Although most adolescents have access to several networks, including the family, the school, and the neighborhood peer group, they vary in the degree to which they are embedded in each of these interaction settings (Farrell, Bissel, Pan, & Barnes, 2001). For example, some adolescents go through the rituals of school; they go to classes and do the required academic work, but they are indifferent to the teachers and most of their fellow students. Still others may be alienated from their families and peers in the neighborhood, but they find support, stimulation, and social models at school. In this paper we examine how embedded adolescents are in family, school and neighborhood networks, and whether the degree of embeddedness in each type of setting influences their sexual behavior. Specifically, in a six year longitudinal study of adolescents, we examine whether the context in which they interact with their support networks - family context, neighborhood context, or school context - affects their rate of sexual activity, the number of sexual partners they have, and the likelihood of a pregnancy occurring. By support network we mean the people that an adolescent turns to for advice, emotional support, instrumental help, and companionship. By degree of embeddedness in a context we mean the proportion of their support network seen in that context. Thus, an adolescent who reports that sixty percent of her support network are seen in a family context is more embedded in her family than one who reports that only fifteen percent of her network are seen in a family context.

Drawing on control (e.g., Hirschi, 1969), social capital (e.g., Lin, 2001), and differential association ( e.g., Matza, 1964) theories, we hypothesize:

H1: The more embedded the adolescent is in a family or school context, the less the likelihood of sexual activity in early adolescence, the less the number of sex partners each year, and the less the likelihood of pregnancy over the course of adolescence.

H2: The more embedded the adolescent is in a neighborhood context, the greater the likelihood of sexual activity in early adolescence, the greater the number of sex partners each year, and the greater the likelihood of pregnancy over the course of adolescence.

Methods

A representative sample of 699 families in metropolitan Buffalo, NY, was obtained using random digit dialing telephone procedures to locate families. To be included, a family had to have at least one parent and one adolescent, age 13 to 16. Families were followed for six successive years, (1989-1996). In year one, families were offered $50.00 to participate, and in years two and three they were paid $75.00. In years four through six, when many adolescents had moved outside of their homes, we paid parents and adolescents separately $25.00 to come to our offices to be interviewed. To allow for comparisons, black families were over-sampled, resulting in 210 (30%) black, 472 (67%) white, and 16 (3%) other families. In year four, when network data were gathered, sample attrition resulted in an overall sample size of 485.

Sexual activity was measured with the following questions answered in a self-report section of the interview:

1. About how many times have you had sexual intercourse in the last 12 months?

2. How many different people have you had sexual intercourse with in the last 12 months?

3. If you’re a female, have you gotten pregnant in the last 12 months?

4. If you’re male, have you gotten a girl pregnant in the last 12 months?

Question one was asked all six years. Question two was asked in years 3 through 6. Questions three and four were asked in years 5 and 6. We did not ask all the questions in all years because questions two through four did not seem appropriate to ask of the younger adolescents. As an indicator of sexual activity, we created a dummy variable: 0 for no activity and 1 for any activity. As an indicator of multiple partners, we created a dummy variable: 0 for zero or one partner, and 1 for two or more partners. The indicator of pregnancy was set at 0 for "no" and 1 for "yes."

In year four, following Fischer’s (1982) method, we asked the adolescents the following questions to elicit the members of their support network: Who would you turn to 1) when you are "emotionally down" or "need to make a difficult decision;" 2) when you need to "borrow money" or "get a ride somewhere;" and 3) when you want to "go to the movies or do something for the fun of it." After obtaining a list of initials in response to each question, we then asked a series of questions about each person named, including – "In what context do you usually see this person (family, neighborhood, church, school, work, or other)." Network size is the total number of people listed at least once in response to all the questions. As an index of the degree to which the adolescent is embedded in a family-centered network, we computed the percentage of the total network seen in a family context. The same procedure was used to compute the degree of embeddedness in school or neighborhood networks. A respondent could indicate they saw a person in more than one context. In this case, the person named is counted as a member of each network indicated. Thus, a cousin who is seen in a family context, in the neighborhood, and in school would be included in each of the lists, and he or she would be included in the calculation of all three indices of embeddedness. Family income was either mother’s income or the combined income of mother and father or stepfather, whichever was applicable.

