13
http://gpi.sagepub.com/ Relations Group Processes & Intergroup http://gpi.sagepub.com/content/7/3/195 The online version of this article can be found at: DOI: 10.1177/1368430204046107 2004 7: 195 Group Processes Intergroup Relations Bernard A. Nistad, Annelies E. M. van Vianen, Wolfgang Stroebe and Hein F. M. Lodewijkx Persistence in Brainstorming: Exploring Stop Rules in Same-Sex Groups Published by: http://www.sagepublications.com can be found at: Group Processes & Intergroup Relations Additional services and information for http://gpi.sagepub.com/cgi/alerts Email Alerts: http://gpi.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: http://gpi.sagepub.com/content/7/3/195.refs.html Citations: What is This? - Jul 1, 2004 Version of Record >> at UVA Universiteitsbibliotheek on May 17, 2014 gpi.sagepub.com Downloaded from at UVA Universiteitsbibliotheek on May 17, 2014 gpi.sagepub.com Downloaded from

Persistence in Brainstorming: Exploring Stop Rules in Same-Sex Groups

  • Upload
    rug

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

http://gpi.sagepub.com/Relations

Group Processes & Intergroup

http://gpi.sagepub.com/content/7/3/195The online version of this article can be found at:

 DOI: 10.1177/1368430204046107

2004 7: 195Group Processes Intergroup RelationsBernard A. Nistad, Annelies E. M. van Vianen, Wolfgang Stroebe and Hein F. M. Lodewijkx

Persistence in Brainstorming: Exploring Stop Rules in Same-Sex Groups  

Published by:

http://www.sagepublications.com

can be found at:Group Processes & Intergroup RelationsAdditional services and information for    

  http://gpi.sagepub.com/cgi/alertsEmail Alerts:

 

http://gpi.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://gpi.sagepub.com/content/7/3/195.refs.htmlCitations:  

What is This? 

- Jul 1, 2004Version of Record >>

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

Persistence inBrainstorming: Exploring Stop Rules in Same-Sex Groups

Bernard A. Nijstad and Annelies E. M. van VianenUniversity of Amsterdam

Wolfgang Stroebe and Hein F. M. LodewijkxUtrecht University

In a laboratory experiment, three-person interactive and three-person nominal groups ofcollege students brainstormed without externally imposed time constraints. All groups werehomogeneous with regard to gender. Half of the participants were instructed to continuebrainstorming until they ran out of ideas (expectancy stop rule), whereas the other half wereinstructed to continue until they were satisfied with their performance (satisfaction stop rule).We found that interactive groups were more persistent than nominal groups in both of the stoprule conditions and thereby compensated for their usual productivity loss. We also found, aspredicted, that women were more persistent in the satisfaction stop rule condition, whereasmen were more persistent in the expectancy stop rule condition. This effect may be due togender differences in self-evaluations.

keywords brainstorming, gender differences, persistence, stop rules

Group Processes &Intergroup Relations

2004 Vol 7(3) 195–206

BEFORE a group can solve a problem or makea decision, potential solutions or ideas have tobe generated. In the 1950s, Alex Osborn sug-gested brainstorming as a method to improvegroup idea generation (Osborn, 1957).During group brainstorming, people areinstructed to generate many ideas, to think ofuncommon ideas, to combine and improveideas, and to refrain from criticism. Brain-storming is an effective method of idea gener-ation relative to other procedures (Parnes &Meadow, 1959). However, research has con-sistently shown that nominal groups (people

who work alone and whose non-redundantideas are pooled) outperform interactivegroups of the same size by a large margin (e.g.Taylor, Berry, & Block, 1958). This productivityloss is found in brainstorming groups with

GPIR

Copyright © 2004 SAGE Publications(London, Thousand Oaks, CA and New Delhi)7:3; 195–206; DOI: 10.1177/1368430204046107

Author’s noteAddress correspondence to Bernard Nijstad,Department of Work and OrganizationalPsychology, University of Amsterdam,Roetersstraat 15, 1018 WB Amsterdam, The Netherlands [email: [email protected]]

01 046107 (bc/d) 30/6/04 2:26 pm Page 195

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

three or more members and increases withgroup size (for reviews, see Mullen, Johnson, &Salas, 1991; Stroebe & Diehl, 1994).

Diehl and Stroebe (1987, 1991) studied thecauses for productivity loss in brainstorminggroups and found strong support for theeffect of mutual production blocking.Blocking arises from the constraint that onlyone person in a group can speak at a giventime. As a result, group members cannotalways express ideas as they occur, but insteadhave to wait for their turn before they canspeak. Obviously, individuals working alone donot have this problem. Consistent with thisexplanation, Diehl and Stroebe (1987, 1991)found that when turn taking is imposed onnominal groups (people work in separaterooms, but must take turns expressing theirideas) a productivity loss similar to that foundin interactive groups occurs.

Paulus and Dzindolet (1993) offered anotherexplanation for the productivity loss in brain-storming groups. They argued that groupmembers compare and match their perform-ance with that of the least productive groupmember. These downward comparisons occurbecause there are typically no strong incentivesfor group members to perform well. Paulus andDzindolet (1993; see also Camacho & Paulus,1995) found some support for this argument.For example, they found that over time, groupmembers in interactive (but not nominal)groups converged around the rate of produc-tion of the least productive group member.

Although groups are less productive thanindividuals, most people are convinced thatgroup brainstorming is very effective, andgroup brainstorming is still widely used inorganizations of all kinds. Moreover, groupmembers enjoy brainstorming more than doindividuals and are more satisfied with theirperformance (e.g. Diehl & Stroebe, 1987).Researchers have found these beliefs in groupsuperiority so unrealistic that they have calledthem ‘the illusion of group effectivity’ (Paulus,Dzindolet, Poletes, & Camacho, 1993; Stroebe,Diehl, & Abakoumkin, 1992).

