Group members tend to be biased in their evaluation of the information discussed. The present study aimed to disentangle the effects of preference consistency, social validation, and ownership on information evaluation in a single experimental design. Participants first received information about a personnel selection task. After having made a decision, they read a transcript of a fictitious discussion. In the transcript, preference consistency, social validation, and ownership of information were orthogonally manipulated as within-subjects factors. As hypothesized, preference consistency, social validation, and ownership all increased the perceived quality of information. Furthermore, participants intended to discuss a larger proportion of their preference-consistent information than of their preference-inconsistent information. This discussion bias was significantly associated with the evaluation bias favoring preference-consistent information. These results provide the first empirical demonstration that the evaluation of information in groups is characterized by three distinct biases and that biased evaluation of information may contribute to biased discussion of information. Copyright # 2009 John Wiley & Sons, Ltd.One of the most consistent findings in the group decision-making literature is that groups talk more about information that is shared and preference-consistent than about information that is unshared and preference-inconsistent (for reviews, see Brodbeck, Kerschreiter, Mojzisch, & Schulz-Hardt, 2007;Wittenbaum, Hollingshead, & Botero, 2004). Shared information is information that all group members are aware of prior to the discussion, whereas unshared information is held by only one member. Preference-consistent information refers to information that supports the group members' initial preferences, whereas preference-inconsistent information contradicts these preferences.Recent evidence suggests that group discussions are not only characterized by biased discussion but also by biased evaluation of information. Three types of evaluation biases have been proposed in the literature: The first one is the bias favoring preference-consistent information . Thus, group members are proposed to evaluate the information in favor of their initial preferences. Second, it has been argued that hearing that others possess the same information provides social validation of that information (Larson,