A Meta-Analysis of the Effects of USB (Universal School-Based) SEL (Social Emotional Learning) Programs for Marginalized Students in the United States

• In Event: Contemporary Approaches to Evaluating Universal School-Based Social Emotional Learning Programs: Effectiveness for Whom and How?

Abstract

This study presents findings of a meta-analysis of universal school-based (USB) SEL intervention studies implemented in the United States (US) K-12 education settings. The synthesized results of the intervention studies in the US educational contexts can provide meaningful suggestions for successful SEL implementations in US schools. We reviewed the SEL intervention studies of K-12 education under twelve SEL-relevant outcomes: (a) skills, (b) attitudes/beliefs, (c) prosocial behaviors, (d) externalizing behaviors, (e) civic attitudes and behaviors, (f) peer relationships, (g) emotional distress, (h) school functioning, (i) disciplinary outcomes, (j) school climate and safety, (k) family relationships, and (l) physical health outcomes.
Meta-Analysis Sample and Analytic Models. Following the PRISMA review processes, we identified 90 USB SEL studies published from January 2008 through December 2020. In this study, we analyzed 602 effect sizes in across the twelve SEL outcome domains (k = 602, N = 90). We applied for the Correlated and Hierarchical Effects (CHE) model and embedded the Robust Variance Estimation (RVE) method as suggested by Pustejovsky and Tipton (2022). We executed the analysis models using the R software, with the packages of "metafor", "clubSandwich", and "robumeta". Results.
Table 1 shows descriptive characteristic of 90 reviewed studies. Also, Table 2 shows the summary of students demographics in the US SEL intervention studies. Slightly more than half of the studies included students in Elementary School (K-5th grade) (N = 55, 62.5%) with 19 studies including Middle School (6th-8th grade) and 14 including High School (9th-12th grade). Student race/ethnicity was unspecified for 13 studies (19.70%). Within the remaining studies there was race/ethnicity was represented by American Indian or Alaskan Native (N = 13), Asian (N = 27), Black or African American (N = 58), Hispanic/Latino(a) (N = 50), Native Hawaiian or Asian/Pacific Islander (N = 7), more than one race (N = 10), white (N = 49). Student language status was only represented in 13 studies, and was indicated primarily by English Language Learner (N = 4), English Second Language (N = 3), and Limited English Proficiency, Non-Native First Language (N = 2 each). Among 90 SEL intervention studies, 33 studies (31.42%) reported on participants' demographic information related to disability issues.
The Q-value of the overall SEL outcomes was significant, indicating heterogeneity among the included SEL intervention effects (Q [601] = 8194.530, p < .001). First, the overall mean effect of the SEL program in pre-post tests was .193 and statistically significant between the intervention and control groups (k = 602, g = .193, 95CI = [.141, .244], p < .001). Table 3 shows the results of the USB SEL intervention effects between the pre and post-test by SEL subdomains. In the subgroup meta-analysis, we identified nine domains (i.e., SEL skills, Attitudes/beliefs, Prosocial behaviors, Externalizing behaviors, Peer relationship, Emotional distress, School functioning, School climate/safety, and Physical health) that met the recommendable number of samples (i.e., k min.= 10; Valentine et al., 2010). Additional moderators will be discussed in light of data set heterogeneity.

Authors

• Cheyeon Ha, Yale University

• Michael McCarthy, Yale Child Study Center

• Miranda Wood, University of Missouri

• Kaveri Sehgal, Northeastern University

• Eliya Ahmad, Yale University

• Christina Cipriano, Yale University