Many African American parents and caregivers raise children in public housing complexes in inner-city communities that have been compared to the world‘s war zones (Gabarino, 1995). Ethnographic data suggest that parents in these public housing communities devote an enormous amount of time and energy to monitoring their children’s whereabouts, friendships, and activities (Jarret, 1990, 1995; Osby, 1993). Within the last decade, soaring rates of HIV/AIDS fueled by risky sexual behavior has become an added risk for urban minority communities. African Americans are disproportionately represented among HIV/AIDS cases in the United States and the majority of cases occur in poor urban communities (CDC, 2002).

Prior research has demonstrated an association between parental monitoring and sexual risk-taking among youth (Miller, Forehand, & Kotchick, 1999; Rodgers, 1999). Parental monitoring appears to limit adolescents’ opportunities to have sex (Kotchick, Shaffer, Miller, & Forehand, 2001; Repetti, Taylor, & Seeman, 2002). Parental monitoring has also been consistently related to delayed initiation of sexual activity (Murry, 1994), development of self-regulation and emotional regulation skills, redirection of adolescents away from peers who engage in high-risk sexual and other risky behavior (Bates & Labouvie, 1995; Crosby et al., 2000) and less willingness to have sex (Crosby et al., 2000). Several studies have examined the impact of parental monitoring on sexual risk-taking specifically among urban African American families (e.g., Crosby et al., 2001; Mandara, Murray, & Bangi, 2003; Tolou-Shams, Paikoff, McKirnan, & Holmbeck, in press). There are, however, scant empirical investigations of the determinants of successful parental monitoring in these settings (see Jones, Forehand, Dorsey, Foster, & Brody, 2005, for an exception).

Parenting research suggests that personal characteristics of a parent, his/her child, and the context in which the parenting occurs affect parenting skills and competencies (McLoyd, 1990; Luster & Okagaki, 1993a,b). For example, a high level of parental distress has been linked to poor monitoring among poor, urban families (Jarret, 1990; McLoyd, 1990). Likewise, having a child as a teen has been associated with inadequate child-rearing strategies and with poor adjustment in children (Black et al., 2002). Adolescent factors are also considered relevant to parenting quality in low-income communities. Being receptive to parental warnings and rules can help to limit negative peer and street-life influences (Osby, 1993). In addition, parental monitoring varies by adolescent age and gender. Boys display higher rates of risk behaviors, such as delinquency, aggression and sexual activity, but also are monitored less by their parents than are girls (Richards, Miller, O’Donnell, Wasserman, & Colder, 2004). As children get older, parents also monitor them less, which may be developmentally appropriate, but in urban public housing, close monitoring may still be needed (Richards et al., 2004).

Contextual factors, such as instrumental and emotional support for parents, can also facilitate more effective parenting (McAdoo, 1982; McLoyd, 1990). Inner-city African American families rely on various configurations of extended households to provide support, which may have important benefits in monitoring youth. Intergenerational households, in particular, seem to have a positive effect on monitoring in that they provided a greater ratio of adults to youth, offering additional eyes and ears to keep track of youth. Similarly, instrumental help (e.g. financial assistance) from other adults may have a direct effect on monitoring by providing additional caregivers (McAdoo, 1982). Such help might also come from the mother’s partner living within or outside the home (Coley, 2001). Jones and colleagues (2005) recently found that African American mothers who reported higher levels of conflict with and lower levels of support from co-parenting adults reported the lowest levels of parental monitoring.

Friendship networks also provide emotional and material help for single, unemployed mothers living in poverty, and reciprocal activities within friendship networks can supplement maternal resources and make survival possible (Scott & Black, 1994). Religious guidance via church members and ministers has also been found to be an important source of support for families living in poverty (Osby, 1993; Taylor & Chatters, 1991). A sense of guidance from religion may also be crucial in helping parents facing poverty to maintain positive self-evaluations and self-worth (Baptiste, 2006). This might indirectly affect parents’ energy and commitment to monitoring their children.

