Collective Information Sampling

Collective Information Sampling Bias

The CIS bias toward discussing shared information is relatively counterintuitive. After all, the assumption is that decision-making groups are formed in order to disseminate information to group members that they may otherwise never discover on their own; groups are then expected to use the collective knowledge of members to make a decision that is better informed than that of any one individual. However, in 1985 Stasser and Titus uncovered the tendency for group discussions to favor shared information over unshared information to a much greater degree than is predicted by simple probabilities. In a group setting, the probability that a certain piece of information will be discussed, p(D), is based on the following model: p(D) = 1 − [1 − p(R)]n. This model relates the probability that a member will recall and contribute an item of information (p(R)) to the number of members who can potentially recall the item (n). This model shows, through basic algebraic computations, that information that is held by more than one member has a significantly higher probability of being mentioned in a group conversation or task. However, actual groups perform much differently than predicted by the model, such that shared information is favored in discussions to a much greater degree than simple probabilities suggest. Unexpectedly, decision-making groups prefer to talk about information that everyone in the group already knows, and they do so significantly more than group organizers likely hope. Essentially, members do not disseminate much new information but instead prefer to discuss information that is likely already known by other group members.

As a result of this lack of new information obtained during discussion, decisional outcomes are often biased toward the prediscussion preferences of individual group members; this bias toward prediscussion preferences is typically found even when the pooled information strongly favors a different decision alternative. Put simply, people often go into groups with a slight bias toward a decision, and other group members do not reveal enough new information to sway that decision; as such, groups are not able to make decisions any better than any one individual within the group.

...