Date of Completion

12-1-2015

Embargo Period

12-1-2015

Keywords

growth mixture modeling, residual variance, heterogeneity, behavior typology, noninvariance

Major Advisor

D. Betsy McCoach

Associate Advisor

Sandra M. Chafouleas

Associate Advisor

H. Jane Rogers

Associate Advisor

Noel Card

Associate Advisor

Faith G. Miller

Field of Study

Educational Psychology

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

This study investigates student classroom behavior changes over one year using multilevel growth mixture modeling to demonstrate how modifying assumptions of invariance affects parameter estimates, number of classes, and proportion of students assigned to each class. Current best practices for growth mixture modeling emphasize the importance of the proper specification, but the impact of these assumptions on the parameters and latent class composition has not been thoroughly addressed in applied research in multilevel growth mixture models. Using the Direct Behavior Rating Single Item Scale measures from 1975 students in lower elementary, upper elementary and middle school, a series of models were compared from full invariance to partial noninvariance. This research provides a description of steps, decisions, and results from testing for noninvariance, and how these affect the resulting subgroups and model parameters. Results indicated a dramatic shift in the students from higher classes to lower classes as the model was relaxed to allow for class level difference in variance parameters. The best fitting models for each grade group contained three latent classes characterized by students with consistently good classroom behavior, students with less consistent moderate behavior, and students with highly variable behavior. Criterion measures provided validation of these results. Research on classroom behavior heterogeneity using GM modeling represents an important addition to the knowledge base by providing a descriptive typology of student behavior, lending information necessary to fine tune behavior interventions. This research suggests that using variability as a criteria for the typology results is a more sensitive screening instrument identifying more students for follow up. This research also suggests that because student classroom behavior is highly variable for the students most at risk due to behavior problems, variability should be measured and tracked in single case studies to identify interventions to reduce this variability.


COinS