Our study seeks to determine the association between parental monitoring and preadolescent sexual risk situations, and in turn, investigates parent, preadolescent and contextual factors associated with parental monitoring among urban, low-income African American families. On the basis of prior literature, our hypotheses are: (1) increased parental monitoring would be directly associated with less exposure to preadolescent sexual risk situations; and (2) that parental risk factors (e.g., more psychological distress), youth risk factors (e.g., more peer pressure), and contextual factors (e.g., less social support) would be directly associated with less parental monitoring.

Method

Participants

The study sample consisted of 309 African American 4th and 5th graders and their parents who were part of the Chicago HIV Prevention and Adolescent Mental Health Project (CHAMP), a longitudinal study of familial and contextual influences on HIV/AIDS risk exposure (Paikoff, Holmbeck, & Brooks-Gunn, 1993). The entire sample of parents and youth lived in two public housing developments in impoverished parts of the city of Chicago.

Procedures

Five elementary schools serving the two public housing projects were approached to participate in the larger CHAMP study. Youth were given flyers at school to take home to parents, who then contacted the research team to express interest. After obtaining informed consent, each parent and youth were interviewed privately at a university site. At a later date, with parental consent, each youth was interviewed privately at school and completed a separate assessment about peer relations and sexual risk behavior. Measures used in the current study were drawn from the larger assessment battery used in the CHAMP study.

Measures

Parental monitoring

Both parent and youth reports of parental monitoring were used in the study. Parent Report of Monitoring was assessed using a 36-item scale (α = .82) developed by Gorman-Smith, Tolan, Zelli, & Huesman (1996) including items related to a) parental awareness of youth’s whereabouts, activities, and friends; b) effectiveness of rules and discipline; and c) parental involvement with youth. Each item was rated on a 5-point scale, with responses ranging from 1 (none/very low supervision) to 5 (very high supervision). Higher scores indicated closer monitoring of youth.

Youth Report of Monitoring (Gorman-Smith et al., 1996) was also assessed via a 36-item scale (α = .80) asking youth about how their parents monitored them. These items were the same as for parents, but were written to assess youth perceptions. Again, items were rated on a 5-point scale with responses ranging from 1 (none/very low supervision) to 5 (very high supervision). Higher scores indicated youth perceptions that parents more closely monitored them.

Sexual risk situations were assessed with the Situations of Sexual Possibility measure (Paikoff et al., 1993), which consists of 14 multi-part questions that together assess (a) the extent to which preadolescent youth were in sexual risk situations, (b) youth’s use of contraception, and c) discussion of these issues with parents and other adults. Only questions about the extent to which youth were in sexual risk situations were used in the present study. This included a total of 26 items coded by five levels of possible sexual situations: (1) no contact with opposite-sex peers outside of school (n = 44, lowest risk); (2) out-of-school, nonprivate contact with opposite-sex youth (n = 107); (3) limited private contact (only one occasion and/or less than an hour alone, n = 72); (4) frequent and/or prolonged private contact (n = 52), but no one-on-one private contact, and (5) one-on-one private contact (n = 10), but no sexual contact. Finally, youth who either engaged in sexual activity or observed others engaging in sexual activity (n = 16) were assigned a score of 6. Therefore, scores ranged from 1 to 6, with higher scores indicating greater exposure to sexual risk situations. Eight youth did not provide sexual risk data.

Parent factors

The anxiety and depression subscales of the Symptom Checklist Revised (SCL-R) (Derogatis, 1992) were used to assess psychological distress. Each of the 40 items (α = .95) was rated on a 5-point scale of distress, from 0 (not at all distressful) to 5 (extremely distressful). Higher scores indicated greater psychological distress. Maternal age at first birth was computed using data provided by biological mothers (n = 269) in the CHAMP demographic questionnaire (Paikoff et al., 1993) and was categorized as 1 (maternal ages 22 and above); 2 (ages 20–21), 3 (ages 18–19), 4 (ages 16–17), and 5 (age 16 or below). Two scores greater than 50, which were determined to be outliers, were coded as missing, yielding a total of 267 valid scores for this variable.

