A Functional Theory of Task Role Distribution in Work Groups

A Functional Theory ofTask Role Distribution inWork Groups

Joachim Stempfle, Oliver Hübner and Petra Badke-SchaubUniversität Bamberg/Germany, Institut für Theoretische Psychologie

In social psychology, much research has been conducted with regard to small groups, focusingon a few recurrent themes. Questions of group structure and group composition, however, havebeen widely neglected in social psychology research, although their importance is beingrecognized (Barrick, Stewart, Neubert, & Mount, 1998; Moreland, Levine, & Wingert, 1996).The following study focuses on explaining the emergence of a task role distribution in self-organizing work groups. A theory of task role distribution in work groups is being proposedcapable of predicting the distribution of task roles in a team based on group members’ skills andpreferences. The basic assumption behind the theory is that teams will strive to organizethemselves in a functional way, taking into account both external demands (tasks that areassigned to the team) and internal demands (needs and preferences of the team members). Acase study has been conducted on a self-organizing team of 15 student software developersworking on a project task over a period of three weeks. The theory has been applied to thestudent project team. For the majority of team members the task role each team member wasgoing to be assigned was predicted correctly by the theory.

keywords roles, self-organization, task role distribution, teams, work groups

Group Processes &Intergroup Relations

2001 Vol 4(2) 138–159

FOCUSING on the study of work groups, a largefield opens up containing many different typesof groups operating in a wide range of differentenvironments. These groups differ in manyattributes. One important attribute of workgroups is the group structure, which McGrath(1984, p. 11) defines as ‘the continuing pat-terned relations among individuals who aremembers’. With regard to aspects constitutingthe group structure such as task roles, hierarchy,group member responsibilities and groupnorms, in many cases a formal, well-definedgroup structure is being pre-establishedthrough organizational norms and rules. This

normative, prescribed structure may entail cleardescriptions of task roles, clear-cut differencesin rights and responsibilities of group membersand a strongly regulated interaction process bymeans of standard interaction procedures.

GPIR

Copyright © 2001 SAGE Publications(London, Thousand Oaks, CA and New Delhi)

[1368-4302(200104)4:2; 138–159; 016530]

Author’s noteAddress correspondence to Joachim Stempfle,Institut für Theoretische Psychologie,Universität Bamberg/Germany, Markusplatz 3,D-96045 Bamberg, Germany. [email:[email protected]]

5 - Stempfle (jk/d) 6/3/01 10:26 am Page 138

http:\\www.sagepub.co.uk

Examples for teams that fit into this descriptionare flight crews (Helmreich & Foushee, 1993)and surgical teams (Helmreich & Schaefer,1994).

There are many work groups, however, suchas project teams, task forces or self-organizingwork groups in the industry, in which the groupstructure is much less predefined. In suchgroups, the group structure must first emergeduring the process of group formation andgroup interaction. Roles, norms and communi-cation patterns are not predefined, but must beestablished through interaction. In such groups,establishing and maintaining a functional groupstructure presents an important requirementthat must be met in order for the group toperform its tasks. Even in groups on which thegroup structure is largely imposed by organiz-ational norms, an informal structure frequentlyarises that may even counter the formal struc-ture imposed by higher organizational levels(Krech, Crutchfield, & Ballachey, 1962; Rosen-stiel, Molt, & Rüttinger, 1995).

This emphasizes the need to study self-organizing processes in work groups, particu-larly with regard to the emergence of the groupstructure. While it is widely recognized that inany self-organizing group some kind of groupstructure will evolve, the mechanisms are stillunclear that describe just how a specific groupstructure arises and why different groupmembers are being assigned specific task rolesand responsibilities. Why is person A in a projectteam being recognized by her colleagues as theinformal group leader and not person B? Whydoes person C frequently interact with cus-tomers but person D doesn’t?

Theory

Self-organization in work groupsThe fact that groups of any kind will establish agroup structure with norms and roles ‘by them-selves’ points to a process of self-organizationtaking place in groups. In self-organizationtheory (Haken, 1995; Prigogine, 1985), self-organization is defined as the spontaneousdevelopment of dynamically ordered states(Schuster, 1987). Self-organization arises in

open systems that are in continuous exchangewith their environments. Self-organization rep-resents an adaptive process, with the systemadapting to its environment through changes inthe system’s structure and organization.

