This paper reports the results of a meta-analysis of studies of DI programs (excluding studies that only examined DI components) that included a comparison group, randomly assigned students to groups, and reported statistical information that could be used to calculate effect sizes. Twenty studies with 95 separate comparisons were identified. Slightly more than half of the studies (n=11) dealt with reading, 5 involved mathematics, 2 regarded language, and 2 regarded other areas. Slightly over half of the studies (n=11) involved general education students, while the rest involved students with disabilities. The ES index, Hedges’s g, was used. The average effect size over all 95 comparisons was .66, exceeding Fashola and Slavin’s level of “educationally importance” (.25) and Cohen’s “moderate” levels (.50). Effect sizes were slightly smaller, on average, for reading and the other programs (.56 and .54, respectively) than for language (.81) and mathematics (1.03. Effect sizes were very similar for studies involving general education (.69) and special education students (.71). Effect sizes were also regressed on methodological characteristics of the studies, including sample size, whether or not fidelity was monitored, the length of the intervention, the type of test used to measure achievement, and the year the study was conducted. The only statistically significant relationship was with sample size, with smaller effects for larger samples. Trends (p<.10) were also apparent for monitoring fidelity, with smaller effects when fidelity was monitored, and test type, with smaller effects for norm referenced than for criterion referenced tests.
Click here to view all documents.

