Longitudinal Predictors of Early and Late School Dropout in South African High School Students

Education is one of the strongest predictors of heath worldwide and school dropout is associated with both short and long term negative outcomes.  In South Africa, school dropout is common with 60% of first graders ultimately dropping out; by grade 12, only 52% of the population remains enrolled. Cross-sectional studies find dropouts more likely to engage in risk behaviors (e.g., substance use) but other contextual factors such as what youth do in their free time are ignored. Furthermore, few longitudinal studies focus on experiences of dropouts in developing countries. Therefore, the current study seeks to better understand a more comprehensive set of predictors of school dropout in a developing country.

The current study analyzed data from South African high school students participating in an effectiveness trial of HealthWise South Africa. Students enrolled in Cohort 1 (having the greatest number of waves) from both control and intervention schools were included in analyses (N=1,131). Data were collected every 6 months between 8^th and 11^th grade resulting in 8 waves of usable data. Dropout was measured in two ways using participation in the school-based survey as a proxy for school dropout. “Early dropouts” were those who provided at least 3 waves of data in Grades 8 and 9 but were not present for Grades 10 and 11 (n=333). “Late dropouts” included those who provided at least 3 waves of data in Grades 8 through 10 but were not present for Grade 11 (n=341). Both early and late dropout groups were mutually exclusive and were compared with “complete” individuals who were present for all Grades (8-11; n=457). 

Logistic regression analysis was performed using demographic, substance use, and free time experience variables to predict membership in early and late dropout groups. Results indicated early dropouts were more likely to have failed a grade (OR=4.1), less likely to have used alcohol (OR=0.4), more likely to have used tobacco in the past month (OR=9.8), and more likely to have increasing trajectories of leisure boredom (OR=2.4) and amotivation (OR=1.3) over time. Late dropouts were more likely to live with their father (OR=0.64) and less likely to have decreasing trajectories of leisure boredom (OR=0.6).  

Findings help to better understand temporal predictors of dropout and emphasize the need for a comprehensive approach when intervening with those students most at risk of dropping out. What youth do in their free time and how they experience that time has been linked with increased risk behavior but also with positive health and developmental outcomes. This study is the first to suggest that what youth do in their free time, along with other demographic factors, may be related to school dropout.