The performance of groups can be greatly influenced by their ability t
o identify their best member. This study investigated this issue in th
e context of a quantitative judgment task. Relative confidence judgmen
ts were used to determine the extent to which group members were able
to identify their best member. The effectiveness of groups was also co
mpared across three experimental conditions that varied in terms of th
e specific instructions group members were given about the group discu
ssion. Results indicate that group members were often able to identify
the most accurate member even though the judgment task did not have a
demonstrable solution. Groups were also at least as accurate as their
best member nearly 40% of the time. The group process intervention th
at encouraged group members to try to determine the most accurate indi
vidual judgment fared better than the intervention that focused on int
raindividual accuracy rankings. Groups in this condition also showed t
he strongest correspondence between identifying the best member and pe
rforming at this level. Groups assigned to an unstructured group discu
ssion condition performed nearly as well as those assigned to the more
successful of the two interventions.