Work groups can be regarded as opensystems. Just like any open system, work groupsexist in a specific environment. This environ-ment (the department, the company, the state,etc.) provides the group with certain resources,expecting certain outcomes in return. Workgroups are created around a particular task: ful-fillment of the group task is the group’s primaryconcern (Argyle, 1983). If the group fails tomeet its goals and doesn’t produce the expectedoutcomes, the group’s very existence is indanger. In this regard, the work group’senvironment pushes the group to structure itselfin a way that will enable the group to functionmost effectively. As McGrath, Arrow, andBerdahl (1999) have pointed out, adaptation tothe environment (achieved through changes inthe group structure and through manipulationsof the external world) is driven by a dynamicinterplay between the group and its embeddingsystems.

To make things even more complicated, thereis not only the group task that needs to be suc-cessfully accomplished. Every group memberalso pursues his or her own individual goalsstemming from individual needs that may ormay not correspond to the group’s goals. Socialexchange theory (Gergen, Greenberg, & Willis,1980; Thibaut & Kelley, 1959) suggests that inany social interaction people try to balance whatthey put into the interaction with what they getout of it. Similar to economics, interaction‘costs’ (e.g. supporting others) are put inrelation to interaction gains (e.g. social accept-ance). If a relationship seems unbalanced in thesubjective eye of the beholder, distress willresult, leading to attempts to reduce distress byrestoring actual or psychological equity.

Social exchange theory has been applied tothe development of groups over time by More-land, Levine, and Cini (1993), who point outthat group members periodically evaluate thebenefits they receive from membership in thegroup by taking into account the contribution of

Stempfle et al. a functional theory of task role distribution

139


the group with respect to satisfaction of theirown individual needs. Depending on the resultof this evaluation process, group members willadjust their level of commitment. A groupmember striving for individual advancement inan organizational team, for example, will expectthe group or the group leaders to support his orher career in exchange for the effort he or sheis contributing to the achievement of the grouptask. If the group fails to recognize the needs ofits member, it is likely that the member willreduce his or her level of commitment, trying torestore actual or psychological equity by reduc-ing his or her efforts, by creating conflict or evenby leaving the group. Thus, it is important forthe group to maintain a high level of equity inthe group. Otherwise the functioning of thegroup is threatened.

Both of the demands proposed above, theneed for the group to adapt to externaldemands (success in fulfilling the group task)and internal demands (recognizing the needsand preferences of the group members) areemphasized in recent literature. McGrath et al.(1999) follow Moreland & Levine (1982) whenpointing out two generic purposes of the group:completing the group’s projects and fulfillingmember needs. In accordance with these find-ings, we propose external and internal demandsto be the driving forces behind adaptation ingroups. The establishment of a group structurecan be seen as an adaptation effort initiated bythe group. External and internal demands willtherefore play an important role in the estab-lishment of the group structure. The strength ofthe influence of internal and external demandson group adaptation and group structure willvary depending on the nature of the group’senvironment, however. Groups operating in arestrictive environment in which success in taskaccomplishment is critical for the group’s veryexistence will be primarily concerned with adap-tation to external demands, potentially at theexpense of internal demands. On the otherhand, in groups in which external pressure islow, adaptation to internal demands will be thegroup’s primary concern.

Building on this theoretical background, ourhypothesis is that adaptation is the driving force

behind the establishment of a group structure inself-organizing groups. Groups that are giventhe opportunity to structure themselves will dothis in a functional way by striving for a ‘goodfit’ with respect to external and internaldemands emerging in the group spontaneously.Note, however, that we do not propose theemergence of a ‘best fit’. Obviously, there aremany different ways for a group to structureitself, some of them more or less functional thanothers. We expect that the group will try to avoidand correct dysfunctional group structures thatendanger the group’s stability or significantlyhamper the group’s progress toward its goals,but we expect that the group will be satisfiedonce it has established a structure that producessatisfactory results in task accomplishment andserves to meet the needs of most of the groupmembers. This proposition is consistent withresearch on behavioral decision theory (Beach,1993) which shows that in decision-making taskspeople do not strive for an optimum decision,but are satisfied once a chosen alternativeexceeds a certain threshold value.

Recapitulating, we have proposed that groupsstrive to adapt to internal and external demandsby establishing a functional group structure. Atthis point, we need to state more precisely whatit means for a group to establish a functionalgroup structure. Specifically, we will predict howtask roles in the group will be distributed on thebasis of internal and external demands, thusestablishing a functional distribution of taskroles in the group.

