Modeling Math Growth Trajectory—An Application of Conventional Growth Curve Model and Growth Mixture Model to ECLS K-5 Data

Abstract

To model students’ math growth trajectory, three conventional growth curve models and three growth mixture models are applied to the Early Childhood Longitudinal Study Kindergarten-Fifth grade (ECLS K-5) dataset in this study. The results of conventional growth curve model show gender differences on math IRT scores. When holding socio-economic status (SES) constant, gender differences reduced on the mean start IRT scores, growth rate, and acceleration rate. Growth mixture modeling applied to ECLS K-5 children reliably identified three classes of children based on their math growth trajectories. Growth mixture modeling results indicate that gender differences are different depending on different math development classes. After controlling for SES, growth mixture modeling results show that gender differences on the mean start IRT scores, linear growth rate, and quadratic growth rate reduced in all subpopulations. Growth mixture modeling result also show that after controlling for gender, the effects of SES on math development are different in different subpopulations.