Search Results

Method: The study used a sample of 10- to 15-year-old youths (N=953) from the 2004 and 2006 waves of the National Longitudinal Survey of Youths and the Mother-Child datasets. Four possible responses to the question "I don't feel safe at this school," were dichotomized to reflect some degree of feeling unsafe in school versus "not at all true." Groups of variables included youth characteristics (age, race/ethnicity, gender, behavior problems), maternal/family characteristics (marital status, mothers' education, poverty), school environment (ease of making friends, teacher's involvement, school rule enforcement, weapon carrying), neighborhood environment (neighborhood safety, area of residence), and parenting practices (discussions with youths, school involvement). Weighted descriptive statistics were computed, and models were estimated using multivariate logistic regression.

Results: Approximately 30% of the youths felt that their schools were unsafe to some degree. Of the youth characteristics, only age was statistically significantly related to feeling unsafe in school (OR=1.15), and none of the maternal/family characteristics were significant. Within the school environment, youths repor ting higher levels of ease in making friends at school (OR=.65), teachers being more involved in their school life (OR=.75), and the school enforces rules (OR=.67) were less likely to perceive their schools as unsafe. Youths observing someone carrying a weapon in school were 1.68 times more likely to perceive an unsafe school. Within the neighborhood environment, youths perceiving their neighborhoods as safe were less likely to feel unsafe in school (OR=.62), and those residing in a metropolitan area not in a central city, compared with youths living in a central city, were .68 times as likely to perceive their schools as unsafe. Finally, youths whose parents were more involved in their schools were less likely to feel unsafe in their schools (OR=.90).

Implications: These findings support the importance of examining interactions within multiple system levels and indicate that school and community environments and parents' involvement in the school have important influences on students' perceptions of an unsafe school. These results suggest that practitioners who assist schools in creating environments that are supportive, yet enforce rules, free of weapons, and involve parents will increase students' perceptions of a safe school. Advocating for community interventions, such as neighborhood watches, and social policies to decrease violence in communities also might increase youths' perceptions that their schools are safe.

To address these hypotheses, secondary data analysis was conducted, using the National Longitudinal Survey of Youth. The sample was drawn from the mother-child dataset, which included youth who in the first of two years, 2002 or 2004 (Time 1), were living with their mothers, enrolled in regular school, responded to at least one of the 13 items from the self-administered survey, and the mothers responded to at least one of the four items measuring peer and school externalizing behaviors in Time 1 and Time 2 (in 2004 for those entering the sample in 2002; in 2006 for those entering the sample for 2004). Multivariate hierarchical logistic regression model were estimated to address the hypotheses.

Findings from the study indicate that youth's learning problems and peer externalizing behavior at Time 1 were significantly associated with peer externalizing behavior at Time 2. When the microsystem variables were included in Model 2, ease of making friends was statistically significant. When the exosystem variables were added in Model 3, the neighborhood environment variables were all statistically significant, but none of the macrosystem variables were significant when added to Model 4. Concerning school externalizing behavior at Time 2, male gender and school externalizing behavior at Time 1 were statistically significant, and two microsystem variables--cognitive stimulation and negative peer influence--were significantly associated with school externalizing behavior at Time 2. None of the exosystem and macrosystem variables were associated with school externalizing behavior at Time 2. With regards to the moderators, I found that for Hispanics, higher levels of cognitive stimulation was associated with an increased risk of exhibiting school externalizing behavior, although the odds ratio indicated little practical significance. I also found that ease of making friends also moderated the effects of learning problems on school externalizing behavior at Time 2. With regards to the mediators, since no direct relationship between poverty and peer and school externalizing behaviors at Time 2 was found, no further tests for mediation were conducted.

Method: The study used a sample of 4,287 children and youth, ages 0 to 14, from the five waves of the National Longitudinal Survey of Youth (2000, 2002, 2004, 2006, and 2008). Dependent variable is the number of times mothers spanked their children in the past week. Independent variables in the model include the following: Children's scores from the Behavior Problems Index; family poverty; mother's health and marital status; community resources (afterschool daycare, recreational centers); and maternal perceptions of neighborhood safety. Covariates include child and mother demographics. Models were estimated using hierarchal linear modeling (HLM) in STATA 11SE to trace the patterns nested inside individual children over time.

Results: Child's age, race, and BPI scores were significantly associated with corporal punishment over time. Younger children, African American, and Hispanic children, and those with behavior problems received more spankings (p<0.001). Poor mothers also appeared to spank their children more (p<0.001); this was the only maternal factor to achieve significance in the model. As for neighborhood variables, the use of community resources did not significantly decrease the number of spankings children received over time. However, mothers who reported their neighborhood is safe for raising children spanked their children less (p<0.05). Lastly, the likelihood-ratio tests indicate that HLM adds more explanatory power to the model compared to OLS (p<0.001). This lends support for accounting for the random variations in the intercept by the individual child, capturing the story that each child starts from a different point