This article concerns the effects of learner control in computer-assisted instruction (CAI). After reviewing previous reviews of research on the topic, twenty-four studies of learner control were subjected to meta-analysis. The results of both the review and meta-analysis are equivocal. Several reviews indicate that learner control works less well with younger, less able students. Other reviews indicate that, given optimal conditions, learner control can work with any students. The meta-analysis, however, yielded an average effect size that was small and negative suggesting that the average student would be slightly better off without it. Although learner control has theoretical appeal, its effects on learning seem neither powerful nor consistent.

This article critically examines the literature reviews of computer-assisted instruction (CAI). Its purpose is to synthesize what is known about CAI at all levels of implementation. Sixteen reviews, three traditional and thirteen quantitative, are compared and evaluated. Results of the synthesis indicated that typical effect of CAI is to raise outcome measures moderately by 0.42 standard deviation units. Thus, the effect of CAI is to place the average student using it at the 66th percentile of the control group distribution.

No abstract