Belief Perseverance, Biased Assimilation, and Covariation Detection: The Effects of Hypothetical Social Theories and New Data

Abstract

Initial beliefs about the relation between a person's risk preference and ability as a firefighter were manipulated using a hypothetical explanation task. Subjects then evaluated new data that displayed either a clearly positive or a clearly negative relation between the variables, presented in scatterplots. Final beliefs were assessed in several ways. The main findings were that general beliefs were influenced by new data but not by explanation, that specific predictions about a group of 7isk-preferring and a group of conservative firefighters were influenced by explanation but not by new data, that specific predictions about 1 00firefighter trainees were influenced by new data and by explanation, and that new data were evaluated in an unbiased fashion. Discussion focuses on the power of hypothetical explanation to produce belief perseverance, boundary conditions of biased assimilation, and the different judgment processes people use to generate answers to different types of belief questions.