In their study on brainstorming, Timothy Larey and Paul Paulus took a different approach to the subject. Instead of looking at the effectiveness of the technique as a whole, they looked at how different group dynamics impact the results of brainstorming. The study looks at differences between people who prefer working in groups or alone as well as actually brainstorming in groups or brainstorming alone.
The basic hypothesis of the study is that, while people who prefer working in groups will do better than people who prefer to work alone when working in a group, people who brainstormed alone will outperform people who worked in groups. A key point of this argument rests on the idea of “groupthink” that originated with Jarvis. The idea is that groups, in a desire to reach consensus quickly, will avoid divergent thinking and instead prefer to agree more quickly than would otherwise be the case. This convergent thinking would limit the performance of small groups in developmental and broad practices such as brainstorming, which demand a certain level of openness, creativity, and even divergence to be effective.
Using 144 volunteers from the University of Texas at Arlington, Larey and Paulus first tested for introverted or extraverted preferences, and then generated groups of four or individuals based on the results of the first test. This allowed for interactive groups and nominal groups of those who either preferred to work in groups or work alone. The experimenters then provided a brainstorming activity for all groups that discussed ways to improve the campus. Measurements were taken on the total number of ideas, the number of different categories discussed (categories being things like improving parking, classes, student activities, etc.), and how much detail was covered in each category.
The results confirmed the original hypothesis of Larey and Paulus by demonstrating that individual brainstorming activities were more effective than interactive small groups. While interactive groups made up of members that preferred working in groups did perform marginally better than groups of members that preferred to work alone, both fell far short of the nominal individual brainstorming groups. The average total number of ideas that interactive groups generated were 23.5 and 20.0 while those that nominal groups generated 45.6 for nominal high interactive preference and 53.2 for nominal low preference. Another key side hypothesis that was proven was that interactive groups discussed fewer categories in depth, while also not moving between categories. Rather, they preferred to spend more time discussing a few categories by repeating ideas. The nominal group, on the other hand, changed categories much more frequently and readily.
The ultimate conclusion of the study is that brainstorming is most effective at generating ideas when it is done by an individual. The authors believe that this is because people tend to prefer convergent thinking and can easily fall prey to “groupthink” mentalities when in small groups. This mentality discourages divergent or “free-wheeling” ideas and limits both creativity and real depth of discussion as people who interact would rather avoid conflict.
This study has some significance for the intelligence field. As brainstorming is a common step early in the modeling and research phases of intelligence, and most intelligence is done in small groups, it is important to know how these two factors interact. According to this study, a direct and interactive brainstorming session is both remarkably ineffective and a waste of the time of individuals. An alternate suggestion would be to have each group member individually brainstorm ideas for a given amount of time. At the end of this period, the group could come together and collectively present their own individual reports to a facilitator through a round-robin session where each idea is considered.
Larey, T.S., Paulus, P.B. (1999). Group preference and convergent tendencies in small groups: A content analysis of group brainstorming performance. Creativity Research Journal, 12 (3) 175-184.Retrieved from: