Donald Lehmann

Other Multivariate Techniques: Meta-Analysis

Abstract:
The author is asked to consider the following: For a meta-analysis, how should one decide if the same independent variable was used in all studies? When one is studying the effect of a drug, it is relatively easy. However, in consumer experiments, all the experimenters may state that they manipulated the same construct, but they actually manipulated the construct in two different ways, and the type of manipulation produces differences in the results. Also, experimenters may state that they manipulated different constructs, but the manipulations may be quite similar, and the effects of the manipulations of the two constructs may also be quite similar. Finally, consider the case in which one has two distinct constructs that could be aspects of a meta-construct and two distinct manipulations of each of these constructs. When would it be reasonable to aggregate the two sets of studies and claim that one is interested in examining the effects of the meta-construct?

Close Window