A functional distribution of task rolesWe have stated earlier that work groups are con-fronted with a group task. The group task is thendivided into several smaller subtasks. Onesubtask or a set of subtasks is assigned to eachgroup member. We refer to the whole of sub-tasks assigned to a single group member as the‘task role’, building on role concepts that havelong been neglected in psychology (Blumberg,Hare, Kent, & Davies, 1983), but limiting theseconcepts in accordance with Emery’s (1978)concept of ‘occupational roles’ to the clearlydefined aspect of tasks assigned to a groupmember. In a work group, each member

Group Processes & Intergroup Relations 4(2)

140


occupies a specific task role that defines what hisor her particular job is in the group. The distri-bution of task roles in a work group we will referto as the ‘task role distribution’.

Teams composed of highly qualifiedmembers such as project teams and cross-func-tional teams are characterized by a heterogene-ity of task roles. In a software engineering team,one person may be working on the programcode, another one may at the same time beworking on the graphical user interface, a thirdone may be interacting with the customer, etc.Each task role will place certain demands on theperson entrusted with the task role. Of course,group members also differ with regard to theirskill profiles. In their recent review of five majorstudies on skills and competencies required inthe workplace today, O’Neil, Allred, and Baker(1997) name four major categories of skillsrequired in the workplace today: basic academicskills, higher order thinking skills, interpersonaland teamwork skills and personal characteristicsand attitudes. Under each one of thesecategories, a large number of subskills can beclassified.

Given the diversity of tasks and skills requiredin today’s workplace, it seems highly likely thatfew people possess extraordinary skills in allrelevant areas. Rather, it seems likely that mostpeople display a skill profile with considerablestrengths in some areas and weaknesses in otherareas. What follows is that not all task role distri-butions are equally functional: if completion ofthe group’s task is the group’s primary concernthen the group should structure itself in a waythat different task roles are assigned to differentgroup members according to their individualskills. In the above example, one group membermay display excellent computer programmingskills, another one may be quite good at relatingto people. In order to maximize group effec-tiveness, the skilled computer programmershould be working on the program code whilethe group member who is good at relating topeople should be handling customer requests.The better the match between task roledemands and skills of the person assigned thistask role, the higher the probability that theperson will successfully complete his or her task.

In the theory, task role demands operate as athreshold value, however, with deviations ofmember skill from task role demands only beingtaken into account if task role demands exceedthe group member’s skills. In this case, failure intask completion is likely. This is taken intoaccount by the theory. On the other hand, oncea group member’s skills match or exceed taskrole demands, the size of the difference betweenthe group member’s skills and task roledemands is not taken into account.

Thus, the first requirement for a functionaltask role distribution is:

A functional task role distribution is one where thedemands of each task role do not exceed the skillsof the group member to which this task role hasbeen assigned.

Elsewhere we have called this requirementthe principle of competence (Stempfle, 1998b). Wewould expect groups that structure themselvesaccording to the principle of competence toattain a good fit with regard to their environ-ment.

As we have already mentioned above, com-pletion of the group task is not the only require-ment the group has to fulfill. Group memberseach strive to achieve their own individual goals.In exchange for the effort they put into thegroup task, group members expect the group toprovide them with something in return. Thedifferent task roles that the group offers to itsmembers do not only require different skills andefforts, but also provide opportunities to fulfilldifferent needs. A tricky programming task mayserve to satisfy a strong need for achievement.The role of a project manager may serve to fulfilla need to exert power and social influence. Interms of Vroom’s (1964) motivation theory,different task roles can be more or less instru-mental to the individual’s needs. Depending onthe instrumentality of the different task roleswith regard to the needs of each individualgroup member, each group member will holdindividual preferences about the different taskroles. Group members will prefer task roleswhich appear highly instrumental to their indi-vidual needs. In terms of social exchangetheory, being assigned a preferred task role will


141


be experienced by a group member as highlyrewarding. Social exchange theory predicts thatthe effort this group member will put intoachieving the task will be equally high.