Youth factors

Responsiveness to parents was assessed with a 12-item scale (α = .75) adapted by Baptiste (2000) from a Gorman-Smith et al. (1996) measure and completed by parents. Parents gauged youth’s responsiveness to their discipline and guidance, ranging from 1 (very responsive) to 3 (not at all responsive). Higher scores indicated less responsiveness to parental discipline . Items assessed areas such as whether punishment worked with the child, if parents thought that the child responded to discipline, and whether the child was easy to control without the help of other adults. Susceptibility to peer pressure was measured by a 42-item scale (Paikoff et al., 1993) (α = .94) that assessed the degree (ranging from 1, not at all, to 4, very likely) to which children were pressured by their friends to maintain friendships, even if it meant engaging in risky behavior. Items assessed pressure experienced in less serious risk behavior (e.g., making fun of another friend) to more serious infractions (e.g., drinking alcohol, smoking marijuana). Child gender was determined from information provided in the demographic questionnaire (Paikoff et al., 1993).

Contextual factors

Parents completed seven measures assessing their supportive networks. The support measures were placed into two categories, instrumental support and emotional support, to capture conceptual distinctions.

Instrumental support

Presence of a partner (no partner = 0, partner = 1) was assessed from information provided in the CHAMP demographic questionnaire. The parenting help scale (Kellam, Branch, Agrawal, & Ensminger, 1975) measured parents’ reports of the amount of actual help they received from individuals within or outside their household. Scores represented the total number of individuals helping with the child. Household type was determined from a household roster parents provided (Paikoff et al., 1993). Households were categorized as either two-generation households (parents lived alone with their children) (coded as 0) or three-generation households (grandparents lived in the household) (coded as 1).

Emotional support

Perception of religious guidance was measured by a single item from the CHAMP parent assessment (Paikoff et al., 1993). Parents were asked, “Does your religion provide you with guidance in daily living?” Reponses were coded on a 5-point scale ranging from of 1 (none) to 5 (a lot). Friendship support was assessed by three separate variables: the proximity of friends, frequency of friendship activities, and friendship quality, taken from the CHAMP parent assessment (Paikoff et al., 1993). Friends’ proximity measured whether parents had at least one close friend living in the area (no = 0, yes = 1). Frequency of friendship activities measured how often parents got together with a friend for a recreational activity, from “never” (1) to “several times a week” (5). Friendship quality was measured using a 6-point scale ranging from “not good” to “very good” with higher scores indicating parent perception of higher friendship quality.

Data analyses

Prior to any statistical analyses, all data were screened for missing data and outliers. Distribution of normality was assessed by testing the distribution skew, using the Fisher skewness coefficient (Pett, 1997). All variables were approximately normally distributed. In addition to the two cases deemed to be data entry error and coded as missing, one score of child responsiveness was also coded as missing for the same reason.

Descriptive statistics, including frequencies, means and standard deviations and correlations for all indicator variables were calculated using SPSS Version 12.0 (2003) during the preliminary data analysis stage. LISREL 8.54 (Joreskog & Sorbom, 2003) was used to test the main hypotheses. The full information maximum likelihood (FiML) feature of LISREL was used to compensate for missing data, which was only 3.6% of total cells. FiML uses all available data to predict model parameters and is preferred to pairwise or listwise deletion procedures in that there is less bias in the parameter estimates (Arbuckle & Wothke, 1999).

Maternal age and parental psychological distress were modeled as observed indicators of the latent construct “parent risk;” susceptibility to peer pressure and youth responsiveness were modeled as indicators of “youth risk;” total help, household type, and partner presence were modeled as indicators of “instrumental support,” and frequency of friend activities, friend proximity, friendship quality, and perception of religious guidance were modeled as indicators of “emotional support.” Both youth report and parents’ self-report served as indicators of parental monitoring. The sexual risk situations construct was represented by a single variable.