Recently, however, some evidence suggeststhat group brainstorming may not always be

ineffective. For example, when special pro-cedures are used to eliminate productionblocking, groups can outperform individuals.Studies of electronic brainstorming, in whichideas are exchanged through a computer inter-face, and studies of brainwriting, in which ideasare exchanged through written notes, haveshown that idea sharing can lead to productivitygains relative to nominal groups (see, forexample, Dennis & Valacich, 1993; Paulus &Yang, 2000). Apparently, when people can writeor type simultaneously, so that there is no pro-duction blocking, reading the ideas of otherscan be stimulating (also see Dugosh, Paulus,Roland, & Yang, 2000; Nijstad, Stroebe, &Lodewijkx, 2002). Production blocking over-rides these stimulation effects in face-to-faceinteraction.

Another reason for optimism about groupbrainstorming comes from two experiments byNijstad, Stroebe, and Lodewijkx (1999). In thetypical brainstorming experiment, equal timeresources are given to individuals and groups, toallow for fair ‘Equal Person-Hour-Comparisons’(Stroebe & Diehl, 1994). Nijstad and hiscolleagues (1999) removed all time constraintsand allowed participants themselves to decidewhen brainstorming sessions should end. Theyfound that groups spent more time at theirtask than did individuals, often brainstormingfor more than 30 or 40 min. Moreover, persis-tence increased with group size. This led to asubstantial reduction of the usual productivityloss for groups up to six persons. In fact,groups of four were no less productive thantheir nominal counterparts. Apparently, oneadvantage of group brainstorming is thatpeople keep each other going, leading tohigher levels of performance. It should benoted that these results did not occur becausepeople had a limited set of ideas, so that (dueto production blocking) groups needed moretime to express all their ideas. When asked tobrainstorm for an additional 5 min after theyended their sessions, both individuals andgroup members were quite capable of produc-ing more ideas. This suggests that participantswere not out of ideas, but rather stopped forother reasons.

Group Processes & Intergroup Relations 7(3)

196

01 046107 (bc/d) 30/6/04 2:26 pm Page 196

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

This article is aimed at extending thefindings of Nijstad et al. (1999). We examinedthe conditions under which brainstorminggroups are particularly persistent and produc-tive, and tested whether these conditions aresimilar for male and female groups.

Persistence and stop rules

Nijstad et al. (1999) reasoned that theirfindings arose from certain stop rules thatparticipants used to decide when to end theirtask. People may stop brainstorming for severalreasons, including boredom (enjoyment rule),satisfaction with their performance (satis-faction rule), or frustration about coming upwith new ideas and the belief that few suchideas will be found (expectancy rule).1 If weassume that there is a dominant stop rule, oneused by the majority of people (whether theywork alone or in groups), then we can derivepredictions about the persistence of brain-storming individuals and groups of differentsizes, depending on what that stop rule is.

Suppose that most people end a brainstorm-ing session when they no longer enjoy it. In thiscase, we would expect groups to be morepersistent than individuals, because membersof brainstorming groups usually enjoy them-selves more than do individuals brainstormingalone (e.g. Nijstad et al., 1999). When groupsare large, of course, they may be less enjoyable(see Levine & Moreland, 1990).

If satisfaction with performance is moreimportant, then a different prediction wouldbe made. Group members are often verysatisfied with their brainstorming performance,whereas individuals often are not. One reasonmay be that the total set of ideas produced by agroup can be impressive, and people may nottake group size into account when evaluatingtheir collective performance (Stroebe et al.,1992; see also Paulus et al., 1993, for otherreasons why group members are more satisfiedthan individuals). The total set of ideasproduced by a single individual might seemsmall by comparison. As a result, groupmembers may reach ‘satisfactory’ levels ofperformance relatively early in a brainstorming

session, and stop brainstorming before indi-viduals do (a similar argument was made byPaulus & Dzindolet, 1993).

Finally, if most people stop brainstormingwhen it becomes difficult to generateadditional ideas and their expectancies forfuture productivity are low, then we wouldexpect groups to be more persistent than indi-viduals. In groups, ideas often follow each otherat a quick pace and periods of silence areuncommon. This may create the impressionthat more ideas are possible, because it seemsrelatively easy to generate them. This wouldlead to high expectancies and thus persistence.Individuals who brainstorm alone must do allthe talking themselves, so they experiencemore difficulty, and there may be periods whenthey are unable to come up with ideas. Thismight create the impression that investingmore time and effort will not produceadditional ideas (low expectancy). Becauseexpectancies for future performance arehigher for groups than for individuals, groupsare expected to be more persistent than indi-viduals at brainstorming.

Several findings from the Nijstad et al. (1999)experiments suggest that most people use theexpectancy rule to decide when brainstormingshould stop. Their first experiment, whichshowed that groups are more persistent thanindividuals, contradicts the satisfaction rule,but is consistent with the expectancy and enjoy-ment rules. However, the results of their secondexperiment were inconsistent with the enjoy-ment rule and provided support for theexpectancy rule. In that experiment, task enjoy-ment was manipulated by using two differenttopics that varied in how stimulating theyseemed to participants. Although task enjoy-ment was significantly affected by topic, persis-tence at brainstorming was unaffected,suggesting that participants did not stop whenthey no longer enjoyed the task. Finally, whenasked to describe their reasons for endingbrainstorming sessions, most people (over 80%,whether they brainstormed in groups or alone)indicated that they were ‘running out of ideas’.This finding also indicates that the expectancyrule was dominant for most participants.

