What If Consumer Experiments Impact Variances as Well as Means? Response Variability as a Behavioral Phenomenon.

This essay discusses the fact that in many cases means and variances of probability distributions are confounded and cannot be separated, which poses particular problems for analysis and interpretation of behavioral response data in consumer research. The discussion focuses on random utility theory to show that the confoundment of means and variances poses problems for published research in many areas of consumer behavior. Examples are provided to show why and how reported empirical differences in means may in fact be differences in variability. Ways to deal with the problem empirically are discussed. Copyright 2001 by the University of Chicago.