Biased Questions in Judgment of Covariation Studies

Abstract

An experiment was conducted to test the hypothesis that the phrasing of a question about the relationship between two events can influence what information subjects feel they, need to answer the question. Subjects were presented with one of two covariation problems and w ere asked a question that explicitly mentioned one type of instance or a second type of instance, or an unbiased question that mentioned all,four relevant types of instances. As predicted, subjects who were asked a biased question most often requested the frequency of instances mentioned in the question. Subjects who were asked an unbiased question most frequently requested positive confirming instances and requested significantly more information to answer the question. The relationship of this study to other studies demonstrating confirmatory hypothesis-testing strategies and implications for conducting research on intuitive judgments about relationships between events are discussed.