Social exchange theory also allows predic-tions for group members who are assigned to atask role that doesn’t meet their own prefer-ence. In their subjective view, these groupmembers receive only little reward by the group.In order to restore actual or psychologicalequity, they will be likely to show some kind ofdisrupting or withdrawal behavior, thus becom-ing marginal members or even ex-members ofthe group (Moreland et al., 1993). In order tosuccessfully complete the group task, however,the group needs highly motivated members whoare eager to complete the tasks that have beenassigned to them. What follows is that a func-tional task role distribution is one where as manygroup members as possible are being assigned atask role that they themselves prefer. In terms ofsocial exchange theory, such a distribution oftask roles would produce only a minimum ofinequity.

Thus, the second requirement for a func-tional distribution of task roles in work groups is:

A functional distribution of task roles is one whereas many group members as possible are beingassigned to a task role that they themselves prefer.

Elsewhere we have called this requirementthe principle of preference (Stempfle, 1998b).

Figure 1 depicts the two principles we havestated and the proposed effects of these prin-ciples on group functioning.

Now that we have stated the two fundamentalprinciples on which the functional theory oftask role distribution is built, the question thatneeds to be answered is how the two principlescombine in producing the evolving task roledistribution in the group. Evidently, the twodemands do not necessarily have to be inaccordance with each other. They may in factpresent the group with an unsolvable dilemma.It is possible that a group member prefers a taskrole that does not match his or her skills. In thesame manner, a group member ideally suitedfor a particular task role may still prefer anothertask role. The group thus must try to achieve


142

Figure 1. Effects of subtask assignment.


some kind of equilibrium between the two prin-ciples, establishing a task role distribution bytaking into account both the skills and thepreferences of its members. But are both prin-ciples of equal importance for the distributionof task roles in the group? We believe that thisnot so. In fact, our hypothesis is that the relativeweight of each of the two principles will vary foreach particular task role, depending on what wewill call the centrality of the task role.

We define centrality by use of a number ofcriteria:

– The more central a task role, the greater itsimportance with regard to achievement of thegroup goal. If a central task is not completed suc-cessfully, the group as a whole will be likely to failin achieving its goals.– The more central a task role, the higher theskill level that is needed to complete the taskrole. Central task roles require expert skills. – The more central a task role, the more diffi-cult it is to reassign the task role to a differentgroup member during the on-going process ofgroup work. Reassignment of a central task rolewill be difficult and time-consuming.

In most cases, we expect the three criterialisted above to be somewhat correlated,although there may be instances when this is notthe case.

In the project team we have observed, anexample for a central task role was the role ofthe chief programmer. The chief programmerwas responsible for establishing the architectureof the software product and for defining themain variables and the interfaces betweendifferent parts of the program. Successful com-pletion of these tasks was indispensable for thegroup in order to solve its task. The tasksrequired a high level of expertise and skill inprogramming and analytical thinking. Once thetask role had been assigned, it could only bereassigned under great difficulty, since most ofthe central concepts and the architecture of theprogram was only stored in the chief program-mer’s memory and did not exist in written form.It would have been really difficult and time-consuming for another person to try to under-stand the architecture that the former chief

programmer had established. On the otherhand, an example for a task role of little cen-trality in the observed project team was the roleof the documenter. It was the documenter’s jobto document the on-going process of projectwork, noting what has been accomplished andwhat problems have occurred. Although aprocess documentation was requested by boththe project management and the customer, theproject task could be accomplished even if thedocumenter failed in his task. The process docu-mentation mainly served as an add-on and wasnot a central ingredient of the project task.

As for the implications of task role centrality,we expect groups to strive strongly toward theoptimization of task role assignments for centraltask roles, whereas for less central task roles sub-optimum task role assignments will be morereadily tolerated. The group will put much moreeffort into optimizing the assignment of centraltask roles than into optimizing the assignmentof task roles of little centrality. Thus, the morecentral a task role, the more effort the group willput into assigning it to a group member whoseskills and preferences closely match the taskrole.

Furthermore, centrality in the theory servesto adjust the weight of skill vs. preference in theassignment of task roles: the more central a taskrole, the more important it is to assign the taskrole according to skills rather than preferences.The group can by no means afford to assign acentral task role to a person who doesn’t possessthe corresponding skills and will be likely to failin task accomplishment, even if this personshows a high preference for the task role. On theother hand, for the assignment of task roles ofvery low centrality it doesn’t really matter who isbeing assigned the task role. For this reason, theless central a task role the more it should beassigned according to preferences rather thancompetencies in order to minimize dissatis-faction among the group members.

