[Skip to Navigation]
Sign In
January 2005

Effects of Out-of-School Care on Sex Initiation Among Young Adolescents in Low-Income Central City Neighborhoods

Author Affiliations

Author Affiliations: Department of Population and Family Health Sciences, The Johns Hopkins Bloomberg School of Public Health (Drs Roche and Astone), and Department of Pediatrics, The Johns Hopkins School of Medicine (Dr Ellen), Baltimore, Md.

Arch Pediatr Adolesc Med. 2005;159(1):68-73. doi:10.1001/archpedi.159.1.68

Objective  To examine the association between patterns of out-of-school care over time and the initiation of sex among young adolescents living in low-income urban families.

Design  A prospective cohort study (using a 16-month follow-up) examining the predictive value of changes in out-of-school-care arrangements on early sex initiation.

Setting and Participants  By using a multistage stratified area probability sampling strategy, we selected 494 subjects aged 11 to 13 years living in low-income central city neighborhoods who did not report having had sexual intercourse by time 1.

Main Outcome Measure  Adolescent report of having had sexual intercourse by time 2.

Results  Bivariate results suggested that being at home with an adult during out-of-school hours was related to less sex initiation than self-care, care at another person’s home, and attendance at an organized/supervised activity. Adjusting for demographic attributes, parental monitoring, parental curfews, and family routines, remaining in out-of-home care or self-care was associated with a 2.5 times (95% confidence interval, 1.3-5.1 times) increase in the likelihood of initiating sexual intercourse when compared with being at home with an adult at both assessment times.

Conclusion  Policies and programs that enable young adolescents to spend their out-of-school hours at home with an adult may help reduce the risk of early sex initiation among youth in low-income urban areas.

Although the proportion of adolescents initiating sex before they are in ninth grade declined from 40% in 1991 to 34% in 2001,1 there are compelling public health reasons to continue to focus on early sex initiation among adolescents. Adolescents who engage in sexual intercourse at an early age experience increased risks of having multiple sexual partners2,3 and using contraception irregularly4 and, in turn, face heightened risks of experiencing unplanned pregnancies and acquiring sexually transmitted infections.5

Much of the research exploring social contextual influences on early sex among adolescents has focused on risks related to parenting practices and neighborhood conditions.6,7 Virtually none of this contextual research has considered how care arrangements during out-of-school hours (ie, after school and during the summer) affect the probability that a young person has sex at an early age. This is despite the fact that many adolescents spend after-school and summer hours away from parents.8 Where and with whom adolescents spend their out-of-school hours provide important information about the nature of youths’ out-of-school social contexts. Out-of-school care is conceptually distinct from parenting practices such as monitoring and involvement, which represent explicit parental efforts aimed at influencing adolescent behavior.

Knowledge about how out-of-school care affects adolescent problem behaviors such as substance use, delinquency, and deviant peer involvement may inform our understanding of out-of-school care’s impact on early sex initiation because early sex often co-occurs with these other adolescent health-compromising behaviors.9 In this regard, some research has shown that substance use is significantly higher among students in self-care (ie, “latchkey children”) than among youth not in self-care, and that the extent of problem behaviors increases with the number of hours spent in self-care.10 Other studies, however, have shown that adolescents in self-care do not differ in terms of behavior problems, self-esteem, social adjustment, and susceptibility to peer pressure when compared with adolescents being cared for by an adult. Rather, these studies have shown that negative health outcomes increase as out-of-school-care arrangements become further removed from adult supervision, such as moving from being at a neighbor’s house, to being at school, to being at a job, and, especially, to “hanging out” with friends.11-13 There is some evidence that high levels of parental supervision and involvement mitigate risks associated with self-care.11,14 These studies did not focus on low-income families and did not consider the extent to which parenting practices mediated associations between out-of-school-care arrangements and adolescent behavior.

In the present study, we examine the relationship between patterns of out-of-school care over time and the early initiation of sex among low-income, primarily racial or ethnic minority, adolescents living in poor, urban neighborhoods. There are several reasons to focus on this population. First, when compared with their socioeconomically advantaged counterparts, these youth are more likely to initiate sex at an early age.15-17 Second, because work mandates included in the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 have resulted in more poor mothers moving into the workplace,18 issues of out-of-school supervision and care have becoming increasingly relevant to poor families.19 Third, families in low-income urban areas face unique challenges in caring for their adolescent youth during out-of-school hours. For example, neighborhood violence and other threats to safety have a heightened salience to families in low-income, urban areas such that self-care may be a less desirable arrangement for these families. The fact that self-care is less prevalent among families in central city areas and among African American compared with white youth20 supports this possibility. Given that families with limited financial resources face constraints in accessing high-quality after-school and summer programs, they face limited options regarding after-school and summer care for their youth. Possibly because of racial or ethnic minority families’ greater economic needs and lack of formal labor market opportunities, African American and Latin American families are more likely than white families to rely on extended kin to care for their adolescents.21

