Statistical Evaluation of an Experimental Design Examining Cognitive Dissonance
Abstract
In 2007, an experiment by Egan, Santos, and Bloom, which adapted Brehm's original free-choice paradigm design for non-verbal subjects, concluded that children and Capuchin monkeys experienced cognitive dissonance. However, Yale economist M. Keith Chen soon came forward to dispute the findings. Chen claimed that the design of the 2007 study was flawed; the findings could simply be attributed to ordering statistics rather than the effects of cognitive dissonance. In particular, the root of the fallacy lay in the researchers' belief that all objects receiving a particular rank on a discrete ranking scale were equally preferred by the subjects. Chen argued each subject actually had an ordered preference among equally-ranked objects. To test Chen's assertion, the researcher conducted an experiment similar to the Egan, Santos, and Bloom study. The subjects (N=130) ranked 20 different colors of M&M candies on a scale from 1 (lowest) to 5 (highest). After the subject ranked the colors, the researcher randomly selected four colors that received the same numbered rank and had the subject make decisions among several pairs from the four. The researcher recorded the rejection patterns to determine if they reflected the values expected from ordering statistics. Also, the subjects answered a qualitative three-question survey to give insight into the motivation behind their choices. Analysis of both the quantitative and qualitative data supported Chen's claims, indicating that the subjects did not experience cognitive dissonance and that they had an ordered preference among "equally-ranked" objects.