Besides the two principles and task role cen-trality, a fourth influencing factor on task roledistribution must be considered. We have statedearlier that the strength of external and internaldemands may vary not only for each single taskrole, based on differences in task role centrality,


143


but may also depend on environmental con-ditions. Consider a research team in which thefunding the team receives is directly dependenton the team’s success in their research activities.The team periodically receives feedback on itsperformance, which determines the team’sbudget. We would not be surprised if this teamplaced somewhat greater weight on the principleof competence than on the principle of prefer-ence, since success or failure in task accomplish-ment is of vital interest for the team. Consider, onthe other hand, a group of friends trying to plantheir evening activities. In this group we wouldexpect tasks to be distributed somewhat more onthe basis of the principle of preference. What theexample demonstrates is that the team’senvironment greatly influences the relativeweight of the two principles in the emergence oftask role distribution. In establishing a generalmodel for task role distribution in work groups,we must take into account the team’s embeddingsystem (McGrath, 1984) and its influence on theteam. We will refer to this influence as environ-mental pressure. Environmental pressure is con-sidered high when the flow of resources the teamand/or individual members receive directlydepends on the team’s success or failure in taskaccomplishment. Environmental pressure is con-sidered low when the team’s success or failure intask accomplishment does not influence theresources the team obtains. The higher theenvironmental pressure, the more relative weightwill be placed on the principle of competence vs.on the principle of preference.

Summarizing, we have now demonstrated thecentral ingredients of the theory, with the prin-ciple of competence and the principle of prefer-ence working as the driving mechanisms in thedistribution of task roles in the group. Central-ity of the different task roles determines theimportance of any single task role assignmentfor the functionality of the overall task roledistribution and serves to adjust the relativeweight of each of the two principles for each taskrole. Environmental pressure functions as anoverall moderating variable, adjusting the rela-tive weight of the principle of competence vs.the principle of preference for all task roles.

In order to allow for predictions resulting from

the theory and for the testing of hypotheses, wewill now state the theory in mathematical terms.

A model of group organizationAs we have already stated, the first requirementfor a functional distribution of task roles is thatthe demands of each task role should notexceed the skill level of the group member thatthe task role has been assigned to (principle ofcompetence). Mathematically, the differencebetween the task role demands and the skills ofthe group member completing the task role canbe calculated as a stress value:

Stress M

D S

c

M

Nij ij ij

j

i

1

1

=

- d=

=

!!

_ i

, with (1)

<

for D S

for D S

1 0

0 0ij

ij ij

ij ij

$=

-

-d *

N = number of group members,M = number of task demands and corre-

sponding skills,Dij = demand j of the task role assigned to

member i,Sij = skill j of the group member i,StressC = stress value resulting from compe-

tence mismatch.

The lower StressC the more the task role distri-bution meets the requirement stated in the prin-ciple of competence.

The second requirement for a functionaldistribution of task roles is that each groupmember be assigned to a task role that matcheshis or her preferences (principle of preference).Mathematically speaking, the differencebetween the maximum preference possible for atask role and the preference of the groupmember actually performing the task role canalso be calculated as a stress value:

,maxP P Stressi i

i

N

P

1

- ==

!^ h with (2)

Pmaxi = maximum preference possible forthe task role assigned to member i,


144



145

Pi = preference of group member i for theassigned task role,

StressP = stress value resulting from a prefer-ence mismatch.

The lower this stress value, the more the taskrole distribution meets the requirements statedin the principle of preference.

We have mentioned that in combining thetwo principles, task role centrality must be con-sidered. Centrality will determine the impact ofany single task role assignment for the function-ality of the overall task role distribution. Cen-trality will furthermore serve to adjust therelative weight of the principle of competencevs. the principle of preference for each singletask role. Integrating equation 1 and equation 2with task centrality as a moderating variableresults in the following equation:

M

D S

c P P c c c Stress1max min

ij ij

j

M

i i i i

i

N1

1

i: : :

-

+ - + - ==

=

!!

J

L

KKKKKKK

_

_ _

N

P

OOOOOOO

i

i i ,

(3)with

ci = centrality of the task role assigned tomember i, varying from 0 to 1,

cmin = minimum possible value of c,Stress = stress value.

