The structure of executive function in early childhood: Evidence from individual and group-based assessments and teacher reports.

• In Event: 1-053 - An integrative approach to understanding executive function development in young children.

Integrative Statement

Much of our understanding of EF development is fragmented across academic disciplines (education, developmental psychology, cognitive science, etc.), each with its own measurement traditions, techniques, and theoretical perspectives (Morrison & Grammer, 2016). To date, there have been few attempts at integrating approaches across disciplines to address fundamental questions relating to the characterization and development of EF. At present, the dominant conceptualization is that EF consists of a unitary set of skills in early childhood, that differentiate into three separate, but related constructs during the transition into late childhood (Zelazo, Blair, & Willoughby, 2016). However, virtually all the studies supporting this framework have relied on a similar set of cognitive tasks, all administered in a highly controlled laboratory setting.
While the use of these methods has provided insight into the development of EF, laboratory-based tasks administered to individual children leave out contextual factors that might influence the ways in which children employ EF in classroom contexts. EF skills often play out in group settings, with natural distractors and social interactions with peers and teachers. The lack of contextually sensitive EF measures can potentially hinder our understanding of how EF manifests in a naturalistic setting and has implications for EF research.
Therefore, the aims of this study are to examine the structure and organization of early EF skills using commonly used cognitive tasks, teacher reports, and a new set of ecologically sensitive group-based assessments. We also examine the degree to which EF assessments across measurement context relate to one another.
Participants were 195 kindergarteners (104 boys, 91 girls) that were assessed individually, using three cognitive tasks (Head-toes-knees-shoulders, Digit Span-Backwards, Pair Cancellation), and in a group context, using newly developed EF tasks that mimic classroom behaviors (Ahmed, & Morrison, 2018). Teacher-reported EF was measured using 19 items from the CBQ (Rothbart, 2001).

We used hierarchical confirmatory factor analysis, using maximum likelihood estimation to examine the dimensionality of EF across levels of analysis. A one factor solution exhibited the best model fit for the individually-assessed cognitive tasks, and 3-factor solutions of working memory, response inhibition, and attention control emerged as the best fitting models for group based and teacher reports of EF (χ2 (197) = 318.8, p=.000; CFI=.92; TLI=.91; RMSEA=.05). Each standardized factor loading for each latent factor was significant and exceeded a cutoff value of 0.40 (Stevens, 2001). We also observed significant, and positive relations among individual, group, and teacher-report latent EF measures (see Figure 1.).

Results confirm that the structure of early EF varies with measurement type. Specifically, we replicated the unitary model of early EF using individually-assessed cognitive tasks. However, componential models best represented group-based and teacher-reported EF. Given that these tasks provide information about individual EF components, and were designed with minimal overlap across components, they have the potential to be more sensitive to early differentiation than are traditionally used EF tasks. Further, the constraints of highly controlled, laboratory EF tasks potentially limits our understanding of the real-world manifestations of these important early skills.

Authors

• Sammy Ahmed, University of Michigan
  Presenting Author

• Frederick J Morrison, University of Michigan
  Non-Presenting Author