Executive function measurement in urban schools: Exploring the links between EF skills and teacher ratings

• In Event: Poster Session 01
  In Poster Session: PS 01 Topic - Education, Schooling

Abstract

Children use executive functions to complete everyday activities; however, higher-order processes pose a difficult construct to measure, partially due to the wide range of skills encompassed by executive functioning (EF; e.g., inhibition, working memory and shifting), as well as the impact that diverse ecological contexts have on development and measurement of EF (Wallisch, Little, Dean, & Dunn, 2018).
The present study explores the utility of teacher evaluations of EF in relation to computerized performance-based EF tasks within the context of low-income, urban schools. To do this, the study uses pre-existing data from a project that primarily explored the malleability of EF in low-income, urban school districts in the United States (Malleability of Executive Control, 2016; R305A110932). The sample consists of ethnic minority students from high poverty backgrounds (n = 232; mean age in years = 9.33, SD = 0.75; n = 252; mean age in years = 10.09, SD = 0.87).
In order to test the predictions of the links between the specific computerised tasks and the latent variables that they measure to the BASC-2 EF screener latent factors and the items that measure those factors, multi-group confirmatory factor analyses and structural equation modelling were conducted. The BASC-2 EF screener items embedded in the teacher rating scale were linked to a four-factor model established by Karr and Garcia-Barrera (2017). These factors- problem-solving, behavioral control, attentional control, and emotional control- were then linked to latent factors measured by computer-based EF Tasks.
The fit indices for Model 1 (Figure 1)demonstrate borderline good fit according to the standard criteria for CFA (Cheung & Rensvold, 2002; Hooper et al., 2008). When looking at the results at the parameter estimate level, it became apparent that there were not individual contributions of EF as originally predicted. The natural approach to correct the issue seemed to be combining the EF variables into one latent (Lee, Bull, & Ho, 2013; Nelson et al., 2016; Wiebe et al., 2011). In Model 2 (Figure 2), the EF variables were combined and demonstrated a better fit. All fit indices meet the outlined criteria, and the AIC and BIC show a decrease from Model 1 to Model 2. Model 2 demonstrated much stronger links between the unitary EF latent and the BASC-2 EF screener factors for the present sample.
From a theoretical perspective, the potential significance of this work goes beyond measurement. When compared to research centered on the EF development of white, middle-class children, relatively little is known about their non-white, lower-SES peers. A critical part of understanding whether what researchers claim is universal development of EF, actually is, includes developing reliable and valid metrics that can be used in schools. These findings open a small window in exploring the links between computerised EF tasks and observer reports of children within the context of low-income, urban schools. It would be useful for further research to look at whether these findings generalize to other computerised EF tasks and other observer reports of EF in children- from both parents and teachers alike.

Authors

• Anne Zonneveld, University of Cambridge
  Presenting Author

• Zewelanji N Serpell, Virginia Commonwealth University
  Non-Presenting Author

• Teresa Parr
  Non-Presenting Author

• Michelle Ellefson, University of Cambridge
  Non-Presenting Author