Ambiguity Aversion and Self-Evaluation

University Of Pittsburgh, Pittsburgh PA

Investigators

Abstract

People often face ambiguous information about their risks, skills, personality attributes, and opinions. Patients may need to cope with seemingly conflicting advice about the value of optional protective behaviors (e.g., prostate cancer screening), employees may receive a range of feedback about their job performance, and students may need to integrate an array of potentially discrepant information in order to make reasoned academic and career decisions. Research in the decision sciences suggests that people are often averse to the sorts of ambiguity described above. On the other hand, research in social psychology suggests that people try to use ambiguity to their advantage. They are more likely to believe they are superior to their peers on dimensions that can be defined many different ways (e.g., leadership ability) than on dimensions that are more unambiguously defined (e.g., punctuality). Moreover, they are more likely to see themselves as being superior relative to ambiguously defined referents (e.g., the "average" person) than relative to more unambiguously defined referents (e.g., a specific acquaintance). These separate lines of research suggest a need to better understand the role that ambiguity plays in self-judgment, reactions to personal feedback, and decision-making. This study examines these issues. The project consists of three experiments. Experiment 1 will present chance-based games and skill-based games with probabilities of winning that vary in ambiguity. It is predicted that people will prefer the skill-based games even when the probabilities of winning the chance games are ambiguous, and particularly when their chances of winning are contingent on ambiguous performance. The design of the experiment is such that this sort of pattern would reflect ambiguity aversion despite perceptions of high competence, and would therefore be consistent with both the decision science and self-evaluation literatures. Experiments 2 and 3 will test the hypothesis that people prefer games linked to performance on tasks for which they have received ambiguous feedback. In order to separate the confounding effects of personal competence and controllability, all experiments will include measures of how participants believe others should behave, and one will vary whether performance has occurred in the future or in the past (where controllability is low because past performance cannot be altered).