If you’ve read my 2007 book Group Genius, you know about the research showing that brainstorming groups perform worse than a comparable number of solitary individuals, working alone. Groups typically generate half as many ideas as the pooled ideas of the solitary individuals.
But most of what groups are asked to do in the real world is a lot different than simply generating lists of ideas. There are many studies showing that on more complex tasks, involving knowledge of conceptual systems, groups perform better than individuals. One study* I just read today compared solitary workers to groups of 2, 3, 4, and 5, on their ability to solve a simple codebreaking task: the individual or group was told that the first ten letters of the alphabet each corresponded to one of the digits, and they had to figure out the mapping by proposing mathematical equations to the experimenter (like A + B = ?) and the experimenter gave the answer in letters. Groups of five typically solved all ten letters in 6.83 guesses–which requires them to figure out that if they use multi-digit equations in a clever way, they solve the answer faster: EED + ECD + EFG = ? This was faster than the best of five comparison individuals. Groups of four and three were also faster than the fastest of a comparison nominal group. Statistically, there was no difference in the performance of groups of 3, 4, and 5.
The performance of groups of two was statistically identical to two people working alone–suggesting that you need at least three people to get the benefits of group dynamics, but adding more above three doesn’t give you an additional benefit–at least, for this particular task.
*Laughlin, P. R., Hatch, E. C., Silver, J. S., & Boh, L. (2006). Groups perform better than the best individuals on letters-to-numbers problems: Effects of group size. Journal of Personality and Social Psychology, 90(4), 644-651.