Evaluating measurement of longitudinal education data using the Measurement Model of Derivatives

Abstract

The Measurement Model of Derivatives (MMOD; Estabrook, 2015) provides the opportunity to evaluate and refine measurement scales used in longitudinal studies to clarify their theoretical distinctions and relationship to academic achievement. We demonstrate this using three teacher-rated scales of child self-regulatory behavior obtained from the Early Childhood Longitudinal Study Kindergarten Class of 2010-11 (ECLS-K:2011; Tourangeau et al., 2019). Data-driven factor structures were generated using a training sample (N = 2821), then compared using the MMOD to the theoretical measurement structure on a holdout sample (N = 2822). Finally, to externally validate their utility, the best-fitting data-driven measurement structure was compared to the theoretical structure in their ability to predict academic achievement on a validation sample (N = 5643). We discuss theoretical implications for self-regulation, as well as the MMODs applicability to other educational data sets.

Keywords: ECLS-K:2011; Educational psychology; Longitudinal data analysis; Measurement model.