The fourth determinant of task role distri-bution that has been elaborated is environ-mental pressure. Environmental pressure wasproposed to determine the relative weight of theprinciple of competence vs. the principle ofpreference for all task role assignments. Supple-menting (3) with environmental pressureresults in the following equation:

M

D S

c e P Pmax

ij ij

j

M

i

N

i i1

1

: : :

-

+ -=

=

i

!!

J

L

KKKKKKK

_

_

i

i

(4)

M

D S

c c e c Stress1 1min

ij ij

j

M

i i1

: :

-

+ - - ==

!_

_ _

N

P

OOOOOOO

i

i i

withe = environmental pressure.

Equation 4 is at the core of a computerprogram we have designed. The computerprogram predicts the task role distribution forany group with given task roles and groupmembers. The logic of the program goes asfollows:

For every job in the group, there is (at least)one group member that, according to his or herskills and preferences, is best suited for this par-ticular job. This fact is reflected in equation 4:equation 4 calculates a stress value that will be ata minimum for the best-suited group memberfor each task role. In order to find the mostfunctional distribution of task roles, however,the task roles cannot be considered one by one.To know which group member is best suited fora single job is not sufficient. After all, the samegroup member might be best suited for threedifferent task roles. However, in the group he orshe can only be assigned one task role at a time.This means that the other two task roles have tobe assigned to other group members who maystill be quite well suited for the task role, but whoare not as well suited for this particular task roleas the first group member. On the other hand,there may be group members that are not reallysuited for any of the group’s task roles. However,each group member still has to be assigned to atask role, including members that are not reallysuited for any of the group’s task roles (unlessthe group decides to exclude these members).Each group member has to be assigned exactlyone task role. For this reason, the computerprogram does not try to minimize the stressvalue calculated for every single task role; itrather minimizes the overall group stress valueresulting as the sum of all stress values resultingfrom single task role assignments. In order to dothis, all theoretically possible task role distri-butions are tested by the computer programusing equation 4. For every possible combi-nation of group members and task roles, a stressvalue is being calculated. The combination withthe minimum overall stress value represents themost functional task role distribution. The com-puter program offers the possibility to rank allpossible task role distributions according totheir degree of functionality reflected by theoverall stress value. An empirical task role


distribution can then be judged according to itsdegree of functionality simply by ranking it onthe ranking list of all theoretically possible taskrole distributions.

The number of possible task role distributionsin a given group can be calculated with thefollowing equation:

!

!Or

N

i

i

O

1

=

=

%, with (5)

O = number of possible task role distri-butions,

N = number of group members,o = number of different roles (the same role

might be assigned to several group members atthe same time),

ri = number of group members that the rolei is being assigned to.

Depending on the group size and on thenumber of different task roles in the group, thenumber of possible task role distributions can bethat high that the problem cannot be solvedeven by the fastest computers available today. Inmathematics, the combination problem we arereferring to is known as the ‘traveling salesmanproblem’ (Dakin, 1997).

Numerous approaches are reported in math-ematics that are concerned with the solution oflarge-number traveling salesman-type problems(for an overview, see Dakin, 1997). All of thesealgorithms are content-free, i.e. they can beapplied to any kind of problem as long as theunderlying formal problem structure corre-sponds to the travelling salesman problem.

In order to solve the problem at hand mostelegantly, however, we have developed a differ-ent abbreviating algorithm. In contrast to algo-rithms reported in the mathematical literature(Dakin, 1997; Holland, 1992), this algorithm isnot content-free, but takes into account thespecific content of the group organizationproblem, thus providing a simpler solution toour specific problem. The algorithm asks theuser to define a threshold value for the matchfor a single person-task role fit. The thresholdvalue a priori limits the number of groupmembers considered for a task role, since only

such members are considered whose match-score does not exceed the threshold value. Thisgreatly reduces the number of task role distri-butions that have to be considered by theprogram, since only such task role distributionsare considered in which the stress value for eachperson-task role fit stays within the boundarydefined by the threshold value. The higher thethreshold value, the better must a groupmembers’ skills and preferences match thedemands of a task role in order for this groupmember to even be considered for this task role.Task centrality again is being added as a moder-ating variable: the more central a task role, thebetter must the match between task roledemands and a person’s skills and preferencesbe in order for the program to consider thisperson for the task role. The threshold value canbe adjusted manually, thus allowing the user todecide on the trade-off between calculationtime and calculation accuracy.

