I am seeking information on the conditions under which face-to-face meetings can improve forecasting or decision making. So far, the evidence suggests that face-to-face meetings produce poorer decisions and  poorer forecasts than alternatives such as nominal groups, Delphi, virtual teams, or prediction markets. Can you provide sources for any evidence from comparative studies showing conditions under which face-to-face meetings have improved forecasting or decision making? A draft of my paper is at

http://marketing.wharton.upenn.edu/ideas/pdf/Armstrong/WisdomofCrowds2.pdf