Measuring the Growth of Social-Emotional Competence across Diverse Student Subgroups using Behavioral Rating Scales

• In Event: Poster Session 08
  In Poster Session: PS 08 Topic - Education, Schooling

Abstract

School-based, universal approaches to social and emotional learning (SEL) have been demonstrated to promote social-emotional competence (SEC) in the overall student body. Yet, it is unclear the extent to which disparities in SEC development exist across diverse socio-demographic subgroups of students and whether universal SEL reduces, maintains, or even exacerbates these disparities over time. To explore these questions, assessment tools first need to be scrutinized for measurement invariance across student subgroups.
This study examines (1) the factorial invariance of a widely-used, teacher-completed behavioral rating scale—the DESSA-Mini (Naglieri et al., 2014)—across student gender, race, and socioeconomic status (SES), and (2) the extent to which student SEC growth trajectories over two years differ by these student characteristics. Data used in this study come from a district-wide SEL initiative using Promoting Alternative THinking Strategies (Kusche & Greenberg, 1994). The analysis sample consists of 5,160 students across 15 schools (grades K-2 in Year 1; 48% female; 64% Hispanic/Latinx, 16% non-Hispanic Black, 14% non-Hispanic White; 86% low-income). Student SEC was repeatedly measured using three alternate forms of the 8-item DESSA-Mini (Fall/Winter/Spring) for two sequential years (2011-2012 to 2012-2013). Four levels of factorial invariance (i.e., configural, weak, strong, and strict) were tested using multi-group confirmatory factor analysis. Then, subgroup differences in SEC growth trajectories were examined using second-order latent growth modeling. Full Information Maximum Likelihood estimator with robust standard errors (MLR in Mplus) was used for missingness and non-normality.
A series of fit comparisons with increasing invariance suggest that at least strong factorial invariance (i.e., factor loading and intercept equivalence) can be assumed across gender, race, and SES, given inappreciable changes in practical fit indices (despite the significant chi-square differences that are sensitive to large sample sizes). This strong factorial invariance model showed an acceptable fit (see Table 1). With strong factorial invariance assumed, significant differences in the SEC latent means were observed over time: male (vs. female), Black or Hispanic/Latinx (vs. White), and low-SES (vs. middle-to-high-SES) students often showed lower SEC. When these characteristics were included as a set of predictors of the SEC growth trajectory, male (vs. female) and Black (vs. sample average) students had a lower baseline and a lower rate of growth; Hispanic/Latinx (vs. sample average) and low-SES (vs. middle-to-high-SES) students had a lower baseline but non-differential rate of growth; and White (vs. sample average) students had a higher baseline but a non-differential rate of growth (see Table 2).
Analysis with the DESSA-Mini raises the possibility that behavioral ratings scales can be used to assess growth of SEC across diverse socio-demographic subgroups, with evidence supporting that the same SEC construct is measured. Under an SEL implementation condition, gender, racial, and socioeconomic disparities in student SEC growth were observed, with a medium-sized gender gap and small-sized racial and socioeconomic gaps across time, and with a pattern of these gaps widening over time for male and Black students. Experimental studies are called for to understand how the benefits of universal SEL are distributed across diverse subgroups.

Authors

• Juyeon Lee, The University of Hong Kong
  Presenting Author

• Valerie Shapiro, University of California, Berkeley
  Non-Presenting Author

• Jennifer L Robitaille, Aperture Education
  Non-Presenting Author

• Paul LeBuffe
  Non-Presenting Author