Since the abbreviating algorithm is of onlyperipheral interest for the theory presented inthis paper, detailed results of the testing of theabbreviating algorithm will not be reportedhere. Because of the small size of the group wehave observed, it has not been necessary toemploy the abbreviating algorithm in runningthe calculations reported in this paper.

Method

ParticipantsA pilot case study has been conducted with astudent software engineering team. The teamconsisted of 15 college students majoring incomputer science at a college in Dresden,Germany. Students were aged from 19 to 22 andparticipated in a special course of studies thatcombines theoretical education with practicaltraining. They received their practical trainingin an international information technologycompany. With regard to the students’ edu-cation and background, differences betweenthe students were small. All of the students werein the same stage of their education (secondyear of college). The educational path thestudents had chosen required them to attend astandardized set of classes in a fixed sequence.


146


Therefore, all of the students had attended thesame classes and had received the same degreeof formal training. The students were alreadyacquainted with each other, but had neverworked together on a task.

As part of their practical training, the studentswere assigned to a project task (a database appli-cation for the administration of an educationalorganization) that they were to accomplish inteamwork in a three-week period. At the begin-ning of the project, the students received a shortintroduction on project management and team-work in the field of software engineering.Following this introduction, they were asked toorganize themselves, defining task roles andassigning them to group members, working onthe completion of the task and interacting witha (simulated) customer. The students’ perform-ance during the time of project work wasmonitored by two supervisors. The studentsperiodically received feedback on their per-formance by the supervisors and by the simu-lated customer. At the end of the project, thestudents were graded by the supervisors accord-ing to their performance during project work.

Following the introductory course, thestudents themselves defined the task roles theywere going to assign among themselves. Ninedifferent task roles were defined. Seven of these

task roles were later on actually assigned to atleast one group member in the course of projectwork. Each group member was assigned onesingle task role. Table 1 lists the seven task rolesthat were actually assigned in the group, alongwith a short description of each task role (taskrole descriptions were given by the team itselfand have been translated by the authors) andthe number of group members that this specifictask role was assigned to.

Data gatheringDue to the fact that the study was conductedwith a real-world work group under conditionsof a tough timetable imposed by the organi-zation, observations could only be made by asingle observer. For the same reason, simple-to-handle measuring instruments were utilized,mostly rating scales.

Before task roles were assigned in the group,each student was asked to rate his or her ownpreference for each of the seven task roles on a10-point rating scale. Maximum preferencepossible as used in equation 2 was set to 10, themaximum on the 10-point rating scale used toassess the students’ preferences.

In order to describe the demands of the taskroles, we proposed two skills: programmingexpertise and emotional intelligence. The two


147

Table 1. Task role definitions

Assigned to howTask role Description many group members?

Project manager Coordinates efforts of the group members,controls outcomes, manages communicationinside the group and with customers 1

Substitute project manager Same as project manager 1Leader of a programming Coordinates efforts in a programming subteam,

subteam communicates with other subteam leaders andwith the project management 3

Chief programmer Defines central variables, defines and implementsinterfaces between program parts, integrates theprogram modules created by differentprogramming subteams 1

Program designer Develops the graphical user interface, graphicallyintegrates program modules created by the subteams 1

Programmer Works on the program code, documents the code 7Documenter Documents the process of project work 1


skills can be considered a specific application ofthree of the four major areas of skills describedby O’Neil et al. (1997) to the field of softwareengineering, with programming expertisebeing comprised of basic academic skills andhigher order thinking skills and emotionalintelligence referring to interpersonal andteamwork skills.

The students’ programming expertise wasrated on a 10-point rating scale by three of thestudents’ supervisors. With intercorrelations of.67, .83 and .74, interrater reliability can be con-sidered to be satisfactory. Construct validity ofthe ratings can be estimated as high: correlatingthe average of the three supervisors’ ratings withthe grade-point average of the students’ firstyear of college results in a correlation of .75.

