The Effects of Sample Size on the Estimation of Regression Mixture Models

• In Event: Roundtable Session 60
  In Roundtable Session: A Mixture of Papers on Heterogeneity

Abstract

Studies of regression mixture models and methodology have increased in recent years. These models maybe useful in an educational setting because groups of individuals are identified based, in part, on having a similar pattern of relationships between variables, as opposed to grouping individuals based on common traits of a single variable. For example, educational researchers may group children based on a relationship among test scores or attitudes toward a subject during an intervention. It is important that we understand their statistical strengths and weakness, in order to provide practitioners with guidance for using these models. This paper will look at the effect that total sample size and class size has on parameter recovery and model selection.

Authors

• Jessalyn Smith, CTB/McGraw-Hill LLC

• M. Lee Van Horn, University of South Carolina

• Litong Zhang, CTB/McGraw-Hill LLC