A Systematic Review of Single-Case Research on Video Analysis as Professional Development for Special Educators

Abstract

Studies using video analysis are being reported more frequently in the literature. Although the body of research suggests that video analysis is effective for changing educators’ instructional practices, questions regarding for whom and under what circumstances it is most effective still remain. This meta-analysis reports on the overall effectiveness of video analysis when used with special educators, as well as on moderator analyses related to participant and instructional characteristics. Tau-U, a nonparametric effect size commonly used in single-case research, was used to aggregate the results from 191 AB phase contrasts across 12 dissertations and 18 peer-reviewed publications. A total of 111 participants across 69 single-case design experiments are reflected in the results. Overall, results indicate that video analysis is effective for changing the instructional practices of special educators and that it can be used across a variety of settings and with a diverse group of educators.

What’s in a Click? The Efficacy of Conditioned Reinforcement in Applied Animal Training: A Systematic Review and Meta-Analysis

“…There are several limitations which should be noted. First, dividing the dataset according to research designs was based on (a) the interest of analyzing studies aimed at changing behavior, which happened to use SCRMs, and (b) the lack of commonly accepted effect size indices that could be used to synthesize SCRM and group research [ 57 ]. Therefore, only a very small number of studies could be included in the meta-analysis, which makes the interpretation of results, especially the moderator analyses, tentative.…”

“…It was selected as the index for the current analysis because it has been found to be consistent with visual analysis of SCRM data, has the ability to control for an undesired baseline trend (i.e., trend in the direction of the intervention or confounding direction), is robust enough for small data sets, and was shown to have greater statistical power and precision compared to other nonoverlap effect size indices [ 55 , 56 ]. Tau-U is interpretable as a continuous index of improvement and the published benchmarks should be considered relative to participants’ needs, intervention comparisons, and settings: 0 to 0.62 = small effect; 0.63 to 0.92 = medium effect; 0.93 to 1.00 = large effect [ 56 , 57 ]. The Tau-U indices were computed by entering baseline and intervention data into a free online Tau-U calculator [ 58 ].…”

“…An effect size value was obtained for each AB contrast (i.e., baseline versus intervention). These experiment-level effect sizes were then combined into one omnibus effect size per study by applying an inverse weighting computation that puts more weight on studies with more data points and stability [ 57 ].…”

“…During the next step, effect sizes were entered into the Comprehensive Meta-Analysis ® software program (Version 3.3; [ 59 ]) to generate a summary effect size which represents all included studies. The summary effect size reflects a weighting computation that assigns relevance based on within-study and between-study variance [ 57 ].…”

“…Although neither a random nor fixed effects model is considered perfectly suitable for SCRM data, the assumed across-studies differences (e.g., subjects, outcome measures, procedures, and settings) made a random effects model the most appropriate [ 57 ]. In other words, the variance between studies was hypothesized to be due to systematic differences rather than sampling error, hence, implementation of a random effects model seemed to be appropriate for the current analysis [ 49 , 60 ].…”