Lecture Notes: Week 13

How can groups make decisions? There are a number of different, commonly used decision rules or processes:

• Consensus: every member must agree on the decision
• Majority: more than half (or 2/3 or whatever) of the members must agree
• Dictator: the decision is made by a group leader, with advice from the group
• Delphi: group members make comments in a series of rounds, often anonymously (using surveys), until consensus is reached or a median judgment is sufficiently constrained.

Plous discusses Sniezek’s work, comparing 5 decision rules: consensus, dialetic (discuss potential biases), dictator (select a group member to make the decision), delphi, and collective (aggregate individual judgments without interaction). She found that the dictator technique was best for quantitative judgment, but the dictator chosen would often be biased toward the group’s collective judgment, and thus less accurate than potentially possible!

Many studies have examined minority group members, their impact on group decisions, and the impact of the majority on them. As an undergraduate, Alan participated in a typical such study. He was introduced to 3 other people, and each was taken into an isolated booth and given headphones. He was shown a color on a screen and asked to name it. Over the headphones, he heard 3 other people name the color "red", but it sure looked orange to him. At the moment of truth, what did he say? What would he have said were it not for hearing the others?

Fiorelli examines medical teams explicitly, and focuses on issues of power. It’s based on self-reports from team members, and suggests that in medical teams, the physician is the most powerful figure in the group and most influential in decision-making, even when the group process is democratic in nature. This may result from the importance of "expert power" -- perhaps physicians usually hold the most relevant knowledge for the decision.

Group discussion can often lead to better problem-solving than individuals. This process is sometimes called "truth wins" -- when one member of the group stumbles onto the right answer, it becomes apparent to the others (a variant of this, "truth supported wins", requires that more than one person discover the truth in order for the group to agree on it.)

Group decisions are rarely as good as the best member, but usually better than the average member. This suggests that groups have a sort of averaging process, along with some enhancement due to discussion, multiple perspectives, etc.

Groups behave in many ways like individuals; Plous discusses group versions of common attributional biases (e.g. attributing success to intrinsic factors and failure to situational factors, perceiving other groups as more homogeneous). Danny Kahneman used to talk about a vivid example of this in the context of a committee authoring a textbook.

Other biases are unique to groups. Members’ initial leanings are often intensified by group discussion. Groups are particularly prone to "groupthink", in which the group’s leanings rapidly become all-consuming and result in a sort of tunnel vision that prevents the group from seeing contradictory evidence. Groups are also highly subject to escalation of commitment in the face of sunk costs.

Finally, Jim Larsen in the UIC Psychology Dept has done considerable work on the issue of information flow in groups. He’s shown, for example, that when group members each possess some shared and some unshared information, the shared information tends to receive more attention in discussion and greater weight in decision making. This is obviously non-normative (why?).