Nijstad et al. stop rules and persistence

197

01 046107 (bc/d) 30/6/04 2:26 pm Page 197

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

Our experiment

Our experiment is meant to extend the Nijstadet al. (1999) findings. In their research, partici-pants received no instructions about which stoprule to apply—they were simply told to stopwhen they felt it was a good time to stop. Underthose conditions, the expectancy stop ruleseemed to be dominant. One goal of ourexperiment was to compare group and indi-vidual persistence when participants wereexplicitly instructed which stop rule to apply. Inparticular, we were interested in the satisfactionand expectancy stop rules, because these rules(as noted earlier) were expected to havedifferent effects on groups and individuals. Wealso suspected that these stop rules would havedifferent effects on male and female groups.We thus performed an experiment in whichmale or female participants worked in eitherinteractive or nominal groups and receivedeither satisfaction or expectancy stop ruleinstructions. All groups were homogeneouswith regard to gender.

PersistenceWhen people are instructed to follow the satis-faction stop rule, they ask themselves duringbrainstorming ‘Am I satisfied now?’ and endthe session when the answer is ‘Yes’ (cf. L. LMartin, Ward, Achee, & Wyer, 1993). Due tothe baseline fallacy and other factors (seePaulus et al., 1993; Stroebe et al., 1992), groupmembers are usually more satisfied with theirperformance than are individuals. We there-fore expected group members to reach satis-factory levels of performance sooner thanindividuals. Consequently, groups should be lesspersistent than individuals in the satisfactionstop rule condition (hypothesis 1).

When people are instructed to follow theexpectancy stop rule condition, they ask them-selves during brainstorming ‘Can I stillgenerate ideas?’ and end the session when theanswer is ‘No’. Group members usually experi-ence less difficulty when generating ideas thando individuals. We therefore expected groupsto be more persistent than individuals in theexpectancy stop rule condition (hypothesis 2).

Nijstad and his colleagues (1999) found nogender effects when participants were not toldwhich stop rule to apply. There are, however,reasons to believe that a particular stop rulecould have different effects on male and femalegroups. Research on gender differences in self-evaluations has shown that women, relative tomen, tend to underestimate their abilities(Betsworth, 1999; Kimball & Gray, 1982), ratetheir task performance less favorably, and evenfeel entitled to less pay for the same amount ofwork (e.g. Jost, 1997). Moreover, they scorelower than men on such ‘core’ self-evaluations(see Judge, Locke, Durham, & Kluger, 1998) asgeneralized self-efficacy and self-esteem(Bergman & Scott, 2001; Cross, 2001; Kling,Hyde, Showers, & Buswell, 1999), locus ofcontrol, and emotional stability (Costa, Terrac-ciano, & McCrae, 2001; Lynn & Martin, 1997;Martin & Kirkcaldy, 1998).

These differences between men and womenmight interact with stop rules to affectpersistence during brainstorming. Under thesatisfaction stop rule, women’s lower self-evaluations might lead them to be dissatisfiedwith their performance and thus continuebrainstorming. Men, in contrast, might bemore satisfied with their performance, and thusend their brainstorming sooner. Consequently,we expected women to be more persistentthan men under the satisfaction stop rule(hypothesis 3). But under the expectancy stoprule, women’s lower self-evaluations might leadthem to be less confident about generatingmore ideas, whereas men might believe thatmore ideas are possible if they just brainstorma bit longer. Consequently, we expected men tobe more persistent than women under theexpectancy stop rule (hypothesis 4).

ProductivityNijstad and his colleagues (1999) showed thatthe usual differences in productivity betweeninteractive and nominal groups were reducedor even disappeared when people could brain-storm as long as they wanted, because inter-active groups were more persistent. Weexpected to replicate this finding in our experi-ment. However, as a logical consequence of

Group Processes & Intergroup Relations 7(3)

198

01 046107 (bc/d) 30/6/04 2:26 pm Page 198

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

hypotheses 1 and 2, an interaction betweentype of group and stop rule was expected. Wepredicted that groups would be less productivethan individuals in the satisfaction stop rulecondition (hypothesis 5), because groups in thatcondition would not compensate for their usualproductivity loss by being more persistent(H 1). However, groups ought to compensatefor their productivity loss in the expectancystop rule condition and thus be as productive asindividuals (hypothesis 6), because groups in thatcondition would be more persistent than indi-viduals (H 2). Given hypotheses 3 and 4, we alsoexpected women to be more productive thanmen in the satisfaction stop rule condition(hypothesis 7), whereas men were expected tobe more productive than women in theexpectancy stop rule condition (hypothesis 8).

Method

Participants, task, and designParticipants were 108 students (51 males, 57females) from Utrecht University who volun-teered to take part in the experiment and werepaid DFL 11.25 (approximately US$5) for theirparticipation. Participants generated ideas onthe topic of how education at Utrecht Uni-versity could be improved. They were randomlyassigned to conditions.

Participants brainstormed in either nominalthree-person groups or in interactive three-person groups. All groups were homogeneouswith regard to gender and contained studentswho were strangers to one another. Members ofnominal groups worked separately, whereasmembers of interactive groups worked togetherin one room. All participants were instructed tostop brainstorming when they ran out of ideas(expectancy rule), or when they were satisfiedwith their production (satisfaction rule). Thus,the design was a 2 (interactive vs. nominalgroups) � 2 (expectancy vs. satisfaction rule) �2 (gender) full factorial.

ProcedureThe experiment was carried out in fourdifferent rooms. One room was reserved for theexperimenter, and the others could be used to

run three parallel sessions. For each session,three male or three female participants eitherworked together as an interactive group orworked individually as a nominal group.

Upon arrival, participants were led to theexperimenter’s room, where instructions weregiven to them. They were told that we wereinterested in the brainstorming performanceof groups and individuals, and so some partici-pants would work in groups and others wouldwork individually (persistence was not men-tioned). Next, they were asked to turn in theirwatches in order to avoid feelings of timepressure. Participants were told that the experi-ment would take one hour and that it wouldconsist of two different tasks (see below). Thefour basic brainstorming rules (generate manyideas, generate ‘wild’ ideas, combine andimprove, and no criticism; Osborn, 1957) werethen described. When these were clear, partici-pants were told that they had to decide forthemselves when to stop brainstorming. Thenone of the stop rules was read to them.

Participants in the satisfaction condition weretold: ‘You should go on until you are satisfiedwith your performance. When you feel youhave generated a sufficient number of ideas,you may stop brainstorming. Please do not stopbefore you are satisfied with your production.’

In contrast, participants in the expectancy con-dition were told: ‘You should go on brainstorm-ing until you feel that you are running out ofideas. Thus, you stop brainstorming when youdo not expect to generate many new ideas ifyou go on. Please do not stop when you thinkthat going on will lead to new ideas.’

We explained to participants that the timethey spent brainstorming did not matter,because everybody would eventually spend onehour in the laboratory, and any time not spentbrainstorming would be spent at the secondtask. This was not true, but we said it to avoidcreating a demand characteristic for partici-pants. This procedure is similar to the one usedby Nijstad et al. (1999). Their research sug-gested that one hour should be sufficient,because no group or individual continuedbrainstorming for more than 50 min.

If participants had no questions, then they

Nijstad et al. stop rules and persistence

199

01 046107 (bc/d) 30/6/04 2:26 pm Page 199

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

were given the topic of their brainstormingsession and led to either one room (interactivegroups) or three rooms (nominal groups). Inevery room, there was one table, with a micro-phone on top of it. The microphone was con-nected to a recorder in the experimenter’sroom so that the brainstorming session could berecorded. There was also a small device with alight and a button that was connected to theexperimenter’s room. Participants were told tostart brainstorming when the light came on.When they wanted to stop, they should push thebutton, sending a signal to the experimenter. Atthat signal, the experimenter recorded the timethat participants had spent on the task.

After they finished, participants went to theexperimenter’s room, where they individuallycompleted a post-experimental questionnaire.Finally, the participants were paid for theirparticipation, debriefed, and dismissed.

Dependent variables and scoringThe two major dependent variables were persis-tence (how much time was spent brainstorming)and productivity (the number of non-redundantideas generated during brainstorming). Thequestionnaire contained three items thatmeasured on 9-point scales how much theparticipants enjoyed brainstorming (1 = not atall to 9 = very much), how many ideas theycould still generate after ending the brain-storming session (1 = none at all to 9 = verymany), and how satisfied they were with theirbrainstorming performance (1 = not satisfiedto 9 = very satisfied).

The tapes were transcribed and ideas weremarked on the transcriptions by a rater who wasblind to conditions and hypotheses. For arandom subset of 12 tapes, this procedure wasrepeated by a second independent rater. Theintraclass correlation of the two raters’ countswas .93 (p < .001). Next, ideas that were men-tioned more than once were removed fromthe counts, to measure the number of non-redundant ideas. This was done for both inter-active and nominal groups. For the nominalgroups, the lists of ideas generated by the threemembers were combined and duplicates were

removed. An independent rater repeated thiscoding task for 12 groups, with good reliability(Cohen’s kappa = .91; p < .001).

Results

Manipulation checks The ratings on the questionnaire items wereanalyzed using 2 (type of group) � 2 (stoprule) � 2 (gender) analyses of variance(ANOVAs). Participants who worked under thesatisfaction stop rule indeed reported moresatisfaction (M = 4.93) than participants whoworked under the expectancy stop rule (M =4.04) (F(1, 100) = 5.72, p < .05). The ANOVAalso revealed, as other researchers have found,that participants who worked together in inter-active groups (M = 5.04) were more satisfiedthan participants who worked alone in nominalgroups (M = 3.93) (F (1, 100) = 8.98, p < .01).It should be noted, however, that satisfactionwas not particularly high. The overall mean was4.48, significantly lower than the scale midpointof 5 (t(107) = –2.68, p < .01).

Expectancy ratings were lower in theexpectancy condition (M = 2.56) than in thesatisfaction condition (M = 3.07), althoughthe difference was not significant (F(1, 100) =2.63, p = .11). This may be due to a floor effect,because the overall mean was very low (M =2.82), significantly lower than the scalemidpoint of 5 (t(107) = 12.81, p < .001). Type ofgroup had no effect on expectancy ratings (F(1,100) = 1.58, ns). No further effects were foundin these ANOVAs. These results suggest thatour manipulation of stop rules was successful.

Finally, a 2 (type of group) � 2 (stop rule) �2 (gender) ANOVA was also performed on theenjoyment ratings. As other researchers havefound, participants who worked together ininteractive groups enjoyed themselves more (M= 6.89) than participants who worked alone innominal groups (M = 6.24) (F(1, 100) = 5.27,p < .05). No further effects were found in theANOVA. Enjoyment and satisfaction ratingswere positively correlated (r = .45, p < .001), butthe other correlations among the questionnaireitems were not significant (p > .10).

Group Processes & Intergroup Relations 7(3)

200

01 046107 (bc/d) 30/6/04 2:26 pm Page 200

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

PersistencePersistence and productivity data were analyzedat the group level (see Table 1). Persistence wasthe total number of person-hours spent at thetask, divided by the number of group members(three). For interactive groups, the totalnumber of person-hours equaled three timesthe group’s persistence, whereas for nominalgroups, it equaled the sum of the persistence ofeach individual in the group. Persistence wasanalyzed using a 2 (type of group: interactivevs. nominal) � 2 (stop rule: satisfaction vs.expectancy) � 2 (gender) ANOVA.