The research question addressed in this study is: “What is the association between patterns of out-of-school care over time and sex initiation among young adolescents living in low-income, urban neighborhoods?” When compared with youth who either remain in self-care over time or move into self-care at time 2, we expect that adolescents in adult-supervised forms of care will have a lower odds of initiating sex. Furthermore, it is expected that positive parenting practices will mediate some of the association between adult-supervised care and reduced sex initiation.



The data for this study derive from a household-based, stratified-random, longitudinal survey22 of more than 2400 low-income families living in low-income neighborhoods in 3 locations (Boston, Mass; Chicago, Ill; and San Antonio, Tex). Participants were interviewed in 1999 and again an average of 16 months later. Data were collected on demographics, family health and well-being, and parenting. A detailed description of the study’s design and methods is available elsewhere.23 Informed consent was obtained from caregivers and youth before each interview. The institutional review boards of The Johns Hopkins University, Baltimore; Northwestern University, Chicago; Harvard University, Boston; The University of Texas at San Antonio; Penn State University, University Park, Pa; Boston College, Boston; and Research Triangle Institute, Research Triangle Park, NC, approved all procedures.

Because of our interest in sex initiation among young adolescents (for whom the health risks associated with sexual intercourse are greatest4-6), we selected a sample of female caregivers and their adolescents who were aged 11 to 13 years at time 1. We refer to female caregivers as mothers in this study because 90% of them were the biological or adoptive mother of the adolescent. Of these 694 youth, we excluded youth who did not participate at time 2 (74 [10.7%]), had already initiated sex by the first time assessment (46 [6.6%]), and had missing or invalid data for study variables (96 [13.8%] of those who had participated in both waves). This resulted in a sample of 494 subjects aged 11 to 13 years (71.2% of the original sample of adolescents in this age group).

As shown in Table 1, most adolescents were African American or Latino. Almost two thirds of the youth lived with a mother only, and in more than a third of families, neither the parent nor spouse had finished high school or received a general equivalency diploma. Just over half of mothers in the sample did not have a paying job; 25.7% worked full time.

Table 1. 
Description of the 494 Study Participants at Time 1*
Description of the 494 Study Participants at Time 1*

Adolescents excluded from the sample because they already had had sex at time 1 were more likely to be male (P<.05) and reported lower levels of parental monitoring (P<.001) and parental curfews (< .05) than did youth who had not had sex by time 1.


Dependent Variable

Adolescent sexual intercourse was measured by the question: “Have you ever had sexual intercourse?” (Response categories were as follows: 0, no; and 1, yes.) The question was preceded by a statement that sexual intercourse is sometimes called “having sex,” “making love,” or “going all the way.” Because adolescents were aged 11 to 13 years at the time of enrollment into the study, we regard the initiation of sexual intercourse by the time 2 interview (an average of 16 months after the time 1 interview) as early sex initiation.

Independent Variables

Out-of-school care was measured using the mother’s time 1 and time 2 answers to the question: “Aside from school, where would you say [child] spent most of [his/her] time when [he/she] was not with you last week?” Original response categories included the following: 1, “child always with respondent”; 2, “in respondent’s home with father or respondent’s partner”; 3, “in respondent’s home with sibling(s)”; 4, “in respondent’s home with a relative”; 5, “in respondent’s home with a nonrelative”; 6, “formal day care center”; 7, “organized/supervised activities (eg, sports or an after-school program)”; 8, “outside of home with a relative”; 9, “outside the home with nonrelative”; 10, “home, playing outside”; 11, “home alone”; and 12, “other (specify).” Because there were few youth in some of these care arrangements, we categorized responses into the following groups at times 1 and 2: 0, “at home in maternal care”; 1, “at home with father or mother’s male partner or another adult”; 2, “in an organized/supervised activity”; 3, “outside the home in the care of a relative or nonrelative”; and 4, “self-care” (includes home without an adult and playing outside). Adolescents in the other category (n = 7) were not included in the sample. (At time 1, we were only able to verify if the sibling at home with the youth was an adult because a subsequent survey item asked about the age of the sibling but not of the oldest person with the child at home. Our categorization of responses at time 1 may, therefore, have some measurement error because we assume that the nonsibling or nonrelative at home is an adult.)