In the first model (Fig. 1), referred to as the “full model,” parent risk, youth risk, and support variables were modeled as predictors of parental monitoring and sexual risk situations, and monitoring was modeled as an additional predictor of sexual risk situations. The second model (Fig. 2), which was hypothesized to be the best-fitting model, proposed parent risk, youth risk, and the support variables as predictors of parental monitoring only, and monitoring as the sole predictor of sexual risk situations. These models are shown as Figs. 1 and 2 respectively.

Fig. 1
figure 1

Full model of interrelationships among study variables

Full size image
Fig. 2
figure 2

Results from best-fitting model

Full size image

Analyses were also run to test whether hypothesized paths differed for boys and girls. Widamon (1985) has advocated using nested models to test the invariance of various paths across groups. Using the full model (Fig. 1), a two-group model in which all paths were modeled to be invariant across males and females was first run to obtain a baseline chi-squared value. Nested models in which each path was allowed to vary across males and females were then run and the difference in chi-squared values with the associated difference in degrees of freedom were plotted on a standard chi-squared distribution table to assess whether freeing the path of interest resulted in a statistically significant improvement in fit. Also, because age was expected to negatively co-vary with the level of parental monitoring (Osby, 1993; Richards et al., 2004) additional analyses assessed age as a predictor of both parental monitoring and youth’s sexual risk situations.

Results

Parents were nearly all women (98%), of which 89% (n=269) were the biological mothers of children in the study. Sixty-four percent (n = 201) of these birth mothers were teenage parents (age 19 or less). Nearly one-half (46%) of parents received their GED or had attended high school, and about one-half had never worked outside the home. The majority of parents (75%) lived “reportedly” without partners in the home and were raising their children in poverty (68% reported an annual income less than $10,000). Within the sample of youth, 175 were female and 134 were male (mean age = 10.92 years, range, 9–12). Approximately one-half (49%) had at least some private contact with an opposite-sex peer. Of these riskier youth, 52% (n = 78) had at least prolonged private contact and 21% (n = 16) of those reporting prolonged or frequent contact specifically reported either observing or having sexual intercourse with an opposite-sex peer. See Table 1 for descriptive data.

Table 1 Descriptives for all study variables (N = 309)
Full size table

Individual associations with monitoring and sexual risk situations

Results of bivariate correlations are presented in Table 2. Parent reports of monitoring were found to be significantly associated with parent psychological distress; youth responsiveness; perception of religious guidance; friends’ proximity; frequency of friendship activities; friendship quality; and youth report of parental monitoring. Owing to higher youth-responsiveness scores (indicating less responsiveness), the negative association of youth responsiveness with parental monitoring suggests that if parents perceived youth as more responsive, they provided more monitoring. Similarly, psychological distress was also negatively related to parental monitoring, indicating that less-distressed parents provided closer monitoring than more-distressed parents. Youth reports of parental monitoring were associated only with parent reports of monitoring and youth responsiveness, with more responsive youth being more closely monitored.

Table 2 Correlations for all study variables (N = 307)
Full size table
Table 3 Loading of each observed variable on respective latent variables
Full size table

Youth sexual risk situations were found to be associated with peer pressure and youth responsiveness. Specifically, greater peer pressure and lower responsiveness were associated with greater sexual risk situations. Furthermore, parent reports of monitoring showed a marginal negative association with youth sexual risk situations.

Structural equation model results

Table 3 shows the loading of each observed variable on its respective latent factor. Neither maternal age nor parental distress loaded significantly on the parent risk factor. Each of these variables had a limited range of scores, which may explain the poor loadings. The nonsignificant loading of peer pressure on the youth risk construct is somewhat puzzling, whereas the high, yet nonsignificant, loading of youth responsiveness may stem primarily from measurement error. All instrumental support variables loaded well, with the loading for total help being especially strong. The same was true for emotional support, with the exception of perception of religious guidance, which that may have loaded poorly owing to its dissimilarity from the friendship variables. Finally, both youth and parent reports of monitoring loaded significantly on the monitoring construct.

