Measuring executive function in early childhood: A case for formative measurement.

Abstract

This study tested whether individual executive function (EF) tasks were better characterized as formative or reflective indicators of the latent construct of EF. EF data that were collected as part of the Family Life Project (FLP), a prospective longitudinal study of families who were recruited at the birth of a new child (N = 1,292), when children were 3, 4, and 5 years old. Vanishing tetrad tests were used to test the relative fit of models in which EF tasks were used as either formative or reflective indicators of the latent construct of EF in the prediction of intellectual ability (at Age 3), attention-deficit hyperactivity disorder symptoms (at Ages 3 to 5 years), and academic achievement (at kindergarten). Results consistently indicated that EF tasks were better represented as formative indicators of the latent construct of EF. Next, individual tasks were combined to form an overall measure of EF ability in ways generally consistent with formative (i.e., creating a composite mean score) and reflective (i.e., creating an EF factor score) measurement. The test-retest reliability and developmental trajectories of EF differed substantially, depending on which overall measure of EF ability was used. In general, the across-time stability of EF was markedly higher when represented as a factor score versus composite score. Results are discussed with respect to the ways in which the statistical representation of EF tasks can exert a large impact on inferences regarding the developmental causes, course, and consequences of EF.