The Journal of Applied Psychology just published a "Meta-Analysis" on the links between information sharing and team performance. This method entails using quantitative analysis to uncover patterns across large numbers of studies -- in this case, 72 studies of nearly 5000 groups. The overall findings aren't a surprise, that groups that engage in more information sharing enjoy better performance, cohesion, knowledge integration, and satisfaction with decisions made -- but given the best bosses are usually masters of the obvious first and foremost, this is useful reminder that getting people to feel safe enough and to have enough time to share their knowledge is worth the trouble.
There was another twist, however, that bosses ought to devote a bit more attention to, a pair of red flags. The first was that the more distributed the information (the more different places it is in, including spread among different people,locations and departments), the less sharing there was and, perhaps most troubling, the more heterogeneity there was (i.e., the more diversity there was on things ranging from gender, to race, to age, to professional background) the less likely they were to share information.
The upshot for people leading teams and in teams is that you've got to remember that in the situations when you need to share information most -- both in terms of avoiding pitfalls and reaching top performance --are times when there are strong impediments. To be more concrete, if information is spread around different people and places, and if people have different backgrounds, you really need to work in getting them to trust each other, take the others' perspective, and listen carefully to each other.
I see this in the best Stanford student teams. Last year, we did project for a major airline to increase the customer service experience at a large airline. The three person team was about as different as you could construct in a three person team (except all were under 35). An African American female MBA, A female Romanian doctoral student in chemistry, and a male Ph.D student in engineering. They went through a brief period of getting to know each other, but once they got past their differences, the speed at which they started talking, listening, and brainstorming, and combining knowledge was breathtaking -- and indeed, they came-up with implemented a prototype to improve the experience of retrieving luggage that was implemented a few months later. The worst groups, however, are composed of people who, say, act like the most stereotypical of MBAs, engineers, and designers, don't listen to each other and don't teach each other. On average, though, I am impressed that, if you create conditions were the differences are raised and people have time to work together on short projects to learn each others strengths and weaknesses, teams that have the most diverse members do the best work -- and to add a note of warning, the worst work too, because if they never get past their differences (as people open-up more slowly to others who are "different"), they are the worst.
So, having the information spread around a team and having diverse team members is a high magnitudes situation -- it brings out the best and worst in teams, and as a boss, part of your job is to make sure it brings out the best.
Finally, the weirdest twist is that groups that tended to TALK more, shared LESS information. My guess is that this happens because some groups are prone to having norms and a status game where the goal is steal the air time and display dominance, rather than to listen to each other, combine what was said, and act. This is especially true in occupations where people aren't paid to talk, but to do stuff. I recall a great old study of British string quartets where the best groups would talk and fight less, and play more. They were more effective partly because you can't fight as easily when you are playing and partly because disagreements about what to play are worked out best while testing what will work, not arguing over it.
P.S. I found this article described at BPS Research Digest. Here is the reference:Mesmer-Magnus, J., & DeChurch, L. (2009). Information sharing and team performance: A meta-analysis. Journal of Applied Psychology, 94 (2), 535-546.
Here is the study of quartets: Murnighan,
J. K. and Conlon, D. J. (1991). The dynamics of intense work groups:
A study of British string quartets. Administrative Science Quarterly,
36, 165-186
"The more heterogeneity there was ... the less likely they were to share information."
Any sales person would probably tell you that to succeed, you need to dress and talk like the group you are trying to sell to. If you are selling to geeks, they wouldn't trust you if you turned up in a suit and talked MBA-language. On the other hand, if you were selling something to a corporate management team, you would probably need to focus on cost-cutting and efficiency. It doesn't matter what you sell, development processes or CASE tools, if it's "lean production" you talk their language and they will trust you.
Posted by: Jan | April 16, 2009 at 10:27 AM
Bob, again an insightful post. Thank you. I gathered from the post that information sharing is directly proportional to team's performance.
How would one justify the ability of deaf children in Nicaragua designing their own gesture system to communicate among themselves?
Have a great day!
Posted by: Sanjay Basavaraju | April 14, 2009 at 09:35 AM
I really enjoyed this post (and I definitely enjoy the blog)!
I've been thinking about many of these kinds of issues with respect to academic biomedical science. Lots of lab group heads seem to foster intragroup competition, which I think is terrible.
I blogged a bit more about it here.
Posted by: Nat | April 14, 2009 at 07:32 AM
Hi Professor Sutton,
A fantastic study and observation indeed. I once read a study that had very similar findings and proposed that information sharing was related to a person's sense of "trusted identity." That is, if a heterogeneous team quickly formed a basis for collective identity - simple things like having a team name, having a team mission, etc - that this would increase information sharing. When such a collective identity didn't exist, team members looked to other signals for knowing who to trust (i.e. gender, race, function, geography, etc), and this is where things break down.
Another question I have that I don't have a lot of data on: what is the impact of having a team lead vs. just a team of peers? One can envision an argument for greater sharing in either scenario for different reasons. It would be interesting to see if this has been studied.
Posted by: Murthy | April 13, 2009 at 10:06 PM