The analysis revealed a large main effect oftype of group (F(1, 28) = 41.46, p < .001, �2 =.60). Interactive groups spent about 31 minbrainstorming, whereas nominal groups spentonly 14 min brainstorming. The main effect ofstop rule was not significant (F(1, 28) = 1.17,p > .25, �2 = .04), nor was the main effect ofgender (F < 1, �2 = .01). The interactionbetween type of group and stop rule was signifi-cant (F(1, 28) = 7.03, p < .025, �2 = .20).According to hypothesis 1, individuals shouldhave been more persistent than groups in thesatisfaction stop rule condition. But contrary tothat prediction, interactive groups (M = 36.41)were much more persistent than nominalgroups (M = 12.45) in that condition (F(1, 14)= 55.96, p < .001, �2 = .80). According tohypothesis 2, groups should have been morepersistent than individuals in the expectancystop rule condition. This prediction was con-firmed: interactive groups (M = 26.27) were

more persistent than nominal groups (M =15.99) (F(1, 14) = 5.69, p < .05, �2 = .29).

Hypotheses 3 and 4 predicted an interactionbetween stop rule and gender in persistence.This interaction was significant (F(1, 28) =14.41, p < .01, �2 = .34). According to hypothe-sis 3, women should have been more persistentthan men in the satisfaction stop rule con-dition. This prediction was confirmed (F (1, 14)= 6.80, p < .025, �2 = .33). Women (M = 27.39)were indeed more persistent than men (M =21.47) in that condition. According to hypoth-esis 4, men should have been more persistentthan women in the expectancy stop rule con-dition. This prediction was confirmed as well(F(1, 14) = 7.75, p < .025, �2 = .36). Men (M =27.87) were indeed more persistent thanwomen (M = 15.74) in that condition. Theinteraction between type of group and genderwas not significant (F < 1, �2 = .03), nor was theinteraction among type of group, stop rule, andgender (F (1, 28) = 1.70, p > .20, �2 = .06).

ProductivityProductivity was the total number of non-redundant ideas produced by each group,whether its members worked together (interac-tive condition) or alone (nominal condition).Persistence and productivity were strongly cor-related. The correlation for the entire samplewas r = .53 (p < .001), and it was even largerwhen interactive groups (r = .69, p < .01) andnominal groups (r = .90, p < .001) were con-sidered separately. Not surprisingly, greater

Nijstad et al. stop rules and persistence

201

Table 1. Persistence and productivity by condition

Persistence (min) Productivity (No. of ideas)

Condition Male Female Male Female

Interactive groupsExpectancy rule 33.58 (15.05) 20.42 (9.15) 57.25 (29.62) 38.20 (17.37)Satisfaction rule 29.83 (8.86) 44.64 (9.44) 64.80 (33.83) 66.75 (41.03)

Nominal groupsExpectancy rule 22.15 (6.40) 11.06 (3.27) 83.75 (18.06) 50.00 (10.10)Satisfaction rule 11.03 (2.32) 13.59 (4.83) 45.00 (15.64) 54.60 (17.24)

Note: Standard deviations are in parentheses.

01 046107 (bc/d) 30/6/04 2:26 pm Page 201

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

persistence was always associated with moreproductivity.

A 2 (type of group) � 2 (stop rule) � 2(gender) ANOVA of productivity revealed nomain effects of type of group (F < 1, �2 = .00),stop rule (F < 1, �2 = .00), or gender (F(1, 28) =1.58, p > .20, �2 = .05). The lack of a main effectfor type of group shows that interactive groupscompensated for their usual productivity losswith greater persistence, as Nijstad and his col-leagues (1999) found.2 The interaction betweentype of group and stop rule was significant (F(1,28) = 4.58, p < .05, �2 = .14). The results weresimilar to those from the persistence data. Inthe satisfaction condition, interactive groups (M= 65.67) were more productive than nominalgroups (M = 50.33), although this differencewas not significant (F(1, 14) = 1.37, p > .25, �2 =.09). This does not support hypothesis 5, whichpredicted that nominal groups would be moreproductive than interactive groups in that con-dition. But note that hypothesis 5 was derivedfrom hypothesis 1, which was not supportedeither. In the expectancy stop rule condition,nominal groups (M = 65.00) were only margin-ally more productive than interactive groups (M= 46.67) (F(1, 14) = 4.37, p < .06, �2 = .24). Thissupports hypothesis 6, which predicted nodifference between interactive and nominalgroups in this condition (because interactivegroups are more persistent).

Hypotheses 7 and 8 suggested that thereshould be an interaction between stop rule andgender for productivity. This interaction wasmarginally significant (F(1, 28) = 3.84, p < .06,�2 = .12). Hypothesis 7 predicted that womenwould be more productive than men in the satis-faction condition. However, men (M = 56.00)and women (M = 60.00) were equally productivein this condition (F < 1, �2 = .01). Hypothesis 8predicted that men would be more productivethan women in the expectancy condition. Thisprediction was confirmed (F(1, 14) = 8.30, p <.025, �2 = .37). Men (M = 70.50) were indeedmore productive than women (M = 44.10) inthis condition. The interaction between type ofgroup and gender was not significant (F < 1, �2

= .00), nor was the interaction among type ofgroup, stop rule, and gender (F < 1, �2 = .02).

Discussion

Our aim was to extend the findings of Nijstadet al. (1999), who found that brainstorminggroups were more persistent than brainstorm-ing individuals, and that this allowed groups tocompensate for (part of) their usual produc-tivity loss. In particular, we studied the effectsof two different stop rules on the persistenceand productivity of same-sex interactive andnominal groups. Participants were eitherinstructed to stop when they were satisfied withtheir performance (satisfaction stop rule) orwhen they ran out of ideas (expectancy stoprule). We predicted two interaction effectsinvolving persistence and productivity, onebetween type of group and stop rule, and onebetween gender and stop rule. We will nowdiscuss these interactions in turn.

