Classroom skills composition and preschoolers’ academic and executive function outcomes

• In Event: School Readiness: Contributions of children's individual differences during interactions in classroom and home environments

Abstract

As preschoolers spend plenty of time with their peers during child-initiated play and group activities, the average and variability of classmates’ skills, i.e., the classroom skills composition, create various opportunities for children to learn from their peers and result in changes in the classroom learning environment. This study examined: 1) the extent to which classmates’ skills composition (both in terms of mean and variability) is associated with preschoolers’ gains in the areas of language, literacy, mathematics, and executive function (EF); and 2) the moderating role of individual children’s characteristics and family background. When considering these questions, we considered both within- and cross-domain associations.
We used two datasets: One sampled 1,498 children across 138 preschool classrooms from a large county in Virginia in 2016-2017; the other included 2,206 children across 264 Head Start classrooms from the 2014 cohort of the nationally-representative Head Start Family and Child Experiences Survey (FACES). Although FACES measured children’s language skills using the Peabody Picture Vocabulary Test, children’s academic (Woodcock-Johnson III) and EF (Pencil Tap Test) skills were assessed with the same measures in both datasets. We calculated classmates’ skills mean and skills variability in the fall by excluding children’s own scores and imputed fifty datasets to address missing data in both data. For the Virginia dataset, which was focused on one larger program, we estimated regression models clustering standard errors at classroom-level and adjusting for a rich set of covariates. For FACES, given the focus on many programs, we adopted program-level fixed-effects models clustering robust standard errors at PSU level.
For RQ1, the within- and cross-domain effect sizes of classmates’ academic skills and EF mean on children’s spring skills were larger in FACES than in the Virginia data (within-domain: |ES| = .004–.060 and .004–.045; cross-domain: |ES| = .003–.124 and .001–.036). However, only the within-domain association between classmates’ math mean and children’s math (ES = .060, p < .05) skills were significant. In terms of cross-domain associations, significant associations emerged between classmates’ language mean and children’s EF and between classmates’ literacy mean and children’s math skills (ES = .124, p < .001; ES = .070, p < .001). For the associations with classmates’ skill variability, largely null patterns emerged across both datasets (within-domain: |ES| = .011- .034 and .008- .039; cross-domain: |ES| = .002 - .044 and .001 - .048).
For RQ2, which considered the extent to which individual children’s initial skills, household income, maternal education, and home language moderated the above associations, only two out of fifty-two interactions were significant for both classmates’ skills mean and variability in the Virginia data. None of the moderation effects were significant in FACES. Given that we would expect as much by chance, these interactions were not interpreted.
Results from our study suggest that, above and beyond children’s own preschool-entry skill levels, their classmates’ skills have little consequence for their development of academic and EF skills. Taken together, continued attempts at replication are necessary to reconcile the inconsistencies in the extant literature on peer effects.

Authors

• Qingqing Yang, Ohio State University
  Presenting Author

• Arya Ansari, Ohio State University

• Kelly M Purtell, Ohio State University

• Robert C Pianta, University of Virginia - Charlottesville

• Jessica E Whittaker, University of Virginia - Charlottesville

• Virginia Vitiello, University of Virginia - Charlottesville