The full model (Fig. 1) in which parent risk, youth risk, instrumental support, and emotional support were modeled as predictors of both parents’ monitoring and youth sexual risk situations, with monitoring as an additional predictor of sexual risk situations, showed adequate fit with the data, χ2 (68, N = 309) = 162.31, RMSEA = 0.067. As hypothesized, instrumental (B = .10, t = 2.22, p < .05) and emotional support (B = .11, t = 3.08, p < .01) predicted greater parental monitoring, but neither parent nor youth risk variables predicted parental monitoring. Furthermore, none of the variables was a significant predictor of youth sexual risk situations.

Figure 2 shows detailed results from the second, more parsimonious model in which parent risk, youth risk, instrumental support, and emotional support were modeled only as predictors of parental monitoring, with monitoring again modeled as a predictor of youth sexual risk situations. This model showed a similar fit with the data when compared with the full model, χ2 (72, N = 309) = 163.98, RMSEA = 0.064. Instrumental (B = .10, t = 2.19, p < .05) and emotional support (B = .11, t = 3.19, p < .01) again predicted greater parental monitoring, and, as hypothesized, greater parental monitoring predicted fewer sexual risk situations (B = −.12, t = −2.44, p < .01). The latter model, therefore, was a more parsimonious and theoretically derived model, as evidenced by a minimal increase in the χ2 value when the paths from youth and parent risk factors to sexual risk situations were deleted.

Results of analyses by gender

The baseline model with all paths held invariant across males and females showed adequate fit, χ2 (173, N = 309) = 292.31, RMSEA = 0.067. We allowed to vary, one at a time, across males and females each of the paths from parent risk, youth risk, instrumental support, and emotional support to both parental monitoring and youth sexual risk-taking behavior, and from monitoring to sexual risk. There were no models in which freeing an individual path resulted in a significant decrease in the chi-square value, therefore indicating that each of the relationships was similar for males and females.

Finally, follow-up analyses with age showed age to be negatively associated with youth (r = −.13, p < .05) and parent (r = −.19, p < .05) reports of monitoring, but it was not associated with sexual risk situations (r = .09, ns.). Furthermore, the relationship of monitoring to sexual risk situations did not vary systematically across age groups.

Discussion

Close parental monitoring is one strategy parents use to buffer youth against the dangerous environment in low-income, public housing, including exposure to HIV/AIDS through risky sexual contact. Approximately 25% of early adolescents in our study reported prolonged or frequent private contact with opposite sex peers, and parental monitoring was found to be a successful strategy in reducing the frequency of these adolescent situations. In addition, instrumental and emotional support from friends and family were important predictors of parental monitoring, which has implications for the development of successful family-based adolescent interventions. The final model did not support youth and parent risk factors as predictors of parental monitoring. However, bivariate associations suggested that youth responsiveness to parental guidance is also associated with both youth and parents’ reports of parental monitoring and, therefore, might be considered for intervention development.

Consistent with the literature, we found that greater monitoring was associated with less exposure to sexual risk situations. Parental monitoring remained a significant predictor of adolescent sexual risk, even when accounting for other important parenting and child attributes. Thus, one avenue for family-based interventions to deter adolescent HIV risk among African American adolescents should focus on improving parenting strategies, such as parental monitoring. This study is unique in that it pinpoints several specific aspects of parental monitoring that could be incorporated into interventions to reduce adolescent risk, such as teaching parents how to gain instrumental and emotional support from individuals within and outside the household. Our findings demonstrate that parents who perceive themselves as having more help with daily parenting tasks and better quality friendships report closer monitoring of their children. This confirms that parents raising children in the difficult circumstances of public housing may indeed benefit from the help of others, who can alleviate the emotional toll and physical tasks associated with monitoring (Miller, McKay, & Baptiste, in press).

Our findings did not, however, suggest that psychological distress was associated with parental monitoring, after accounting for other youth and parent factors, such as parental support. It is plausible that the presence of multiple avenues of support might mitigate the effects of psychological distress. Prior studies have found that even if parents are psychologically distressed, accessing support may improve their ability to carry out key tasks, such as supervision and monitoring (McAdoo, 1982; McLoyd, 1990). Facilitating positive neighborhood connections, bringing neighbors together to talk about dangers in their community, networking with institutions that can provide recreational opportunities for youth, and looking out for each other’s children may be beneficial interventions for youth safety.