We predicted that interactive groups would be(prematurely) satisfied with their brainstormingperformance sooner and thus be less persistentthan nominal groups in the satisfaction rule con-dition. However, nominal groups should be lesspersistent than interactive groups in theexpectancy rule condition, because periods ofsilence are more prevalent for individuals thanfor groups.3 Interactive groups indeed weremore persistent than nominal groups in theexpectancy rule condition. But contrary to ourpredictions, interactive groups were also morepersistent than nominal groups in the satis-faction condition, and the difference was evenlarger than in the expectancy condition. As aconsequence, there was a main effect for type ofgroup—interactive groups were much morepersistent than nominal groups. This greaterpersistence completely (satisfaction condition)or partially (expectancy condition) compen-sated for the usual productivity loss experiencedby interactive groups. This was only partially con-sistent with our predictions, because weexpected that nominal groups, because ofgreater persistence, would be more productivethan interactive groups in the satisfaction con-dition. This did not happen.

Why were interactive groups more persistentthan nominal groups in the satisfaction con-dition? Perhaps because the satisfaction rule is

Group Processes & Intergroup Relations 7(3)

202

01 046107 (bc/d) 30/6/04 2:26 pm Page 202

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

more ambiguous than the expectancy stop rule.The expectancy rule is relatively unambiguous,because there are clear indicators that someoneis running out of ideas (e.g. periods of silence,repeating ideas). It also seems relatively easy todefend, either to oneself or to others (othergroup members or the experimenter), that onehas stopped because additional ideas can nolonger be generated. The instruction to go onuntil someone is satisfied with his or herperformance is more ambiguous, because it isnot clear how satisfied that person should be.We suspect that people start asking themselveswhether their performance is satisfactory onlyafter some time has passed, perhaps only afterthey seem to be running out of ideas. If it is stillrelatively easy to generate ideas, then whyshould people stop even if they are satisfied?Satisfaction becomes more of an issue as ideasbecome more difficult to generate. Thishappens sooner for individuals than for groupmembers, and so individuals may decide thatthey have done enough before group membersdo. Group members may also be reluctant totell each other that they are satisfied, becausethis suggests that they are not interested inperforming well. Again, this would lead togreater persistence of groups relative to indi-viduals in the satisfaction stop rule condition.

Informal observations of the discussionstoward the end of the group sessions were con-sistent with this analysis. All of the discussionsin the expectancy conditions were relativelystraightforward—someone said that he or shewas running out of ideas, and the othersagreed. In the satisfaction condition, however,participants often began to wonder whetherthey had enough ideas the moment they beganto experience difficulty generating new ideas.

This analysis is also consistent with thefindings from our post-experimental question-naire. Expectancy ratings were relatively low inall conditions (i.e. participants said it would bedifficult to generate additional ideas), whichindicates that many participants were runningout of ideas. And satisfaction ratings werehigher in the satisfaction condition than in theexpectancy condition, and for members ofinteractive groups than for members of nominal

groups, but not particularly high overall. Hadsatisfaction been greater in the satisfaction con-dition, maybe our groups would have stoppedearlier. Instead, they persisted, and only con-sidered ending their sessions when they haddifficulty generating more ideas.

Based on gender differences in self-evalua-tions, we also predicted an interaction betweengender and stop rule, and this prediction wasconfirmed. Men were more persistent thanwomen in the expectancy condition, whereaswomen were more persistent in the satisfactioncondition. And men were more productive inthe expectancy condition than women.However, although women persisted longerthan men in the satisfaction condition, this wasnot reflected in greater productivity, as weexpected it would. Instead, men and womenwere equally productive in that condition. Menand women apparently had similar norms forreaching satisfactory performance, but womenwere less productive per time unit and neededmore time to reach this level of performance.This finding is consistent with other researchreporting that women in groups focus both ontheir task and on relationships, whereas menfocus more exclusively on their task (see Aries,1998; Hoyenga & Hoyenga, 1993). Thus, thepersistence of women in interactive groups waspotentially due to their greater investment inrelationships, which absorbed additional time.Yet the male approach and the femaleapproach eventually led to similar levels ofgroup productivity.

Implications and future researchRecently, evidence has begun to cumulate infavor of group brainstorming. Although manyresearchers have identified problems associatedwith group work in general (e.g. Steiner, 1972)and with group brainstorming in particular(e.g. Diehl & Stroebe, 1987), recent evidencesuggests group interaction can be beneficial.One such benefit is that group interaction canencourage task persistence. According to ourresults, and those reported by Nijstad et al.(1999), people brainstorming alone feeldepleted after a relatively short period of time,which makes them want to stop working. Group

Nijstad et al. stop rules and persistence

203

01 046107 (bc/d) 30/6/04 2:26 pm Page 203

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

members, in contrast, can keep each othergoing, which eventually can compensate fortheir usual productivity loss, at least whengroups are relatively small. As an additionalbenefit, group members also report greatertask enjoyment and satisfaction.

The finding that groups are more persistentthan individuals at brainstorming may underliethe so-called illusion of group effectivity (seePaulus et al., 1993; Stroebe et al., 1992). Thereis a widespread belief that group brainstormingis more effective than individual brainstorming,even though laboratory research has clearlyshown the opposite. Perhaps people associate‘being effective’ with ‘being persistent’ duringgroup brainstorming. Outside the laboratory,there are often no time constraints (or muchweaker constraints) imposed on participants inbrainstorming sessions. Consequently, theproductivity of brainstorming groups outsidethe laboratory may be better than that ofgroups inside the laboratory, and the produc-tivity loss researchers have found in the labora-tory may not always occur in the real world(also see Sutton & Hargadon, 1996). Althoughresearchers have compared the performance ofinteractive and nominal groups in real worldsettings, and found a productivity loss in inter-active groups, this was done with fixed timelimits (e.g. Paulus, Larey, & Ortega, 1995).Future research should make this comparisonwithout time limits, to see whether a produc-tivity loss still occurs.

