Blinder–Morgan group decision experiments

The Blinder–Morgan group decision experiments were two controlled studies designed to test whether small groups could make sound decisions as quickly as individuals. Conducted in 2000 by Princeton economists Alan S. Blinder and John Morgan, they found that groups outperformed individuals on both a changing-conditions detection task and a simulated central-banking task. The experiments challenged the assumption that committees are necessarily slow or less capable than their strongest members, while highlighting a condition for collective intelligence: members’ judgments must be brought together in a real decision process.

Background and Research Question

Blinder’s doubts about committee decision-making grew from experience as vice chairman of the Federal Reserve Board in the mid-1990s. Interest-rate changes were set by the Federal Open Market Committee, a twelve-member body composed of the seven members of the Federal Reserve Board and five presidents of regional Federal Reserve banks. Committees were often criticized as inefficient, and Blinder wanted to test not only whether a group could make intelligent decisions, but whether it could do so quickly.

Blinder and Morgan built two experiments that crudely reproduced challenges faced by the Federal Reserve. Both asked whether discussion among participants could improve decisions under changing conditions, rather than merely adding time or allowing one especially capable person to determine the group’s answer. The studies therefore addressed both the quality and speed of group judgment, and whether collective performance depended on the smartest member.

Detecting a Change in Conditions

In the first experiment, students drew balls from urns that initially contained equal numbers of red and blue balls. They were told that sometime after the first ten draws, the proportions would shift: either 70 percent of the balls would become red and 30 percent blue, or the reverse. Their task was to identify as quickly as possible which color had become more prevalent. The setup represented, in simplified form, the challenge of recognizing that economic conditions had changed and deciding whether a policy response was needed.

The participants faced a cost for delay: they were penalized for every draw made after the changeover had occurred. They completed the task individually, then in groups with free discussion, then individually again, and finally once more as groups. This sequence was intended to control for learning. Under those conditions, group decisions were both faster and more accurate. Groups identified the direction of the change correctly 89 percent of the time, compared with 84 percent for individuals, and outperformed even the best individual.

Simulating Central-Bank Decisions

The second experiment placed students in the role of central bankers. They had to set interest rates in response to changes in inflation and unemployment, judging whether the economy was slowing or gaining momentum and whether rates should move accordingly. The task tested whether participants could recognize a shift in economic conditions and choose the appropriate direction for monetary policy.

Again, groups made better decisions than individuals, while acting just as quickly. Individuals moved interest rates in the wrong direction far more often. The results also undercut the possibility that the group was simply following its most capable member: the performance of the smartest person in a group did not correlate with the performance of that group. The collective result was not merely an individual expert’s judgment passed along to others; in the experiment, the group as a whole performed better than its smartest participant.

A Bank of England study modeled on the Blinder–Morgan experiment reached identical conclusions: groups could make intelligent decisions quickly and could outperform their smartest members. The replication strengthened the experiments’ central finding beyond the original student exercises, although the book describes the first study’s simulations as crude approximations of the Federal Reserve’s actual challenges.

Implications for Collective Decisions

The experiments form part of a wider argument about the conditions that make a group’s judgment useful. A wise crowd depends on diversity of opinion, independence of judgment, decentralization that lets people draw on specialized or local knowledge, and aggregation that combines private assessments into a collective result. The Blinder–Morgan studies show how group discussion can improve performance in small groups, where accuracy cannot simply be assumed to arise from a large crowd canceling out individual errors.

Their findings also bear on the familiar claim that committees are inherently inefficient. In these experiments, groups were not slower than individuals, and deliberation was associated with better answers. This does not establish that discussion is always beneficial or that every committee will perform well; it demonstrates that well-functioning group deliberation can produce gains in both speed and accuracy on the tested tasks.

The experiments also distinguish a decision-making group from a team used only to advise someone else. A small group has little advantage if its members’ opinions are never actually aggregated or if it is not allowed to make a decision. The broader lesson is institutional as much as psychological: including a team in a leadership structure does not by itself capture collective wisdom. The group needs a mechanism through which its members’ judgments become a decision.