Abstract

The usefulness of brainstorming as a group problem solving procedure was assessed by comparing the effects on performance of brainstorming instructions and nonbrainstorming instructions in real groups. There were four independent groups of subjects, each group working on two problems, one under brainstorming instructions and another under nonbrainstorming instructions. The order of problems and instructions were varied from group to gioup. Sixty-four undergraduates worked in 16 groups of four. The study was essentially a replication of Parnes and Meadow (1959) except that (1) as a measure of quality, quality ratings of the best ideas rather than the number of good ideas were used, and (2) group means rather than the means of individual subjects were used as data points in order to provide for more appropriate error term in the analysis of variance. Results showed that more ideas were produced when working under brainstorming instructions relative to non-brainstorming instructions, but the quality of the best ideas produced by individuals was not significantly different between the two kinds of instructions. The results on total output of ideas are consistent with the findings of the earlier study, but the results on quality are not. Brainstorming, which was originated and first used by Alex F. Osborn in 1939, has been subjected to experimental tests for its efficacy for a number of years. The methods used to test the technique, which allegedly promote creative thinking, varied from investigator to investigator. Hoffman (1965) and Zagona, Willis, and MacKinnon (1966) who reviewed experimental studies in this area identified three different categories of Studies, namely (1) those studies which compare individual problem solving with group problem solving; (2) those studies which compare between different group problem solving situations, and (3) brainstorming in individuals. The basic features or rules of brainstroming, which consist of (1) ruling out criticism, (2) encouraging "free-wheeling,"(3) an emphasis on quantity rather than quality, and finally (4) encouraging combination of and improvement over others' ideas (See Taylor, Berry, 8c Block, 1958), presuppose interaction among group members, and therefore the brainstorming in individuals does not represent a valid example of brainstorming technique. The studies which compared individual problem solving with groups solving problems under the condition of brainstroming typically used one or more nominal groups which are contrasted with real groups working under brainstorming instructions (e.g., Taylor, Berry, &: Block, 1958; Dunnette, Campbell, & Jaastad, 1963; Parnes & Meadow, 1963). A consistent finding seems to be that individual problem solving is superior to group problem solving when both are done under brainstorming instructions. When the effect of brainstorming is assessed by comparing a nominal group and a real group working under brainstorming instructions, what is actually evaluated may be a function of the difference between individual work and group work rather than the effects of brainstorming itself since individual working condition as against group working condition is the dominant feature separating a nominal group and a real group. A more appropriate procedure for assessing the effects of brainstorming instructions would be to compare two real groups, one performing under brainstorming instructions and another under non-brain storming instructions. There are two studies which studied brainstorming using the procedure recommended here. Meadow, Parnes, and Reese (1959) found that the group working under brainstorming instructions produced more ideas and more of better ideas. In this particular study, all subjects had previously received a semester of training in "creative problem solving methods." In a second study (Parnes & Meadow, 1959), the subjects did not receive this training prior to the experiment, but the study essentially duplicated the earlier study in its major findings. The present study is basically a replication of the Parnes and Meadow study. The present study differs from the study by Parnes and Meadow in two respects. As a measure of quality of the products, Parnes and Meadow (1959) resorted to counting the number of "good" ideas, a good idea being defined as an idea which is up to or exceeds a given standard. As will be developed in greater detail later, the present writers are of the opinion that a more adequate measure of the quality is not the number of good ideas but the goodness of the very best or of a select few best ideas, for in practical situations only the best idea or ideas count and less good ideas, no matter how good they may be, are by and large ignored so long as there is a definitely better idea. The second difference has to do with statistical procedure. In the last mentioned two studies, the investigators used individual subject's score as the unit of analysis instead of using group mean as the unit. Such a statistical procedure is likely to underestimate the within-group variance and thus become a cause for a Type I error. In the present study, this oversight was corrected by using group means as data points.