Academic engagement and achievement in students with disabilities: A systematic review and meta-analysis

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

Abstract

This meta-analysis aims to estimate overall associations among academic engagement and achievement in elementary students with disabilities (SWDs) and explore substantive and methodological predictors of individual study effect size. Implications for research, policy, and practice will be discussed relative to the strengths and limitations of meta-analysis.

Literature Review
Research highlights the importance of academic engagement in accessing instruction and long-term educational outcomes (Fredricks et al., 2004; Greenwood et al., 2002; Wang & Eccles, 2013). For this reason, engagement and on-task behavior are common proximal measures of intervention effectiveness (Al-Hendawi, 2012), especially among SWDs (e.g., Severini et al., 2018). The relation between challenging behavior and academic risk (see Chow & Wehby, 2018) highlights a need for interventions that maximize both access to instruction and eventual learning of academic content for students of all abilities. However, little is known about the relation between academic engagement and more distal measures of academic achievement in special education research. Thus, the purpose of this meta-analysis was to answer the following research questions: (1) what is the association between academic engagement and academic achievement in K-12 SWDs? and (2) what are potential moderators that contribute to heterogeneity within and between studies?.

Method
We searched databases for published and unpublished studies reporting the correlation between engagement and achievement. We propose to estimate an overall weighted mean effect size (r) and a series of meta-regressions via random-effects models, accounting for within-study effect size dependency using robust standard errors. Iterations of our preliminary search yielded 13,340 articles for abstract-level screening which has just been completed. We are coding sample-level (age, gender, race/ethnicity, disability category, ELL status), measure-level (e.g., direct observation, rating scales, rater), and content area (e.g., reading, mathematics) moderators of the overall weighted mean effect size.

We will use random-effects meta-analysis to synthesize eligible studies. Following main effects analysis, we will test for between-study heterogeneity and use meta-regression analyses to examine potential moderators. Additionally, we will conduct post-hoc tests to provide information about the robustness of our dataset and contextual information to inform the interpretation of our findings. To test for publication bias, we will triangulate results from visual analysis of funnel plots, trim and fill analyses (Duval & Tweedie, 2000), Egger’s regression (Eggers et al., 2001) tests to examine for potential bias as a function of effect size and study variance. Analyses will also include selection models (Hedges & Vevea, 2005), comparison of effects in published and unpublished studies, and a series of sensitivity tests to identify study- or design-level features related to the overall mean effect size.

Pre-Registered Hypotheses
We hypothesize that engagement and achievement measures are moderately correlated, suggesting that they are related but distinct constructs in the context of the present corpus of studies. Further, we anticipate significant variability in how engagement and achievement are measured across studies. We discuss potential study limitations and directions for future research.

Authors

• Esther Ritva Lindström, Lehigh University
  Presenting Author

• Jason C Chow, University of Maryland - College Park
  Non-Presenting Author