Why so little faith? A reply to Blanton and Jaccard's (2006) skeptical view of testing pure multiplicative theories

Blanton and Jaccard (2006) questioned the 4-test regression method used by Greenwald et al. (2002) to test a pure multiplicative theory. The present authors address Blanton and Jaccard's concerns with a combination of simulations and meta-analysis. Simulations show that (a) Blanton and Jaccard's preferred simultaneous regression method has a severe power loss in testing multiplicative theories when predictor variables' means deviate from rational zero values, and (b) Greenwald et al.'s 4-test method has a more limited weakness when predictor means deviate from rational zero in the positive direction. Metaanalyses showed that aggregate analyses of Greenwald et al.'s 5 experiments confirmed a multiplicative theory regardless of which analysis method was used. However, only the 4-test method could confirm a pure multiplicative theory.