extending the stepladder technique: an examination of self ......that begins preliminary discussion...

Group Dynamics: Theory, Research, and Practice1998, Vol. 2, No. 2, 82-91

Copyright 1998 by the Educational Publishing Foundation1089-2699/98/$3.00

Extending the Stepladder Technique: An Examinationof Self-Paced Stepladder Groups

Steven G. Rogelberg and Matthew S. O'ConnorBowling Green State University

The stepladder technique was developed to facilitate group effectiveness by sequen-tially entering members into a group. Unlike past research where the experimenterregulated the entry of members into a group (Rogelberg, Bames-Farrell, & Lowe,1992), this study examined 26 4-person stepladder groups that proceeded through thestepladder process at a self-determined pace (i.e., group members decide how muchtime to take at each step). Self-pacing stepladder groups produced significantly higherquality group decisions than did the 26 4-person conventional groups studied. Further-more, members with the best individual decisions in self-pacing stepladder groups, onaverage, exerted the greatest amount of influence in their respective groups and hadmore influence on group decision quality than best members in conventional groups.

Despite the increased use of project teams,focus groups, autonomous work groups, qualitycircles, multifunctional work groups, and teamchief executive officers, groups are not alwayseffective. In response, a myriad of techniques(e.g., stepladder, delphi, nominal group, consen-sus decision making, dialectical inquiry) havebeen proposed to improve group performance.Each of these techniques affects how a group isstructured and organized to increase the probabil-ity of productive group processes, which in turnare expected to lead to effective performance.The present study examines the usefulness of aproposed, more practically oriented adaptationof the stepladder technique for facilitating groupperformance. In addition, the viability of thebest member influence explanation for steplad-der success broached in the original stepladderresearch is examined.

Steven G. Rogelberg and Matthew S. O'Connor, Depart-ment of Psychology Bowling Green State University.

We wish to thank Bill Balzer, Anita Cowley, MikeDoherty, and Scott Highhouse for their comments on a draftof this article. Thank you to Alison Elder, Tim Gramlich.Bob Yonker, and Carrie Williamson for their data collectionassistance. We would also like to acknowledge die generousfellowship support from the Bowling Green State UniversityAlumni Association.

Correspondence concerning this article should be ad-dressed to Steven G. Rogelberg, Department of Psychology,Bowling Green State University, Bowling Green, Ohio43403. Electronic mail may be sent to [email protected].

The Stepladder Technique

The stepladder technique (Rogelberg, Barnes-Farrell, & Lowe, 1992) is a decision-makingapproach intended to facilitate group effective-ness by structuring the entry of members into agroup. Step 1 of the technique involves thecreation of a two-person subgroup (the core)that begins preliminary discussion of the grouptask. After a fixed time interval, another groupmember joins the core group and presents hisor her ideas concerning the task. The three-person group then discusses the task in apreliminary manner. The process continues insteps until all members have systematicallyjoined the core group. When this occurs, thegroup arrives at a final solution. Figure 1displays the stepladder technique as applied to afour-person group.

The stepladder technique has four require-ments. First, each group member must havesufficient time to think about the group's taskbefore entering into the core group. Second, theentering member must present his or herpreliminary solutions before hearing the coregroup's preliminary solutions. Third, with theentry of each additional member to the coregroup, sufficient time must be provided todiscuss the task. Fourth, a final decision cannotbe reached until the group has formed in itsentirety.

In Rogelberg and colleagues* (1992) examina-tion of the stepladder technique, each four-

82

STEPLADDER TECHNIQUE 83

SteplCORE GROUPTwo group members discuss the task.

Step 2CORE GROUPThe two group members listen to the entering member

ENTERING MEMBERA third group memberpresents his/her initialideas to the core group.

CORE GROUPThe third group member joins the core group. The core group discusses the task.

Step 3CORE GROUPThe three members listen to the entering member.

ENTERING MEMBERA fourth group memberpresents his/her initialideas to the core group.

CORE GROUPThe fourth group member joins the core group. The group discusses andProduces a final group decision.

