Abstract
Groups often focus their discussions on information that all members know at the outset. To test how robust the sampling advantage for shared information is, a meta-analysis was conducted. The analysis integrated findings from 20 publications (45 independent effects), in which information sharedness was manipulated. Groups discussed more shared than unshared information overall. However, the observed sampling advantage was smaller than expected. Groups attenuated the discussion bias in particular when they had to choose among a small number of decision alternatives and when they had less than 30 minutes discussion time. Moreover, groups performing a hidden-profile task tended to display a smaller discussion bias than groups performing tasks with equally attractive alternatives.
Acknowledgements
We would like to thank the Editor Mike Allen and two anonymous reviewers for their helpful comments on our manuscript.