Measures of parenting included parental monitoring, parental curfews, and family routines. Parental monitoring was a 5-item measure indicating adolescent reports of the mother’s knowledge of who the child’s friends are, where the child is after school and at night, what the child does during free time, and how the child spends money. Response categories were as follows: 1, “doesn’t know”; 2, “knows a little”; and 3, “knows a lot.” Parental curfews was a 2-item measure indicating adolescent reports of the latest time that the mother allows the youth to stay out on school and weekend nights. Original response categories were as follows: 1, “as late as I want”; 2, “before 8:00”; 3, “8:00 to 8:59”; 4, “9:00 to 9:59”; 5, “10:00 to 10:59”; 6, “11:00 or later”; and 7, “I am not allowed out Sunday through Thursday [Friday or Saturday] nights.” These data were recoded so that higher values indicated stricter or earlier curfews. Family routines was a 5-item measure indicating the mother’s reports of the extent to which the family had daily routines for talking or playing quietly, for eating meals, and for children doing homework and going to bed. Response categories were as follows: 1, “almost never”; 2, “sometimes”; 3, “usually”; and 4, “always.” Scales demonstrated adequate reliability (α = .69 for curfews, α = .71 for parental monitoring, and α = .68 for family routines). Mean (SD) scores were 0.65 (0.19) for parental curfews, 0.89 (0.12) for parental monitoring, and 2.76 (0.67) for family routines.

Because previous research indicates an increased risk of early sex initiation among male youth,24 older youth, African American and Latin American (vs white) youth,1 and youth from lower socioeconomic status backgrounds,15 we included variables to indicate the adolescent’s sex (0 indicates male; and 1, female), age, race/ethnicity (0 indicates African American; 1, Latino; and 2, white), and parent education (0 indicates less than high school; 1, high school degree or general equivalency diploma; and 2, more than high school). Although early initiation of sex is associated with living in a “nonintact” family structure,25 many racial and ethnic minority youth live in extended kin households that potentially provide the same level of supervision found in 2-parent families. Thus, we created the following family structure categories: 0, 2 biological or adoptive parents; 1, mother plus stepfather or male partner; 2, mother plus extended kin; 3, mother only; and 4, no biological or adoptive mother. We also included a measure of maternal work hours (0, not working; 1, work <40 hours; and 2, work ≥40 hours) and of the city in which the interview took place (dummy coded with Boston as the reference group).

Data analysis

We ran cross tabulations with χ2 tests of significance and comparisons of means with 1-way analysis of variance to examine bivariate associations between early sex initiation and the following variables: times 1 and 2 measures of out-of-school care; and time 1 measures of adolescent age, adolescent sex, adolescent race or ethnicity, parental monitoring, parental curfews, family routines, family structure, parent education, maternal work hours, and city of interview. We used findings from bivariate analyses of sex initiation by out-of-school-care arrangements at both assessment times to develop a variable that describes patterns of out-of-school care over time. We next ran a logistic regression model in which sex initiation by time 2 was regressed on patterns of out-of-school care over time, all time 1 parenting measures (because of their conceptual relevance to our model), and time 1 adolescent and family demographic variables that were statistically significant at P<.05 in bivariate analyses. Analyses were conducted using a commercially available software program (SPSS, version 11.0).26


As shown in Table 2, 16.4% (n = 81) of the adolescents in this study initiated sex by time 2. Significantly more adolescents initiated sex if, at the time 2 assessment, they were in an organized/supervised activity, outside the home with a relative or nonrelative, or in self-care, compared with being at home in maternal care or at home with the father or mother’s male partner or another adult. Although the overall χ2 value did not reach statistical significance for the time 1 assessment of out-of-school care, the pattern of association with early sex initiation was similar to that shown at time 2. When compared with those who did not initiate sex at time 2, adolescents who did initiate sex were older and more likely to be male.

Table 2. 
Sex Initiation by Out-of-School Care at Times 1 and 2 and Demographic and Parenting Characteristics for the 494 Participants
Sex Initiation by Out-of-School Care at Times 1 and 2 and Demographic and Parenting Characteristics for the 494 Participants

Based on findings for bivariate associations between sex initiation and out-of-school-care arrangements at each assessment time, we developed a measure indicating 4 patterns of change in out-of-school care from time 1 to time 2. These patterns were coded as follows: 0, remained in adult care at home; 1, moved from self-care or out-of-home care into adult care at home; 2, moved from at-home adult care to self-care or out-of-home care; and 3, remained in self-care or out-of-home care at both times. Results from multivariate analyses (Table 3) show that adolescents who remained in out-of-home care or self-care at both assessment times were more likely to initiate sex than were adolescents remaining at home with an adult. These effects held when accounting for the adolescent’s sex and age and levels of parental monitoring, parental curfews, and family routines. Consistent with bivariate results, the likelihood of sex initiation was greater among older compared with younger adolescents and was lower among females. There were no statistically significant associations between sex initiation and parenting variables or family structure.