Figure 1. The three steps of the stepladder process applied to four-person groups.

person group proceeded through the stepladderprocess at a pace determined and regulated by anexperimenter. For example, the initial two-person core group discussion lasting for exactly7 min was followed by a 7-min, three-persondiscussion. The group task was Johnson andJohnson's (1987) winter survival exercise,where a group rank-orders 12 items remainingfrom an airplane crash in terms of their orderof importance for survival. Results indicatedthat stepladder groups produced significantlyhigher quality solutions to the exercise thangroups using the conventional approach todecision making (all members working on thetask collectively and at the same time). Further-

more, decisions provided by stepladder groupssurpassed the quality of their most expertmember's individual solution significantly moreoften (56%) than did the conventional groups(13%).

The Present Study

Given the increasing prevalence of groups,the initial documented success of the stepladdertechnique, and the costs associated with groupineffectiveness, further research is needed toconfirm, extend, and begin to examine why thestepladder technique may be effective. Thisstudy begins to address these research needs by

84 ROGELBERG AND O'CONNOR

making two unique contributions to the litera-ture. The first contribution, an applied contribu-tion, examines what happens when groups areallowed to regulate themselves when using thestepladder technique. That is, when groups areallowed to proceed through the stepladderprocess at a self-determined pace (i.e., theydecide how much time to take at each step)rather than at the previously used experimenter-regulated pace, do they still outperform conven-tional groups? The second contribution, thetheoretical contribution, examines the bestmember influence explanation that was postu-lated by Rogelberg et al. (1992) to explainstepladder success. Specifically, we examinewhether best members in stepladder groupsexert the greatest amount of influence in theirrespective groups and have more influence, ingeneral, on group decision quality than their bestmember counterparts in conventional groups.

Self-Pacing of the Stepladder Technique

Rogelberg et al.'s (1992) initial testing of thestepladder technique occurred in a highlytime-regulated environment such that groupsknew explicitly how long they were going tospend on each step of the stepladder process.Although this original research was an impor-tant first step in establishing the value of thestepladder technique, if the technique's effective-ness depends on particular and enforced timeregulation at each step, its practicality andfeasibility for organizations may be called intoquestion. First, explicit time regulations runcontrary to the notions of empowerment andautonomy. Second, the expectation that groupmembers will closely monitor elapsed time ateach step may be unreasonable and may result inthe group bringing in a facilitator to manage thestepladder process. Third, explicit time con-straints may appear forced and unnatural fordecision makers. Finally, the 7-min incrementsused in the initial stepladder research certainlywill not be applicable to all tasks, therefore, howdoes the team approach other tasks (e.g., whatare the time requirements)? Taken together, itfollows that for the technique to be effectivelyused in applied settings, stepladder groupsshould be able to work at whatever pace theychoose and feel is appropriate for the task athand. Specifically, the group itself shoulddetermine how long to spend at each step, in the

process. Unfortunately, past research has notexamined how self-pacing stepladder groupsperform relative to conventional groups. Notonly is this problematic given the aforemen-tioned practicality concerns, but, furthermore,applied psychologists cannot assume that self-pacing is merely an innocuous adaptation of theexperimenter-regulated stepladder procedure.When groups are allowed to self-pace, they aregiven complete control over their group pro-cesses. As a result, group members may choosenot to embrace the stepladder technique andinstead may speed hastily through the steps toinclude everyone in the group, thus creating the"feel" of the "natural" conventional groupstructure. Although theoretical work on thestepladder technique does not exist to substanti-ate this claim, unfortunately for advocates of thestepladder technique three bodies of researchwould suggest that this "not-embracing-the-stepladder-technique-inorder-to-be-more-like-a-conventional-group" concern is quite valid.First, individuals have a tendency to accept thestatus quo and avoid new ways of doing things;this is known as a status quo bias (Samuelson &Zeckhauser, 1988; Schweitzer, 1994). Second,research has found that people have a tendencyto repeat solutions and approaches that haveworked in the past; this is a type of fixationcalled mental set (e.g., Luchins, 1942). Third,resistance to change literature suggests thatpeople resist change due to habit and fear of theunknown (e.g., Nadler, 1983).

Taken together, these lines of research sug-gest that it cannot be assumed that a self-pacedadaptation of the original experimenter-regu-lated stepladder group will outperform conven-tional groups. Consequently, if application is anissue, establishing the value of self-pacedstepladder groups over conventional groups iswarranted.

