What Instruction Works for Students With Learning Disabilities? Summarizing the Results From a Meta-Analysis of Intervention Studies
H. Lee Swanson
University of California-Riverside
In this chapter, I review some of the empirical evidence collected from a comprehensive educational intervention research synthesis for students with learning disabilities ( Swanson, Hoskyn, Lee, & O'Shaughnessy, 1997). The synthesis used meta-analytic techniques to aggregate the research literature on intervention. Meta-analysis is a statistical reviewing technique that provides a quantitative summary of findings across an entire body of research ( Cooper & Hedges, 1994; Hedges & Olkin, 1985). The results of individual studies are converted to a standardized metric or effect size. These scores are then aggregated across the sample of studies to yield an overall estimate of effect size. Particular attention is given to the magnitude of the effect size estimate. According to Cohen ( 1988), .80 is considered a "large" effect size estimate, .50 a moderate estimate, and .20 a small estimate.
One of the major purposes of our ( Swanson et al., 1997) meta-analysis was to identify effective instructional models that yield high effect sizes, as well as the components that make up those models. As an extension of an earlier synthesis ( Swanson, Carson, & Sachse-Lee, 1996), we tested whether treatments that include components of strategy instruction (SI), direct instruction (DI), and/or both instructional models yield significant difference in effect size. Our earlier synthesis found that strategy and direct instruction models yielded higher effect size than competing models, but the instructional components that made up those models were not analyzed.
Because the terms strategy instruction and direct instruction are used in various ways by researchers and practitioners alike, further clarification