There are several other interesting issues thatcould be studied in subsequent research. Onesuch issue involves mixed-gender groups. Ourresearch suggests that gender composition maymatter when there is no fixed time limit forbrainstorming. One hypothesis is that malesmight be more influential than females whendecisions are made about ending sessions.Research has shown that males tend todominate in face-to-face mixed-gender groups(Aries, 1998) as well as in online groups(Postmes & Spears, 2002). This suggests thatthe results for mixed-gender groups wouldparallel the results we found for male groups.That might depend, however, on whether thediscussion topic is more masculine or feminine

(Deaux & Lafrance, 1998; Postmes & Spears,2002).

Another interesting issue is what wouldhappen if explicit performance criteria were set(i.e. groups and individuals were instructed toproduce a specific number of ideas, instead ofsimply brainstorming until they were satisfied).Paulus and Dzindolet (1993) found that settingdifficult and specific goals for brainstormingsessions of fixed length enhanced performanceconsiderably. However, this was true for bothinteractive and nominal groups, so the produc-tivity loss of interactive groups was not reduced.If there were no time limits, however, then theproductivity loss suffered by interactive groupsmight disappear if specific goals were set.

Overall, our results suggest that if groupbrainstorming is preferred, for whatever reasons(e.g. greater persistence, satisfaction, and enjoy-ment), then the satisfaction rule should beapplied, because it seems to encourage persis-tence and productivity. However, the satis-faction rule would primarily benefit groups withhigh performance standards, because perform-ance standards determine how satisfied peopleshould be before ending their brainstormingsessions. Groups with low performance stan-dards may stop relatively early in their sessions(cf. Paulus & Dzindolet, 1993). Further, thesatisfaction rule seems particularly effective forfemales in interactive groups. Different stoprules seem to matter less for males in suchgroups. For individual brainstormers, adifferent pattern emerges. Males seem morepersistent and productive when their abilitiesare questioned (with an expectancy stop rule),whereas different stop rules matter less forfemales. In sum, our results regarding genderdifferences in brainstorming persistence andproductivity under different stop rules suggesttwo things. First, men and women persist fordifferent reasons. Second, these reasons havemore impact on women who brainstormtogether and on men who brainstorm alone.

Notes1. The term expectancy is derived from Expectancy

Value models of motivation (e.g. Vroom, 1964).

Group Processes & Intergroup Relations 7(3)

204

01 046107 (bc/d) 30/6/04 2:26 pm Page 204

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

In those models, expectancy refers to theexpected future relation between effort andperformance (i.e. whether additional effort willlead to better performance). If someone isrunning out of ideas, and it becomes difficult togenerate additional ideas, then that person maynot expect additional effort to improveperformance much. As a result, his or hermotivation to continue weakens.

2. A 2 (type of group) � 2 (stop rule) � 2 (gender)analysis of covariance of productivity scores withpersistence as a covariate, revealed the usualproductivity loss for groups (F(1, 27) = 17.58,p < .001). Thus, corrected for their greaterpersistence, interactive groups (adjustedM = 36.97) were much less productive (per timeunit) than were nominal groups (adjustedM = 75.97). This suggests that interactive groupsindeed compensated for their usual productivityloss with greater persistence.

3. Measuring these periods of silence reliably wasnot feasible, because they were often very brief,especially in the interactive groups.

AcknowledgmentsThis research was funded by grant 575-31.007 of theNetherlands Organization for Scientific Research(NWO), awarded to W. Stroebe. The help ofD. Bijleveld, S. Gomperts, and J. van Middelkoopduring data collection and preparation is gratefullyacknowledged. An earlier version of this article waspresented at the Annual Meetings of the AmericanSociological Association, August 21–25 1997, in SanFrancisco, CA.

ReferencesAries, E. (1998). Gender differences in interaction:

A reexamination. In D. J. Canary & K. Dindia(Eds.), Sex differences and similarities incommunication: Critical essays and empiricalinvestigations of sex and gender in interaction(pp. 65–81). Mahwah, NJ: Erlbaum.

Bergman, M. M., & Scott, J. (2001). Youngadolescents’ wellbeing and health-risk behaviors:Gender and socio-economic differences. Journal ofAdolescence, 24, 183–197.

Betsworth, D. G. (1999). Accuracy of self-estimatedabilities and the relationship between self-estimated abilities and realism for women. Journalof Career Assessment, 7, 35–43.

Camacho, L. M., & Paulus, P. B. (1995). The role of

social anxiousness in group brainstorming. Journalof Personality and Social Psychology, 68, 1071–1080.

Costa, P., Terracciano, A., & McCrae, R. R. (2001).Gender differences in personality traits acrosscultures: Robust and surprising findings. Journal ofPersonality and Social Psychology, 81, 322–331.

Cross, S. E. (2001). Training the scientists andengineers of tomorrow: A person-situationapproach. Journal of Applied Social Psychology, 31,296–323.

Deaux, K., & Lafrance, M. (1998). Gender. In D. T.Gilbert & S. T. Fiske (Eds.), The handbook of socialpsychology (4th ed., Vol. 2, pp. 788–827). Boston:McGraw-Hill.

Dennis, A. R., & Valacich, J. S. (1993). Computerbrainstorms: More heads are better than one.Journal of Applied Psychology, 78, 531–537.

Diehl, M., & Stroebe, W. (1987). Productivity loss inbrainstorming groups: Toward the solution of ariddle. Journal of Personality and Social Psychology,53, 497–509.

Diehl, M., & Stroebe, W. (1991). Productivity loss inidea-generating groups: Tracking down theblocking effect. Journal of Personality and SocialPsychology, 61, 392–403.