Research Question 1 asks, How does theperformance of self-pacing stepladder groupscompare with that of conventional groups on adecision-making task? It is important to notethat using experimenter-regulated stepladdergroups for comparison purposes instead ofconventional groups was considered. However,the original stepladder findings have not, to date,been replicated. Therefore, it cannot be taken asfact that stepladder groups (in any form) doindeed outperform conventional groups. The useof conventional groups as the comparison group


works to provide additional evidence as to thereplicability of the original stepladder findings(even though an adaptation was made). Further-more, if experimenter-regulated stepladdergroups served as the comparison group, ourresearch aim would be to establish that self-pacing stepladder groups performed on par withthe comparison groups. As a result, we would beattempting to find support for the null hypoth-eses. As discussed by Keppel (1991), attemptingto prove the null hypothesis is statistically andmethodologically a limited proposition.

Why the Stepladder TechniqueMay Be Effective

Most models of group performance (e.g.,Gladstein, 1984; Sundstrom, De Meuse, &Futrell, 1990) emphasize input factors andprocess factors as important determinants ofgroup performance. One key input factor is thelevels of task-related expertise members bring tothe group (McGrath, 1984). One key processvariable is the group's ability to recognizeexpertise and to weigh member inputs to thegroup accordingly (Bottger, 1984; Bottger &Yetton, 1988; Einhorn, Hogeath, & Klempner,1977; Henry, 1995; Libby, Trotman, & Zimmer,1987; Littlepage, Robison, & Reddington, 1997;Littlepage, Schmidt, Whisler, & Frost, 1995). Inother words, members' influencing the groupcommensurate with their expertise positivelyaffects group performance.

Rogelberg et al. (1992) explained stepladdergroup success by postulating that members withthe best individual decisions in stepladdergroups were able to exert the greatest amount ofinfluence in their respective groups. Specifi-cally, the stepladder structure creates an environ-ment where individuals, to some extent, are"forced" to participate (a communication man-date). For example, it is difficult to avoidcommunication if you are in the initial interact-ing dyad, and it is nearly impossible to avoidcommunication when you are the group memberwho enters and is required to present your initialideas to the core group. Furthermore, becausethe entering member presents his or her ideasbefore hearing the group's ideas, conformitypressures may be decreased. Taken together,member expertise may become more apparent instepladder groups. As a result, a stepladdergroup may allow members with the best

individual decisions (best members) to exertinfluence in the group, which, ultimately,positively impacts group performance.

Rogelberg et al. (1992) found some prelimi-nary evidence supporting this contention. In asurvey administered after the group task, bestmembers across the stepladder group morestrongly endorsed the statement "They had achance to say what they wanted to say" than didother stepladder group members. Trie originalresearch, however, did not directly assesswhether best members actually exerted in-creased influence (instead it relied on post hocassessment of one survey item). The presentstudy directly assesses the amount of influencethe best members exert.

Research Question 2 asks, Do best membersin self-regulated stepladder groups exert agreater amount of influence on group decisionquality than other less expert members? andResearch Question 3 asks, Do best members inself-regulated stepladder groups have moreinfluence, in general, on group decision qualitythan best members in conventional groups?

Overall, this study's research questions haveboth practical and theoretical implications.Together, they are an important step in extend-ing and understanding the stepladder technique.

Method

Participants and Design

Students in undergraduate sociology andpsychology courses at a large state university inthe midwestern United States were recruited forthe study (61 men and 147 women), by MatthewS. O'Connor, during class time. Interestedindividuals (they were offered extra credit bytheir instructor) provided their names, sched-ules, and phone numbers to the experimenters.The experimenters then created four-persongroups by randomly sampling four individualsfrom each available time slot (individuals notchosen were moved to another time slot). Aftergroups were formed, groups were randomlyassigned to either the stepladder group condition(n = 26 groups) or to the conventional groupcondition (n = 26 groups). An equal proportionof men and women existed across each of thetwo conditions.


