Meta-analysis to estimate the relative effectiveness of TBLT programs: Are we there yet?

Abstract

Bryfonski and McKay published a meta-analytic review of reports on the effectiveness of task-based programs relative to other types of programs. The aggregated effect in support of task-based programs was substantial. However, Boers et al. re-examined the 27 comparative studies from which this effect size was calculated and argued that, with one exception, they did not serve the intended purpose of the meta-analysis well. However, in a recent publication, Xuan et al. argue that a good number of the primary studies used by Bryfonski and McKay are in fact suitable for a meta-analysis if the programs they assess are re-defined as task-supported rather than task-based programs. Xuan et al. then conducted a meta-analysis of 16 of the 27 comparative studies that were originally included in Bryfonski and McKay’s analysis, confirming the benefits of programs that use tasks, albeit with a smaller aggregated effect size. The present article revisits these 16 studies and, in addition, examines a handful more recent ones published since Bryfonski and McKay’s original meta-analysis. The conclusion remains that the field is not ripe yet for a meaningful meta-analysis of the relative effectiveness of either task-based or task-supported programs. Interpreting the outcomes of the meta-analytic endeavours reported so far is especially difficult because of a lack of clarity in both the primary study reports and the meta-analyses of what constitutes a task-supported program and of what is understood by ‘task’.