We first use logistic regression to examine the effects of the three types of network embeddedness on sexual behavior, controlling for family income, race, gender, and adolescent age. In the logistic regressions presented here, the indices of embeddedness were split at the median to aid in interpreting the odds ratios. A value above the median was designated with ‘1’; a value below the mean was designated with ‘0’.

The means and standard deviations for the network properties of black and white, male and female adolescents are displayed in table 1. Means are adjusted for the effects of the control variables (income and age). Note that black adolescents interact with a larger portion of their network members in a family context. White adolescents report a greater percentage of their networks are seen in a school setting. Looking at the gender differences, we find that males report a larger portion of their networks seen in the neighborhood. In figure 1 we present the proportion of black and white, male and female adolescents who are sexually active each year. Note that black males are most likely to be active in year one, and white females are least likely to be active. Because network properties and sexual activity are correlated with race and gender as well as income and age, in our analyses we control for these factors.

(Tables 2 and 3 about here)

In table 2 we present logistic regression coefficients showing the effects of the embeddedness indicators and controls on sexual activity for each of the six years observed, controlling for income, age, race, and gender. In table 3 we present the logits (odds ratios) based on this regression. Reflecting the same findings displayed in figure 1, we see that in years one and two blacks are more likely to be sexually active than whites; and in years one through three males are more likely to be sexually active than females. Looking at the effects of the context indicators, we see that embeddedness in the family reduces the odds of sexual activity in years one and two. Likewise, embeddedness in school networks reduces the odds of sexual activity during the high school years (years 1 through 4). Finally, embeddedness in the neighborhood networks increases the odds of sexual activity in years 4 through 6.

(Tables 4 and 5 about here)

Looking at the multiple partners data gathered in years 3 through 6 (Tables 4 and 5), we see that blacks are more likely to have multiple partners than whites over the past 12 months for all four years, and males are more likely to have multiple partners than females. Looking at the effects of the embeddedness indices, we see that the odds of multiple partners are less for those highly embedded in a family context across all four years observed. Those embedded in a neighborhood context have greater odds of having multiple partners in the later years of adolescence. Those embedded in school-based networks are less likely to have multiple partners while in school (i.e., year 3, when the ages ranged from 16 to 18).

(Tables 6 and 7 about here)

Finally, looking at the pregnancy data (Tables 6 and 7), we see that in years 5 and 6, the odds of black adolescents becoming pregnant are 2 to 3 times greater than the odds for whites. Looking at the effects of the context indices, we see that embeddedness in the neighborhood context increases the likelihood of pregnancy, but the other network context indices have no effects.

Discussion

In Larson and Richard’s (1994) imaginative studies of the daily and weekly cycles of emotions of adolescents, they used pagers to beep respondents at random points during the course of each day throughout a typical week in their lives. When the adolescents received a beep, they were prompted to note in diaries where they were, whom they were with, and what they were feeling. Not surprisingly, they found that over the course of the week adolescents cycled in and out of a set of networks: family, school, extracurricular activities, neighborhood groups, work, and so on.

Our findings suggest that adolescents vary in the degree to which they become attached to people or "embedded" in each of these networks, and this has consequences in their lives. The more embedded they are in networks seen in a family context, the less likely they are to become sexually active at an early age and the less likely they are to have multiple partners over the course of a year. Likewise, the more embedded they are in networks based at school, the less likely they are to be sexually active while in high school, and the less likely they are to have multiple partners. However, the more embedded they are in a peer network in the neighborhood, the more likely they are to have multiple partners, and, if female, the more likely they are to become pregnant, or, if male, get some one pregnant. These findings have implications for interventions and for future research. Strengthening attachments to family-based and school based networks, and increasing interaction in those settings can contribute to less risky sexual behavior. Reducing the embeddedness in neighborhood peer networks can also contribute to less risky behavior. However, to clarify what it is that has the positive or negative effects, it is important to examine the intervening processes that occur in each of these network contexts that contribute to the effects.

Six Years of Observation.

Table 1. Means and Standard Deviations for Network Properties of Black and White, Male and Female Adolescents.

Note: respondents can indicate more than one location for each network member.

Sex, Age, Family, Neighborhood, and School.