Materials and Measures

Experimental task. The problem-solving taskused was Johnson and Johnson's (1987, p. 110)winter survival exercise. To solve the wintersurvival task, participants read a vignette thathad them imagine that the airplane in which theywere traveling crashed in a remote northern areaduring the winter. The group members are theonly survivors and are stranded at least 30 milesfrom any known habitation. Participants thenrank ordered 12 items remaining from the crash(e.g., a hand ax) in terms of their order ofimportance for survival. There were multipleand subtle uses for each item alone and incombination with other items. Consistent withRogelberg et al. (1992) and other research (e.g.,Littlepage, Robison, & Reddington, 1997),performance on the winter survival task wasdefined as the sum of the absolute differencesbetween the ranks assigned by participants foreach item and those advocated by threewilderness experts (i.e., the criterion). Bysubtracting the calculated decision score (thesum) from 100, higher scores were made toreflect better quality decision than lower scores.The worst decision quality score was 28 and thebest possible score was 100.

Identification of the best member. Individu-als completed the winter survival task prior tothe group's collectively completing the sametask. At the conclusion of the study, decisionquality scores were calculated for each groupmember. Consistent with past research, theindividual with the highest decision qualityscore in each group was considered the bestmember (relatively speaking). It was never thecase that a group contained two individuals whoshared the designation of being the best member.

Influence. Influence was defined as thedeviation of the individual rank ordering ofitems from his or her group's rank ordering ofitems, summed over the 12 items;

Influence = /j ~ O.-12

where /,- is the rank assigned to the ith item by anindividual and Gi is the rank assigned to the ithitem by the group in which the individual was amember. A low score indicates high influence.Past research has found this influence index

positively related to participants' perceptions ofinfluence, participation rates, and content ofcommunications (Bottger, 1984; Littlepage &Mueller, 1997).

Time working on task. Total amount of timeworking as a group was calculated for steplad-der and conventional groups in people minutes.People minutes reflect the sum of the amount oftime each individual spent working in orpresenting to the core group. In the case ofconventional groups, people minutes would bethe total elapsed time multiplied by four (allmembers work in the core group from start tofinish).

Procedure

When a four-person group arrived at theexperimental location, the numbers 1 through 4(without replacement) were randomly assignedto participants. In stepladder groups thesenumbers served as the order of entry forparticipants.

Stepladder technique. The experimental pro-cedure used was nearly identical to the originalstepladder procedure conducted by Rogelberg etal. (1992), with one major exception: Groupmembers were informed that time limits orrestrictions did not exist and that they wouldproceed through the stepladder process at theirown chosen pace. First, the stepladder process,requirements, logistics, and the self-regulationconcept were explained to participants. Then, allfour participants were given the pregroup packet(this packet contained two copies of the wintersurvival exercise) to complete individually. Nocommunication among participants was permit-ted. After 7 min, Participants 1 and 2 turned inone copy of the task (participants kept the othercopy for reference during their group meeting)and Participants 3 and 4 turned in both copies.Participants 1 and 2 were then taken to anadjoining room (3.048 m X 3.048 m, with asquare table and four chairs) to work on theproblem together. When ready for their nextgroup member, they were instructed to informthe experimenter. While Participants 1 and 2were discussing the task, Participants 3 and 4watched a video documentary that was unrelatedto the task. Participants 3 and 4 were alsoinstructed not to discuss the task while waitingto enter the group. After Participants 1 and 2notified the experimenter of their readiness for


the next member, the amount of time theyworked together was recorded. Prior to beingtaken to the group, Participant 3 was given 30 sto privately read his or her individual solution.Then, Participant 3 was brought to the group.When the group was ready for their next groupmember, they were instructed to inform theexperimenter. After notification, the experi-menter recorded the amount of time Participants1, 2, and 3 worked together. Participant 4, whowas watching the video, was given 30 s toprivately review his or her individual solutionprior to entering into the group. Groups werethen instructed to use as much time as needed tocreate a final group solution. The amount of timethat the entire group worked together wasrecorded.

Conventional approach. After summarizingthe conventional approach (e.g., individualswould work together collectively to come upwith the one best decision), participants com-pleted the pregroup packet individually. Aftercompleting the experimental task, one of the twocopies of their individual responses was given tothe experimenter (participants kept the othercopy for reference), and the entire group wastaken to the experimental room that was used inthe stepladder group condition. Participants inthe conventional condition were instructed towork together to develop the one best solutionfor the task. The group was instructed that theycould come up with their final solution any waythey wanted and that they had as much time asneeded to generate a final solution. Timeworking together was recorded.

