Detection of Multiplicative Synergisms in Simulated Data for Nonorthogonal Designs: What Lies Beyond Linearity?
Abstract
A three stage simulation was conducted to evaluate the impact that multiplicative synergisms have within nonorthogonal (regression) designs. In the first stage, various designs with differing cue intercorrelations and number of stimulus cases were generated. In the second stage, several response strategies were simulated, including adding, multiplying, and adding-multiplying strategies. In the final stage, various research techniques, involving both descriptive and inferential analyses, were applied to analyze the simulated data. The results revealed that (1) while synergisms produce notable effects throughout the range of intercorrelation values, the most striking influence was found for negative intercorrelations, (2) descriptive indices (such as R2) were relatively unrevealing about the presence of synergisms, (3) in contrast, inferential indices (particularly the hierarchical test) were much more revealing, and (4) there were considerable between-stimulus difference in terms of how easily synergisms could be accurately detected. These findings imply that, because synergisms have been frequently overlooked by investigators, they should be specifically tested for in future research using nonorthogonal designs. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1981
- Accession Number
- ADA104004
Entities
People
- James Shanteau
Organizations
- University of Colorado Boulder