Table 3. 
Data for Early Initiation of Sex by Patterns of Out-of-School Care Over Time
Data for Early Initiation of Sex by Patterns of Out-of-School Care Over Time


Our understanding of how social contexts affect adolescent sexual risk has emanated largely from studies of parenting and, more recently, of neighborhoods. This study’s results suggest that out-of-school care represents an additional social context salient to the risk of early sexual intercourse among young adolescents living in low-income urban areas. Contrary to our expectation that latchkey adolescents would be at greater risk for early sex initiation than adolescents in any form of adult-supervised care, results from our bivariate analyses suggested that early sex initiation was similarly higher among adolescents in out-of-home care and self-care. Findings from prospective analyses indicated that young adolescents who spent their out-of-school hours either in self-care or outside their own home, such as at a friend’s house, in an organized activity, or at the home of a relative, at each assessment time were 2.5 times more likely to initiate sex than those who remained at home with an adult. We did not find support for the idea that positive parenting practices helped account for the relationship between at-home adult care and reduced sex initiation.

In interpreting findings for out-of-school care, it is useful to consider that families in this study were all low income and lived in either poor or near-poor central city neighborhoods of Boston, Chicago, and San Antonio. Based on previous research, adolescents in this study likely faced considerable neighborhood dangers and risks, such as violence, drug use, early childbearing, and joblessness.27 In this way, being at home with an adult may have reduced a youth’s opportunities for engaging in early sex by reducing their exposure to risky social contexts external to the family. These structural barriers seemed to represent a more effective means of deterring early sexual intercourse than did explicit parental efforts aimed at changing the adolescent’s behavior. Parents’ knowledge of the youth’s whereabouts and activities, having rules about curfews, and maintaining familial routines showed no statistically significant relationships with early sex initiation.

Our study does not account for the possibility that adolescents manifesting more problem behavior (often correlated with early sex initiation) elicit negative feelings and reactions by their families such that they are less likely to be cared for at home by an adult when compared with youth exhibiting fewer problem behaviors. In this way, being cared for outside of the home might represent a consequence rather than a predictor of early sex initiation. We were also unable to account for possible changes in the specific nature of out-of-school care occurring during the 16-month period between assessment times. Despite limitations regarding measures of out-of-school care and adolescent sex initiation over time, issues of reverse causality (ie, mothers altering out-of-school care in response to the youth initiating sex after time 1) do not affect the validity of our findings regarding out-of-school patterns and sex initiation. This is because changes in out-of-school care were unrelated to the likelihood of sex initiation. Only when youth remained in self-care or out-of-home care did they experience an increased likelihood of sex initiation compared with those remaining in at-home adult care. It is unclear whether findings from this study would generalize to more economically advantaged families or to families living in suburban and rural areas.

As welfare reform legislation continues to mandate that poor mothers join the labor force, future research should consider factors affecting low-income families’ decisions regarding out-of-school care for adolescents and, in turn, the health of adolescents. It may be useful, for example, to understand how low-income mothers’ decisions regarding out-of-school care are influenced by the availability of kin and adult support networks, by perceived neighborhood conditions, and by the structure and nature of the job market. In this way, it may be possible to develop policies and programs that can support low-income families’ efforts in raising their adolescent youth and, in turn, potentially help reduce rates of early sexual intercourse.

Correspondence: Kathleen M. Roche, MSW, PhD, Department of Population and Family Health Sciences, The Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe St, Baltimore, MD 21205 (kroche@jhsph.edu).

Back to top
Article Information

Accepted for Publication: August 19, 2004.

Funding/Support: This study was supported by grant RO1 AI36986 from the National Institutes of Allergy and Infectious Diseases, Bethesda, Md (Dr Ellen).