It should be noted that in both experimentalconditions participants were told that the goalwas for the group to come up with a solution thatwas as close as possible to the solution that thewinter survival experts developed for this sametask. To increase motivation, participants weretold they would be given feedback regardinghow their group solution compared with theirpeers* and the experts* solutions. A detailedexperimental protocol is available from StevenG. Rogelberg.

Results

the three steps that included the stepladderprocess (see Figure 1) was examined. In Step 1,the initial two-person core group discussionlasted, on average, 5.02 min (SD = 1.42). Step 2lasted, on average, 6.40 min (SD = 1.90). Step3, on average, was completed in 14.81 min(SD = 8.49). Time spent at Step 1 was posi-tively correlated with time spent at Step 2,r(24) = .49,/? < .05, and at Step 3, r(24) = .42,p < .05. Time spent at Step 2, however, was notsignificantly correlated with the time spent at

Time Spent

Before examining decision quality, the timeself-pacing stepladder groups spent at each of

Examination of total time spent working onthe task reveals that, on average, stepladdergroups (M = 87.76 people minutes, SD = 37.95)worked significantly longer, *(50) = — 4.62, p <.05, r\2 = .29, than conventional groups(M = 49.50 people minutes, SD = 18.38).

Group Decision Quality

With regard to Research Question 1, steplad-der groups' decisions (M = 59.07, SD = 10.25)were of significantly higher quality, t(50) =-2.33, p < ,05, TI2 = .10, than conventionalgroups (M = 53.65, SD = 5.83). Furthermore,decisions provided by stepladder groups sur-passed the quality, \(h N = 52) = 7.89, p <.05, of their best member's individual solutionmore often (62%) than did the conventionalgroups (23%).

It is important to note that conventionalgroups and stepladder groups did not differ withregard to member expertise. Specifically, bestmember decision quality scores across thestepladder groups (M = 58.00, SD = 7.85) didnot differ significantly, /(50) = .06, p > .05,from the best member scores across conven-tional groups (M = 57.85, SD = 8.35). In addi-tion, an index of group resources (the average ofthe four individual scores in the group) wascreated. Group resources for the stepladdergroups (M = 50.12, SD = 3.83) did not differ,f(50) = — .54, p > .05, from the group resourcesof the conventional groups (M = 50.85,SD = 6.06). Overall, groups in the two condi-tions appeared equivalent with regard to priorindividual resources. In fact, stepladder groupsstill produced significantly higher quality deci-sions (R2 = .10, p < .05) than conventionalgroups after controlling for group resources andbest member scores by way of a hierarchicalmultiple regression.


Time working on the task was not controlledfor in the aforementioned decision qualityanalyses. However, time in and of itself cannotbe credited solely for stepladder success. If timewere the causal agent, then longer workingconventional groups would outperform shorterworking conventional groups. This was not thecase. Instead, time working on the task andgroup performance were unrelated in conven-tional groups, r(24) = .07, p > .05. Further-more, time spent in Step 1, Step 2, and Step 3 inthe stepladder groups was not significantlycorrelated (p > .05) with group decision qual-ity, rs(24) = .04, .36, and .35, respectively.Finally, total amount of time (across all threesteps) was not significantly related to groupdecision quality for stepladder groups, r(24) =.37, p > .05. It is worth noting that althoughthese correlations were not statistically signifi-cant using our decision rule (a = .05), thosecorrelations greater than .33 would have beendeemed significant using a more liberal decisionrule (a = .10).

Before concluding the decision quality analy-sis section, an exploratory analysis was con-ducted whereby a solution diversity score wascalculated for each group and subsequentlyrelated to decision quality. Solution diversityrefers to variation (i.e., the standard deviation)across team members' pregroup individualdecision quality scores (Wanous & Youtz,1986). A high-solution diversity score suggeststhat individuals in a group were very heteroge-neous with regard to how they approached thedecision-making task. Conversely, a small-solution diversity score suggests that individualswere very homogeneous with regard to howthey approached the task. Solution diversity instepladder groups (M = 6.83, SD = 3.70) wasfound to be strongly related to group perfor-mance, r(24) = .64, p < .05. Solution diversityin conventional groups (M - 7.09, SD = 3.72)was unrelated to group performance, r(24) =.14,p>.05.

