Group Genius and Collective Intelligence October 28, 2010Posted by keithsawyer in New research.
Tags: Alex pentland, Anita Williams Woolley, Christopher F. Chabris, collective intelligence, Nada Hashmi, sociometric badges, Thomas W. Malone
A new study in Science magazine* provides additional evidence for group genius. My own research with collaborating groups has repeatedly demonstrated that groups manifest emergent properties, that are not reducible to the individual characteristics of the group members; this new study confirms my own findings, using a novel qualitative approach combined with “smart badges” designed by MIT’s Alex “Sandy” Pentland.
The researchers studied 699 people that were placed into groups of between 2 and 5 people. Then, they had the groups solve visual puzzles, engage in brainstorming, solve collective moral judgments, and negotiate over limited resources. From the group performance on these tasks, they derived a measure of “collective intelligence”. Using statistical methods, they found a common factor that accounted for more than 43% of the variance on all of the group tasks; they called this measure of group genius c, following the longstanding use of g to indicate “general intelligence” of an individual as measured by standard intelligence tests.
They also had each individual group member take a standard intelligence test, to identify the g score of each group member. And guess what? The average intelligence of all of a group’s members is not significantly correlated with collective intelligence! They also found that the maximum g score was not correlated with c.
When they combined the results of this first study with a second study, they found a moderate relationship between average g and maximum g and c. But c was still a much better predictor of a group’s performance than either average g or maximum g.
What I particularly like about this study is that they also looked at what factors caused a high c. Group cohesion did not; motivation did not; satisfaction did not. The factors that resulted in a high c were: the average social sensitivity of the group members; and the extent to which participation in the conversation was equally distributed across group members (they used the sociometric “smart” badges to measure this). (Some news stories have reported that the presence of women in the group increased its c score, but this effect was largely mediated by social sensitivity, because women score higher on social sensitivity. In other words, it’s not having a female per se that increases c; it’s having members with higher social sensitivity, and they could be male or female.)
This finding confirms the message of my research, as reported in my book Group Genius: effective creative groups display emergent properties that cannot be explained in terms of the aggregated properties of the individual members. I use a different methodology, interaction analysis, to identify the group processes associated with effective group genius; using that methodology, I also found that equal participation results in higher group performance. The key advance of this study is the development of a quantitative measure of “collective intelligence” that meets many of the criteria required of an effective psychometric (or sociometric) assessment. My own methodology of interaction analysis allows a richer explanation of what’s going on; I identified nine other properties of group interaction associated with group genius.
I look forward to reading more great studies from this research team!
*Evidence for a Collective Intelligence Factor in the Performance of Human Groups. Anita Williams Woolley,1,* Christopher F. Chabris,2,3 Alex Pentland,3,4 Nada Hashmi,3,5 Thomas W. Malone3,5. Originally published in Science Express on 30 September 2010; Science 29 October 2010: Vol. 330. no. 6004, pp. 686 – 688