Information gathering on social targets: the trade-off between frequency and diagnosticity

Previous studies showed that prior expectancies bias information search about social targets towards asymmetric testing, wherein the “yes” answer and the “no” answer are not equally diagnostic. This tendency to asymmetry was interpreted in terms of confirmation bias. However, people might be more sensitive to the frequency of expected answers than to their diagnosticity. Given that weak diagnostic evidence is likely to occur whereas highly diagnostic information is rare, the present contribution aims at investigating the consequence of the asymmetric testing on the impression formation. In Study 1 (N=50) we analyzed the impact of yes and no answers to questions focused on personality traits of a social target. The pattern of multiple regression revealed that the questions level of asymmetry did not affect the final impression. In Study 2 (N=41) participants were presented with pools of questions -and relative answers- manipulated by symmetry and balanced in terms diagnosticity. The results showed asymmetric pools(extreme but rare information) to have lower impact on final impression than symmetric pools (less informative but frequent information). In Study 3 (N=60) participants were presented with evidence obtained from different members of a social category through symmetric or asymmetric questions. People were more prone to generalize the trait to the whole category in the symmetric condition. Possible consequences for the stereotype change process are discussed.