Brainstorming groups in context: effectiveness in a product design firm

Administrative Science Quarterly, Dec, 1996 by Robert I. Sutton, Andrew Hargadon

Osborn's (1957) book Applied Imagination fueled the spread of group brainstorming as a tool for increasing creativity in organizations. He proposed four rules for these sessions: don't criticize, quantity is wanted, combine and improve suggested ideas, and say all ideas that come to mind, no matter how wild. Much research has since been conducted on brainstorming effectiveness. Psychological Abstracts indicates that brainstorming was considered in 208 articles between 1967 and 1994. Findings from numerous experiments, however, have led researchers to conclude that face-to-face brainstorming is ineffective. We propose that this conclusion is suspect because it is based largely on a single effectiveness outcome from studies that do not examine how and why organizations use brainstorming. We draw on an ethnography of an organization that uses face-to-face brainstorming to develop a broader view of brainstorming effectiveness.

Much of the brainstorming literature, especially experiments, focuses on two accepted facts. On the one hand, face-to-face brainstorming is portrayed as a popular tool for doing creative tasks in organizations like developing products, overhauling business systems, and improving manufacturing. On the other hand, experiments show that, compared to when people generate ideas alone and output is aggregated to the group level (called "nominal groups"), face-to-face groups generate fewer nonoverlapping ideas per person. These articles often suggest that such lower productivity is surprising given the widespread faith in brainstorming and given Osborn's (1957) claim that, compared with working alone, the average person in a brainstorming group could generate twice as many ideas.(1)

Recent reviews conclude that this "productivity loss" is observed consistently in brainstorming groups with more than two members (e.g., Mullen, Johnson, and Salas, 1991; Stroebe and Diehl, 1994; Paulus, Brown, and Ortega, 1996). In Mullen, Johnson, and Salas's (1991) meta-analysis, they concluded further that nominal groups generate higher quality ideas than face-to-face groups. Diehl and Stroebe (1987) showed, however, that there is little evidence across brainstorming studies that face-to-face groups generate better or worse ideas. Fewer studies examine quality than quantity. Those that do so typically use quality measures confounded with quantity of ideas, and we were unable to uncover any study that has found significant differences in the mean quality of ideas generated by the two kinds of groups. Diehl and Stroebe concluded that face-to-face brainstorming produces fewer ideas and fewer good ideas (outcomes correlated at about .8) but, beyond this evidence of productivity loss, research on quality is otherwise equivocal.

Much research has considered why productivity loss occurs. Diehl and Stroebe (1987) proposed that three factors may explain this loss: (1) evaluation apprehension (group members may not express some ideas because they worry about what others think); (2) free-riding or social loafing (compared with working alone, individuals in groups do not feel as accountable for producing ideas, so they devote less effort); and (3) production blocking (compared with working alone, individuals' idea generation is "blocked" while they wait their turn to talk, and listening to others hampers thinking).

Evaluation apprehension has received some support: Productivity loss is greater when an authority figure is present (Mullen, Johnson, and Salas, 1991) and when some or all group members are disposed toward high anxiety about social interaction (Camacho and Paulus, 1995). Little support has been found for free-riding or social loafing. The strongest support has been found for production blocking, especially in Diehl and Stroebe's series of experiments (for reviews, see Stroebe and Diehl, 1994; Paulus, Brown, and Ortega, 1996). Paulus and Dzindolet (1993) found that productivity loss may also be explained partly by social comparison processes. Members of face-to-face brainstorming groups tended to match their productivity to members who generated the fewest ideas, resulting in lower productivity than nominal groups.

Related research uses the insight that people are more efficient when they work alone to develop (presumably) better idea-generating procedures. The Delphi (Dalkey, 1968) and nominal groups techniques (Van de Ven and Delbeq, 1974) entail first generating ideas alone, then using several rounds of interaction to refine and evaluate ideas. Recent research focuses on electronic brainstorming, in which people type solutions into a computer that has a second "window" of ideas generated by others who are simultaneously working on the problem. Advocates suggest that this method allows people to build on and be inspired by others' ideas without the blocking that occurs in face-to-face interaction. Groups that use electronic brainstorming produce more ideas per person than face-to-face groups and do not appear to suffer from individual productivity loss when compared with nominal groups (e.g., Gallupe et al., 1992; Valacich, Dennis, and Connolly, 1994).