Best Member Influence

An influence score was calculated for eachmember in each respective group. Lower scoresreflected greater influence on the group'sperformance. Complete individual influence onthe group would be reflected by a score of zero

(i.e., a group member's item rankings wereidentical to the group's final decision).

The first set of analyses focused exclusivelyon stepladder groups. Influence scores ofindividuals in the initial two-person core group(M = 32.25, SD = 10.40), group members enter-ing in Step 2 (M = 29.57, SD = 12.50), andgroup members entering in Step 3 (M = 31.65,SD = 10.65) were not significantly differentfrom one another, F(3> 99) = .50, p > .05. Withregard to Research Question 2, average influ-ence scores for best members were calculatedand compared with average influence scores forthe other three members in the group. Across the26 groups, best members in stepladder groupsexerted significantly greater influence(M = 23.03, SD = 8.11) on the group than thethree other stepladder group members(M = 34.27, SD = 8.45), 1(50) = 4.87,^ < .05,T\2 = .32. Conversely, best members in conven-tional groups did not exert significantly moreinfluence (M = 31.61, SD = 9.10) on the groupthan the other conventional group members{M = 31.79, SD = 6.73), f(50) - .08, p > .05.

In response to Research Question 3, averageinfluence scores of best members in stepladdergroups (Af = 23.03, SD = 8.11) were comparedwith average influence scores of best membersin conventional groups {M = 31.61, SD = 9.10).On average, best members in stepladder groupsexerted more influence on the group than bestmembers in conventional groups, *(50) = 3.58,/><.05 ,V = -20.

Finally, the relationship between best memberindividual performance and group performancein stepladder groups was examined and com-pared with the corresponding observed relation-ship in conventional groups. A strong relation-ship between individual performance and groupperformance would be another indication thatthe individual positively influenced the groupwith regard to decision quality. In stepladdergroups, best member individual scores andgroup scores were highly correlated, r(24) =.75, p < .05. Conversely, in conventional groupsbest member individual scores and group scoreswere unrelated, r(24) = .02, p < .05.

Discussion

The present study examined the quality ofdecisions produced by self-pacing stepladdergroups in comparison to conventional groups


and why they were effective. Despite theconcern that self-pacing stepladder groupswould advance hastily through the steps toachieve the look and feel of a conventionalgroup, the technique's effectiveness for facilitat-ing group performance was found to generalizebeyond the setting of specific time limits at eachstep.

By embracing the stepladder technique itappears that the potential process benefits of theexperimenter-regulated technique, as discussedby Rogelberg et al. (1992), may have occurredin self-paced stepladder groups. Specifically,data strongly suggest (i.e., large effect sizes) thatbest members in stepladder groups not onlyexerted the greatest amount of influence in theirrespective groups but also had more influence,in general, on group decision quality than theirbest member counterparts in conventionalgroups. Therefore, it appears that Rogelberg andcolleagues' (1992) initial postulation concerningthe role of the best member and stepladdersuccess was correct. The "forced" participation(e.g., having an entering member present his orher ideas before hearing those of the group)associated with the stepladder technique mayindeed facilitate the recognition of expertise,which subsequently results in best membersexerting increased influence.

Despite the influence the best members wereable to exert, 62% of the stepladder groups stillsurpassed the quality of their best members*individual decisions. A number of potentialinterrelated explanations exist to explain whystepladder groups accomplished this rare feat(e.g., Hill, 1982). Moreland, Argote, andKrishnam (1996) argued that groups possessingtransactive memory systems (a memory of whatknowledge is possessed by various groupmembers; who knows what) are primed toperform more effectively than groups notpossessing such a memory system. Although itwas suggested that these systems develop overtime and interactions, we suggest that based onthe fact that stepladder group members system-atically enter and present ideas, member inputsare made salient to the group. As a result of thesesalient member inputs, transactive memorysystems may readily develop in stepladdergroups (even in short amounts of time), which,in turn, leads to enhanced group performance.Furthermore, when member inputs are madesalient to the group, stepladder groups may be

