Predicting high school graduation rate using clustered multilevel modeling

The present research investigated the predictive variables at level-1 and level-2 that impact the prediction of high school graduation rate—the outcome variable—in the U.S. state of Tennessee. Multilevel models were fitted to 241 schools nested in 81 school districts to examine within-district school-level effects of student cohort population and student absenteeism, as well as contextual-district effects of staff count and district classification (urban or rural). Results show that within-district student cohort population and student absenteeism and contextual-district staff count, and district classification are good predictive variables of high school students' graduation rate.