Given that this is one of the few studies to examine predictors of parental monitoring in relation to adolescent sexual risk taking among African American families in public housing, factors that were not significant in model analyses, but were associated with parental monitoring in bivariate analyses deserve some mention. For example, adolescent responsiveness to parental discipline and guidance was independently related to parental monitoring. This finding suggests that particular youth characteristics, relating to responding and heeding parental advice and guidance, may affect how closely a parent will keep track of the youth. As hypothesized, youth who are less responsive to parental guidance and discipline may be more vulnerable to risks in public housing, in part because they may rebuff parents’ attempts to guide them. This result should be interpreted with caution, however, as the relationship between youth responsiveness and parental monitoring diminished when accounting for other factors, such as parental support. Nevertheless, in addition to a focus on the characteristics of parents and families that facilitate or deter youth risk behavior, programs to remediate or prevent risk would be wise to attend to bidirectional interactions between parents and youth and how these interactions may influence monitoring and risk.

The importance that many parents place on religious guidance as a source of support may also influence monitoring. Parents’ perceptions of guidance from their religion were independently related to their, but not youth, reports of monitoring. Although the measure of religious guidance in this study is not robust, the finding does indicate a need to further explore the place of religion in strengthening parental behavior. For many reasons, sources of religious support are not often targeted in prevention and intervention activities with urban families. This study, however, suggests it is worthy of exploration as a potentially protective factor for parents and their children (Baptiste, 2006).

The negative skew in parent reports of monitoring should be noted. The majority of parents reported close monitoring of their children, but this is consistent with the view that parents in distressed public housing communities provide “stringent” and perhaps restrictive watch over their children (Jarret, 1990, 1995; Osby, 1993). Other limitations in the study should be taken into account both in interpreting the results and in contemplation of future research. First, measures selected for this study were not predictive of youth perceptions of parental monitoring and this was disappointing. Indeed, a youth perception of being monitored may have a more direct impact on their own behavior (Statin & Kerr, 2000). Therefore, factors that may predict youth perceptions of monitoring are worthy of additional exploration. Second, measures of parental monitoring only focused on the monitoring activities of one individual, the primary caregiver. However, a network of primary caregivers may monitor youth growing up in public housing, involving kin and non-kin living within and outside the home. Assessing the strength of monitoring networks in which youth may be involved could yield a more accurate picture of monitoring in public housing. Third, our study is cross-sectional and only examines monitoring when children were preadolescents. But parental monitoring is critical during all stages of childhood and adolescence, and parents and children do change. Further, risks in the inner city may be increased for older adolescents. Therefore, a longitudinal analysis of parental monitoring would be a useful undertaking by providing a developmental understanding of monitoring, which would indicate factors affecting parental monitoring across the stages of childhood and adolescence.

In inner-city public housing, stringent parental monitoring is one strategy to buffer youth against the socially toxic conditions that surround youth. This includes significant of sexual risk-taking, which can increase exposure to HIV/AIDS. This study found preliminary support for a model that could inform primary and secondary efforts to strengthen and support parents’ monitoring capabilities as a way to reduce HIV/AIDS risk (and other sexually transmitted diseases) among low-income, urban African American adolescents. The focus on individual (i.e., parent and youth) and contextual (i.e., at the family level) factors in the study, however, should not obscure the need to evaluate the macro-social conditions that sustain poverty and social disorganization in inner-city areas. Inner-city parents continue to be held accountable for raising children who will stay healthy, stay out of trouble, and grow up to be responsible adults. Strengthening and supporting the monitoring capabilities of parents is one way to help families to carry out this critical responsibility. Supporting parental monitoring activities will not only have an impact on children’s physical and emotional well-being, it may well improve the quality of parents’ lives. Closely monitored children may take fewer sexual risks and avoid other potentially harmful behaviors that stir concern (Jarret, 1995). Having fewer worries can redirect parents’ energies to attend to their own development and improved parenting.