less subject to the common knowledge effect(Gigone & Hastie, 1997). A common knowledgeeffect occurs when a group weighs information,when making a decision, on the basis of howwell-known the information is across groupmembers (Gigone & Hastie, 1997). Becausemember inputs are highly salient, and the groupcannot make a final decision until all membershave entered the group, stepladder groupmembers may be more willing to weighunshared information (information not com-monly known) when making a final decision;this in turn may lead to superior group decisions.Finally, a number of the reasons suggested byRogelberg et al. (1992) on why stepladdergroups outperform conventional groups mayalso be relevant in explaining why stepladdergroup decisions surpass the quality of their bestmembers. Specifically, Rogelberg and col-leagues (1992) suggested that the communica-tion mandate created by the stepladder processincreases the range of ideas expressed andreduces social loafing, both of which may enablestepladder group decisions to surpass the qualityof their best members. Furthermore, three forcesmay be working to increase critical decisionmaking in stepladder groups. Critical decisionmaking, in turn, leads to superior groupperformance (e.g., Janis, 1982). First, enteringmembers present ideas prior to hearing thegroups' ideas and, as a result, blind conformitymay be prevented. Second, task-related conflictmay be imposed on the stepladder groups whenthe entering member presents ideas counter towhat the group has already discussed, thusincreasing the groups' propensity to evaluatedifferent courses of action. Finally, because thegroup is not formed in its entirety until the lastmember joins, pressure to reach a prematuresolution may be delayed.

Future Research and Limitations

Additional conditions under which the steplad-der technique may be appropriate still need to beidentified. For example, the optimal number ofpeople to include in the stepladder process, afterwhich the process is no longer effective and timeefficient, must be determined. Although thestudent groups studied here are perhaps analo-gous to many temporary ad hoc groups (e.g.,project teams) found in industry, a need exists toreplicate these findings using established groups,


preferably in an organizational setting. At thesame time, research should examine the willing-ness of organizational members to use thestepladder technique instead of the conventionalapproach in certain circumstances and overtime. The types of tasks most conducive to beingcompleted effectively by stepladder groups(e.g., problem-solving tasks) and the types oftasks least conducive to being completedeffectively by stepladder groups (e.g., lessdefined tasks or tasks that require multiple groupmeetings) should also be identified. Finally,research should examine the effectiveness of thestepladder technique in facilitating the perfor-mance of nontraditional groups. One such groupis one asked to "interact" by means of thetelephone (i.e., teleconferencing). The steplad-der process may be very appropriate and natural(e.g., when the core group is ready for their nextmember, they can telephone him or her) forfacilitating and improving teleconferencinggroups' performance.

In addition to these generalizibility types ofresearch questions, future work should addresshow to optimize stepladder success. Past re-search and this research randomly assignedindividuals to the various entry positions instepladder groups (e.g., an individual is ran-domly assigned to be the first entering member).It is possible, however, that certain individualsmay be better suited for different "spots" in theprocess, given their personality and ability. Forexample, a creative type of person may be betterpositioned in the initial interacting core group,whereas a person who pays attention to detailmay be better suited as the final enteringmember. Additional research should examinewhether stepladder success can be optimized byassigning individuals to various spots in thestepladder process based on a personality andability profile.

An additional research question of interestconcerns the value of the stepladder technique infacilitating the performance of diverse workteams. For diverse work groups to be successful,they must use their diversity as a valuableresource as opposed to a hindrance. Exploratorydata from this study indicates that stepladdergroups dealt effectively with one type of teammember heterogeneity: causing solution diver-sity. Specifically, the strong positive correlationbetween solution diversity and group perfor-mance in stepladder groups suggests that the

stepladder technique may be able to facilitatethe performance of groups containing teammembers with diverse perspectives on the grouptask. Future work should examine the stepladdertechnique's effectiveness in facilitating theperformance of demographically diverse (e.g.,age), psychologically diverse (e.g., values), andorganizationally diverse (e.g., occupation) workgroups.

Conclusions

Group decisions that do not incorporate theknowledge and resources of the group members,and that are made prematurely in a noncriticalmanner, can result in negative consequences forindividuals, projects, and organizations. Typi-cally, groups operate under the assumption thatthe conventional approach to group decisionmaking is the only tool for decision making.Although the conventional approach certainlyhas its place in organizational decision making,group members should think critically about thetask at hand to determine the most appropriatetool for decision making. Data from this studysuggest that the self-paced stepladder techniqueis one such tool that demonstrates promise forincreasing the quality of group decision making.

