| This meta-analysis synthesized the findings from 23 published and five unpublished experimental or quasi-experimental group design studies on word problem-solving instruction for K-12 students with learning disabilities (LD) and mathematics difficulties (MD). A secondary purpose of this meta-analysis was to analyze the relation between treatment effectiveness and various study features including (a) participant characteristics, (b) study design characteristics, (c) outcome measure characteristics, and (d) contextual characteristics of instruction. Results of a random effects model synthesizing the 31 independent effect sizes extracted from the 28 included studies showed an overall mean effect size of 1.03 (SE = 0.15). Grade level of participants, type of report (published vs. unpublished), assignment to conditions, reliability of outcome measure, instructional setting, interventionist, instructional arrangement, mathematics task, and intervention duration were found to moderate treatment effectiveness. Given that effect sizes in two studies (Fuchs, Fuchs, Finelli, et al., 2004; Fuchs, Fuchs, Prentice, Hamlett, et al., 2004) were over four standard deviations above the mean, analyses performed without these two influential points produced a lower mean effect size (g = 0.77, SE = 0.10) and impacted the results of moderator analyses. Specifically, only two of the six variables found to moderate intervention effectiveness when all 31 independent effect sizes were included remained significant after removing the two influential effect sizes. Those two variables were reliability of outcome measure and intervention duration. The results extend the findings of previous meta-analysis with regard to the effectiveness of word problem solving interventions for students with LD and MD. Limitations and directions for future research are discussed. |
