Socioeconomic Risk, Academic Engagement, and Community Characteristics: A Moderated Mediation Approach to Understanding High School Dropout

Background and Purpose: Failure to complete high school continues to be a problem in the United States. Although the extant literature has consistently identified a number of factors implicated in the decision to drop out of high school, studies testing empirical models of observable pathways to dropout over time are lacking. Using a life-course perspective within the context of Bronfenbrenner’s theory of human development, we test an empirically based model of high school dropout. Several factors spanning multiple ecological levels were included in the model to predict dropout status: socioeconomic risk, academic engagement (i.e., attendance, academic performance, behavior problems), and community characteristics (i.e., rural v. non-rural environment).

Methods: The data for this study were taken from the 1996–2012 Louisiana Department of Education administrative dataset (N=295,491). To be eligible for the study, students had to be old enough to drop out of school as of the last observation period. The primary predictor variable was socioeconomic risk, which is a composite score of the following socioeconomic factors: low income, male, African American race, and being overage. The mediating variables were the students’ average yearly school attendance level, experience of school behavior problems (number of suspension and expulsion episodes), and academic performance. The dependent variable was dropping out of high school. Path analyses were conducted to test the effects of socio-economic risk on dropping out of high school through the mediation of each type of school disengagement variable (attendance, suspension/expulsion, and academic performance), using Mplus 7.4. In addition, we tested for group invariance to compare the direct and indirect associations between rural and non-rural school districts.

Results: All modeled direct and indirect relationships examined were statistically significant at the .001, two-tailed level.  However, the comparison between rural and non-rural students (the moderated effect) demonstrates noteworthy variation in group differences.  These significant differences are found only for the following relationships: 1) the direct effect of socio-economic risk on attendance (t=9.191, p<.001); 2) the direct effect of socio-economic risk on academic performance (t=2,354, p<.05); 3) the direct effect of socio-economic risk on school dropout (t=5.957, p<001); 4) the direct effect of school behavior on school dropout (t=3.928, p<.001); 5) the indirect effect of socio-economic risk on attendance as this impacts school dropout (t=10.725, p<.001); and 6) the indirect effect of socio-economic risk on academic performance as this impacts school dropout (t=5.940, p<.001).  In each case, the effect size is larger for non-rural students over their rural counterparts, indicating that these relationships are more pronounced in non-rural areas of Louisiana than in rural areas.

Conclusions and Implications: The constellation of risk-factors impacting rural and non-rural students in this analysis were comparable. However, results suggest that risk-factors have a differential impact on students’ academic engagement and decision to drop out of high school based on rural or non-rural school location. Taken together with previous findings, when developing interventions and surveillance systems targeting at-risk students, certain risk-factors may be prioritized dependent upon community characteristics (i.e., urban, suburban, rural).