Dugosh, K. L., Paulus, P. B., Roland, E. J., & Yang,H. C. (2000). Cognitive stimulation inbrainstorming. Journal of Personality and SocialPsychology, 79, 722–735.

Hoyenga, K. B., & Hoyenga, K. T. (1993). Gender-related differences: Origins and outcomes. Boston:Allyn & Bacon.

Jost, J. T. (1997). An experimental replication of thedepressed-entitlement effect among women.Psychology of Women Quarterly, 21, 387–393.

Judge, T. A., Locke, E. A., Durham, C. C., & Kluger,A. N. (1998). Dispositional effects on job and lifesatisfaction: The role of core evaluations. Journal ofApplied Psychology, 83, 17–34.

Kimball, M. M., & Gray, V. A. (1982). Feedback andperformance expectancies in an academic setting.Sex Roles, 8, 999–1024.

Kling, K. C., Hyde, J. S., Showers, C. J., & Buswell,B. N. (1999). Gender differences in self-esteem: Ameta-analysis. Psychological Bulletin, 125, 470–500.

Levine, J. M., & Moreland, R. L. (1990). Progress insmall group research. Annual Review of Psychology,41, 585–634.

Lynn, R., & Martin, T. (1997). Gender differences inextraversion, neuroticism, and psychoticism in 37nations. Journal of Social Psychology, 137, 369–373.

Martin, L. L., Ward, D. W., Achee, J. W., & Wyer,R. S., Jr. (1993). Mood as input: People have tointerpret the motivational implications of their

Nijstad et al. stop rules and persistence

205

01 046107 (bc/d) 30/6/04 2:26 pm Page 205

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from

mood. Journal of Personality and Social Psychology, 64,317–326.

Martin, T., & Kirkcaldy, B. (1998). Genderdifferences on the EPQ-R and attitudes to work.Personality and Individual Differences, 24, 1–5.

Mullen, B., Johnson, C., & Salas, E. (1991).Productivity loss in brainstorming groups: A meta-analytic integration. Basic and Applied Social Psychology, 12, 3–24.

Nijstad, B. A., Stroebe, W., & Lodewijkx, H. F. M.(1999). Persistence of brainstorming groups: Howdo people know when to stop? Journal ofExperimental Social Psychology, 35, 165–185.

Nijstad, B. A., Stroebe, W., & Lodewijkx, H. F. M.(2002). Cognitive stimulation and interference ingroups: Exposure effects in an idea generationtask. Journal of Experimental Social Psychology, 38,535–544.

Osborn, A. F. (1957). Applied imagination. New York:Scribner.

Parnes, S. J., & Meadow, A. (1959). Effects of‘brainstorming’ instructions on creative problemsolving by trained and untrained subjects. Journalof Educational Psychology, 50, 171–176.

Paulus, P. B., & Dzindolet, M. T. (1993). Socialinfluence processes in group brainstorming.Journal of Personality and Social Psychology, 64,575–586.

Paulus, P. B., Dzindolet, M. T., Poletes, G., &Camacho, L. M. (1993). Perception ofperformance in group brainstorming: The illusionof productivity. Personality and Social PsychologyBulletin, 19, 78–89.

Paulus, P. B., Larey, T. S., & Ortega, A. H. (1995).Performance and perceptions of brainstormers inan organizational setting. Basic and Applied SocialPsychology, 17, 249–265.

Paulus, P. B., & Yang, H. C. (2000). Idea generationin groups: A basis for creativity in organizations.Organizational Behavior and Human DecisionProcesses, 82, 76–87.

Postmes, T., & Spears, R. (2002). Behavior online:Does anonymous computer communicationreduce gender inequality? Personality and SocialPsychology Bulletin, 28, 1073–1083.

Steiner, I. D. (1972). Group process and productivity.New York: Academic Press.

Stroebe, W., & Diehl, M. (1994). Why groups are less

effective than their members: On productivitylosses in idea-generating groups. In W. Stroebe &M. Hewstone (Eds.), European review of socialpsychology (Vol. 5, pp. 271–303). London: Wiley.

Stroebe, W., Diehl, M., & Abakoumkin, G. (1992).The illusion of group effectivity. Personality andSocial Psychology Bulletin, 18, 643–650.

Sutton, R. I., & Hargadon, A. (1996). Brainstorminggroups in context. Administrative Science Quarterly,41, 685–718.

Taylor, D. W., Berry, P. C., & Block, C. H. (1958).Does group participation when brain-stormingfacilitate or inhibit creative thinking?Administrative Science Quarterly, 3, 23–47.

Vroom, V. (1964). Work and motivation. New York:Wiley.

Paper received 23 October 2002; revised version accepted14 January 2004.

Biographical notesB E R N A R D A. N I J S TA D is assistant professor at the

department of Work and OrganizationalPsychology of the University of Amsterdam. Hisresearch interests include group creativity andinnovation, group decision making, and indecision.

A N N E L I E S E. M. VA N V I A N E N is an associateprofessor in Work and Organizational Psychologyat the University of Amsterdam. Her researchinterests include gender and work, teamcomposition, career development and person-environment fit.

W O L F G A N G S T R O E B E is professor of psychology atUtrecht University. His research interests includedeterminants of group producitivity, attitudechange, and various areas of health psychology.

H E I N F. M. L O D E W I J K X is assistent professor at theDepartment of Social and OrganizationalPsychology at Utrecht University, and at TheNeherlands Open Univerisity. His researchinterests include intergroup competition, groupintitations, group productivity, and, lately, issues ofsenseless violence.

Group Processes & Intergroup Relations 7(3)

206

01 046107 (bc/d) 30/6/04 2:26 pm Page 206

at UVA Universiteitsbibliotheek on May 17, 2014gpi.sagepub.comDownloaded from