Using Longitudinal Structural Equation Modeling to Study the Development of Intelligence and Its Relation to Academic Achievement

Résumé du livre

In 2013, Drs Allen and Adele Gottfried shared the Fullerton longitudinal data with us. The data provided a unique opportunity to investigate the intellectual development and the longitudinal relation between intelligence and academic achievement. Previous studies have seldom addressed the latent nature of intelligence and academic achievement. Instead, they commonly used observed IQ scores and achievement test scores directly. Therefore, the stability of intelligence and the relation between intelligence and achievement were underestimated to some extent due to measurement error in the observed scores. Using structural equation models, in which both intelligence and academic achievement were measured as latent constructs (theoretically free of measurement error), we found that intelligence was very stable from infancy to adolescence. Furthermore, the effect of intelligence at earlier stages on the current intelligence was fully mediated by the intelligence at the adjacent preceding stage. Not surprisingly, intelligence was very predictive for students initial achievement at school. However, after controlling for the previous achievement, intelligence was not predictive of subsequent achievement. This case study guides readers through the whole process of conducting a data-driven research from preparing data to selecting appropriate methodologies, and then from interpreting significant results to reporting important findings. This case study demonstrates the potential value of reexamining classic findings using modern analytic techniques. This case study pays particular attention to making reasonable modifications to models, determining the best models, and interpreting the results to answer research questions.