Controlling Variables: A Meta-Analysis of Training Studies

Abstract

Meta-analysis techniques were used to synthesize research on teaching students how to control variables (CV). Sixty-five controlled studies were included in the review. The mean effect size on posttest scores was .73. The main findings, not all of which were based on statistically significant differences in effect sizes, were: (a) Outcome measures that were demanding in task complexity, that focused on the specific elements of the CV schema, and that required performance across a range of context domains were more likely to reveal treatment impacts; (b) The effectiveness of treatments was enhanced by using motivational strategies based on cognitive conflict and appeals to student interest; (c) Strategies for introducing new learning were more effective if students were provided with explicit rules describing how to control variables; (d) Consolidation strategies had a greater impact when explicit attention was given to the elements of the CV schema in practice tasks and in feedback procedures; (e) Treatment impact increased with the range of contexts in which CV practice was provided. The results of the training studies are of high potential rather than actual use to teachers. Nonuse of the findings is attributed to an over reliance upon inadequate textbooks.