC: Author. Retrieved April 23, 2009, from http://www.thenationalcampaign.org/resources/pdf/SS/SS20_Religionand Experience.pdf

Pedersen, W., Samuelsen, S. O., & Wichstrom. L. (2003). Intercourse debut age: Poor resources, problem behavior, or romantic appeal? A population-based longitudinal study. The Journal of Sex Research, 40, 333–345.

Perkins, D. F., Luster, T., Villarruel, F. A., & Small, S. (1998). An ecological, risk-factor examination of adolescents’ sexual activity in three ethnic groups. Journal of Marriage and the Family, 60, 660–673.

Potvin, R. H., & Lee, C. F. (1982). Adolescent religion: A developmental approach. Sociological Analysis, 43, 131–144.

358 R. K. Caputo

Ramirez-Valles, J., Zimmerman, M. A., & Juarez, L. (2002). Gender differences of neighborhood and social control processes: A study of the timing of first inter- course among low-achieving, urban, African American youth. Youth & Society, 33, 418–441.

Rashad, I., & Kaestner, R. (2004). Teenage sex, drugs, and alcohol use: Problems identifying the cause of risky behavior. Journal of Health Economics, 23, 493– 503.

Regan, P. C., Durvasula, R., Howell, L., Ureno, O., & Rea, M. (2004). Gender , eth- nicity, and the developmental timing of first sexual and romantic experiences. Social Behavior and Personality, 32, 667–676.

Roche, K. M., Mekos, D., Alexander, C. S., Astone, N. M., Bandeen-Roche, K., & Ensminger, M. E. (2005). Parenting influences on early sex initiation among adolescents: How neighborhoods matter. Journal of Family Issues, 26, 32–54.

Rostosky, S. S., Regnerus, M. D., & Wright, M. L. C. (2003). Coital debut: The role of religiosity and sex attitudes in the Add Health Survey. The Journal of Sex Research, 40, 358–367.

Rostosky, S. S., Wilcox, B. L., Wright, M. L. C., & Randall, B. A. (2004). The impact of religiosity on adolescent behavior: A review of the evidence. Journal of Adolescent Research, 19, 677–697.

Santelli, J. S., Lowry, R., Brener, N. D., & Robin, L. (2000). The association of sexual behaviors with socioeconomic status, family structure, and ethnicity/race among US adolescents. American Journal of Public Health, 90, 1582–1588.

Smith, C. A. (1997). Factors associated with early sexual activity among urban ado- lescents. Social Work, 42, 334–346.

Steinberg, L. D. (2004). Adolescence (7th ed.). New York: McGraw-Hill. Thornton, A., & Camburn, D. (1989). Religious participation and adolescent sexual

behavior and attitudes. Journal of Marriage and the Family, 51, 641–653. Trenholm, C., Devaney, B., Fortson, K., Quay, L., Wheeler, J., & Clark, M. (2007).

Impacts of title v, section 510 abstinence education programs. Final report. Princeton, NJ: Mathematica Policy Research and Washington, DC: U.S. De- partment of Health and Human Services. Retrieved August 27, 2007, from http://mathematica-mpr.com/publications/pdfs/impactabstinence.pdf

Whitbeck, L. B., Yoder, K. A., Hoyt, D. R., & Conger, R. D. (1999). Early adolescent sexuality: A developmental study. Journal of Marriage and the Family, 61, 934– 946.

Wu, L. L., Martin, S. P., & Long, D. A. (2001). Comparing data quality of fertility and first sexual intercourse histories. The Journal of Human Resources, 36, 520–555.

Copyright of Journal of Human Behavior in the Social Environment is the property of Taylor & Francis Ltd and

its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.