 Sexually experienced teens: the percentage of students in grades 9 through 12 who have ever had sexual intercourse, selected years 1990-2000. Available at:http://www.childtrendsdatabank.orgAccessed March 1, 2004
Koyle  PJensen  LOlsen  J  et al.  Comparison of sexual behaviors among adolescents having an early, middle, and late first intercourse experience.  Youth Soc 1989;20461- 475Google ScholarCrossref
Seidman  SNMosher  WDAral  SO Predictors of high-risk behavior in unmarried American women: adolescent environment as risk factor.  J Adolesc Health 1994;15126- 132Google ScholarCrossref
Smith  CA Factors associated with early sexual activity among urban adolescents.  Soc Work 1997;42334- 346Google ScholarCrossref
Moore  KAMiller  BCGlei  DMorrison  DR Adolescent Sex, Contraception, and Childbearing: A Review of Recent Research.  Washington, DC Child Trends Inc1995;
Small  SALuster  T Adolescent sexual activity: an ecological, risk-factor approach.  J Marriage Fam 1994;56181- 192Google ScholarCrossref
Romer  DStanton  BGalbraith  JFeigelman  SBlack  MMLi  X Parental influence on adolescent sexual behavior in high-poverty settings.  Arch Pediatr Adolesc Med 1999;1531055- 1062Google ScholarCrossref
Smith  K Who’s Minding the Kids? Child Care Arrangements Current Population Reports Fall1995; Washington, DC US Dept of Commerce, Economics and Statistics Administration, US Census Bureau2000;Google Scholar
French  DCDishion  TJ Predictors of early initiation of sexual intercourse among high-risk adolescents.  J Early Adolesc 2003;23295- 315Google ScholarCrossref
Richardson  JLDwyer  KMcGuigan  K  et al.  Substance use among eighth-grade students who take care of themselves after school.  Pediatrics 1989;84556- 566PubMedGoogle Scholar
Galambos  NLMaggs  C Out-of-school care of young adolescents and self-reported behavior.  Dev Psychol 1991;27644- 655Google ScholarCrossref
Rodman  HPratto  DJNelson  RS Child care arrangements and children’s functioning: a comparison of self-care and adult-care children.  Dev Psychol 1985;21413- 418Google ScholarCrossref
Steinberg  L Latchkey children and susceptibility to peer pressure: an ecological analysis.  Dev Psychol 1986;22433- 439Google ScholarCrossref
Richardson  JLRadziszewka  BDent  C  et al.  Relationship between after-school care of adolescents and substance use, risk taking, depressed mood, and academic achievement.  Pediatrics 1993;9232- 38PubMedGoogle Scholar
Baumer  EPSouth  SJ Community effects on youth sexual activity.  J Marriage Fam 2001;63540- 554Google ScholarCrossref
Brewster  KL Neighborhood context and the transition to sexual activity among young black women.  Demography 1994;31603- 614Google ScholarCrossref
Brewster  KL Racial differences in sexual activity among adolescent women: the role of neighborhood characteristics.  Am Soc Rev 1994;59408- 424Google ScholarCrossref
Fomby  PEstacion  AMoffitt  R Welfare, children, and families: A Three-City Study: characteristics of the Three-City Study: time 1 and time 2. Available at:http://www.jhu.edu/~welfare/CharacteristicsReport.pdfAccessed March 1, 2004
Zaslow  MJMoore  KABrooks  JL  et al.  Experimental studies of welfare reform and children.  Future Child 2002;1278- 95Google Scholar
Cain  VSHofferth  SL Parental choice of self care for school-age children.  J Marriage Fam 1989;5165- 77Google ScholarCrossref
Uttal  L Using kin for child care: embedment in the socioeconomic networks of extended families.  J Marriage Fam 1999;61845- 857Google ScholarCrossref
Cherlin  AJ Welfare, children, and families: A Three-City Study, wave 1 (data set 01-04, Ruiz RO, McKean EA, Peterson JL) machine-readable data file and documentation.  Johns Hopkins University (Producer) Los Altos, Calif Sociometrics Corp, Child well-being and poverty data archive (Producer & Distributor)1999;Google Scholar
 Welfare, children and families: A Three-City Study. Available at:http://www.jhu.edu/~welfare/overviewanddesign.pdfAccessed March 1, 2004
Upchurch  DMAneshensel  CSSucoff  CA  et al.  Neighborhood and family contexts of adolescent sexual activity.  J Marriage Fam 1999;61920- 933Google ScholarCrossref
Davis  ECFriel  LV Adolescent sexuality: disentangling the effects of family structure and family context.  J Marriage Fam 2001;63669- 681Google ScholarCrossref
 SPSS for Windows: Standard Version 11.0.01: Release 15.  Chicago, Ill SPSS Inc2001;
Brooks-Gunn  JDuncan  GJAber  JL Neighborhood Poverty: Volume I: Context and Consequences for Children.  New York, NY Russell Sage Foundation1997;