Abstract:
The purpose of this study was to use classification trees and logistic regression to identify
subgroups of students more likely to be retained. The National Educational Longitudinal
Study of 1988 (NELS:88) was used to identify the sociodemographic, family background
and school related factors associated with grade retention. The sample size for this study
consisted of 10,140 students, 1,570 of which had been held back. The NELS data were
obtained from student questionnaires and surveys with the students’ parents, teachers,
and school administrators. In order to identify the predictors of students more likely to be
held back, models were built using classification trees and logistic regression. Overall,
the current study identified the predictive factors of grade retention. Moreover, this study
demonstrates the effectiveness of using classification trees in conjunction with stepwise
logistic regression in educational research.