Overall, the self-pacing findings represent animportant step in assessing the flexibility of thestepladder technique. If the technique was notaccommodating to self-pacing, the technique'sfuture for organizational decision making mayhave been suspect. Self-regulation of the steplad-der technique offers additional control to groupsand is more natural and generalizable toorganizations.

References

Bottger, P. (1984). Expertise and air time as bases ofactual and perceived influence in problem-solvinggroups. Journal of Applied Psychology, 69, 214—221.

Bottger, P. C , & Yetton, P. W. (1988). An integrationof process and decision scheme explanations ofgroup problem solving performance. Organiza-tional Behavior and Human Decision Processes,42, 234-249.

Einhorn, H. J., Hogeath, R. M., & Klempner, E.(1977). Quality of group judgment. PsychologicalBulletin, H4, 158-172.

Gigone, D., & Hastie, R. (1997). The impact of


information on small group choice. Journal ofPersonality and Social Psychology, 72, 132-140.

Gladstein, D. (1984). Groups in context: A model oftask group effectiveness. Administrative ScienceQuarterly, 29, 499-517.

Henry, R. A. (1995). Improving group judgmentaccuracy: Information sharing and determining thebest member. Organizational Behavior and HumanDecision Processes, 62, 190-197.

Hill, G. W. (1982). Group versus individual perfor-mance: Are N + 1 heads better than one?Psychological Bulletin, 91, 517-539.

Janis, I. L. (1982). Groupthink: Psychological studiesof policy decisions and fiascoes. Boston: HoughtonMifflin.

Johnson, D., & Johnson, F. (1987). Joining together:Group theory and group skills. Englewood Cliffs,NJ: Prentice-Hall.

Keppel, G. (1991). Design and analysis: A research-er's handbook. Englewood Cliffs, NJ: PrenticeHall.

Libby, R., Trotman, K. T., & Zimrner, I. (1987).Member variation, recognition of expertise, andgroup performance. Journal of Applied Psychol-ogy, 72, 81-87.

Littlepage, G. E., & Mueller, A. L. (1997). Recogni-tion of expertise in groups: Expert characteristicsand behavior. Group Dynamics: Theory, Research,and Practice, 1, 324-328.

Littlepage, G. E., Robison, W., & Reddington, K.(1997). Effects of task experience on groupperformance, member ability, and recognition ofexpertise. Organizational Behavior and HumanDecision Processes* 69, 133-147,

Littlepage, G. E., Schmidt, G. W., Whisler, E. W., &Frost, A. G. (1995). An input-process-output modelof influence and performance in problem-solving

groups. Journal of Personality and Social Psychol-ogy, 69, 877-889.

Luchins, A. S. (1942). Mechanization in problemsolving. Psychological Monographs, 54(6, WholeNo. 248).

McGrath, J. E. (1984). Groups: Interaction andperformance. Englewood Cliffs, NJ: Prentice Hall.

Moreland, R. L., Argote, L., & Krishnam, R. (1996).Socially shared cognition at work: Transactivememory and group performance. In J. L. Nye &A. M. Brower (Eds.), What's social about socialcognition? Research on socially shared cognitionsin small groups (pp. 57-84). Thousand Oaks, CA:Sage.

Nadler, D. A. (1983). Concepts for the managementof organizational change. In J. R. Hackman, E. E.Lawler III, & L. W. Porter (Eds.), Perspectives onbehavior in organizations (pp. 551-561). NewYork: McGraw-Hill.

Rogelberg, S. G., Barnes-Farrell, J. L., & Lowe, C. A.(1992). The stepladder technique: An alternativegroup structure facilitating effective group decisionmaking. Journal of Applied Psychology, 77,730-737.

Samuelson, W., & Zeckhauser, R. (1988). Status quobias in decision making. Journal of Risk Uncer-tainty, 1, 7-59.

Schweitzer, M. (1994). Disentangling status quo andomission effects: An experimental analysis. Orga-nizational Behavior and Human Decision Pro-cesses, 58, 457^76.

Sundstrom, E., De Meuse, K. P., & Futrell, D. (1990).Work teams: Application and effectiveness. Ameri-can Psychologist, 45, 120-133.

Wanous, J. P., & Youtz, M. A. (1986). Solutiondiversity and the quality of group decisions.Academy of Management Journal, 29, 149-159.

extending the stepladder technique: an examination of self ......that begins preliminary discussion...

Documents