In order to measure emotional intelligence,we have constructed a behavior-based measure-ment instrument based on Salovey and Mayer’s(1990) construct that has become famousthrough Goleman’s (1995) best-selling book.We defined four measures that together shouldserve as a measure for emotional intelligence.The four measures are: emotion management,display of positive emotions, empathy and uti-lization of emotions (using one’s emotions tomobilize others). Prototypical behavior wasdescribed for all four measures. Students wereobserved during the course of project work.Their behavior was classified according to thefour measures and rated on all four measures atfour points in time during the course of projectwork, based on actual behavior that has beenrecorded during observation of the group. Anoverall measure ‘EQ’ was then calculated as theaverage of all four measures. Table 2 shows theintercorrelations between the overall measure‘EQ’ for the four points in time. With a medium

intercorrelation between the ratings at differentpoints in time of .84, stability of the ratings canbe considered as satisfactory. As stated above,due to limitations of the case study approachobservations could only be made by oneobserver. Thus, we do not have any informationon the interrater reliability of the EQ measure.We have tested the measure for constructvalidity, however, in order to compensate forlimitations in interrater reliability. We havestated that the EQ score was used as a measurefor the interpersonal and teamwork skills of thestudents. In order to ensure construct validity ofthe EQ score, we obtained sociometric measuresby the students both before and after theproject. Each student was asked to name thethree peers that he or she most preferredworking with and the three peers that he or sheleast preferred working with.

A ‘sympathy score’ and an ‘antipathy score’were calculated simply by counting how manytimes a student has been named by his or her co-students in one of the two categories. If the EQmeasure really measures the interaction skills ofthe students, students with a high EQ scoreshould be more popular as shown in the socio-metric ratings than students with a low EQ score.Correlating the sociometric sympathy score(average of score obtained before and after theproject) with the EQ score results in a corre-lation of .69. The correlation of the antipathy-score with the EQ score is �.82. In addition,before the project three of the students’ super-visors were asked to rate the students accordingto their degree of ‘social competence’. Theaverage of the supervisors’ ratings correlatedwith .76 with the EQ score. Given these results,construct validity of the EQ score we haveemployed can be considered as satisfactory.

Each of the task roles that the studentsdefined was discussed and then rated by experts(two psychologists with computer programmingexperience) on a 10-point-scale (10 being thehighest, 1 being the lowest rating) according tothe degree of programming expertise and thedegree of emotional intelligence required forsuccessful completion of the task role. Differ-ences in the judgment of the two raters weresettled by discussion. Table 3 depicts the results


148

Table 2. Intercorrelations of the EQ measure at fourpoints in time

T1 T2 T3 T4

T1 X X X XT2 .93 X X XT3 .70 .85 X XT4 .82 .88 .87 X


of the expert ratings. We have already men-tioned that task roles were also rated by expertsaccording to their degree of centrality for accom-plishing the group task. The centrality rating foreach task role is also displayed in Table 3.

Unfortunately, we could not obtain a reliablemeasure for environmental pressure. Environ-mental pressure was therefore not taken intoaccount for the testing of the main hypothesismentioned below. The value for environmentalpressure was set to the default value of .5,placing equal weight on both the principle ofcompetence and the principle of preference.Further results on the influence of environ-mental pressure obtained through a number ofsimulation runs will be reported later on in thepaper, however.

HypothesisWe predict that, assuming an average degree ofenvironmental pressure, the group will structureitself in a functional way, i.e. that the stress valuecalculated with equation 4 will be close to theminimum for the empirical task role distribution.

Results

Task role distributionGiven the 15-member group described above,the number of possible task role distributionscan be calculated with equation 6:

!

!! ! ! ! ! ! !

!Or

N1 1 3 1 1 7 1

1543243200

i

i

O

1

: : : : : := = =

=

% (6)

Using equation 4, the program calculates astress value for every possible task role distri-

bution. As a result, the program displays themost functional task role distribution possible,i.e. the task role distribution where the overallstress value calculated with equation 4 is at aminimum. The program also generates aranking list with all possible task role distri-butions being ranked according to their degreeof functionality. The empirically observed taskrole distribution of the real-world team isranked on the ranking list of all possible taskrole distributions by the computer program.The rank of the empirical task role distributioncan be used as a measure for the degree of func-tionality of the empirical task role distribution.

Table 4 depicts the most functional task roledistribution as predicted by the computerprogram and the empirical task role distri-bution. As stated above, calculations were doneassuming an average degree of environmentalpressure by setting e to 0.5. For each groupmember, the table depicts the match betweenthe group member and the task role he or shewas assigned. This match represents the stressvalue calculated with equation 4 for a singlemember/task role-fit. The result has been trans-formed into a percentage value by use of (7):

Match StressStress

100 1max

ii

i

:= -c m, with (7)

Stressi = stress value for member i calculatedwith (4),

Stressmaxi = maximum possible stress value formember–task role match i calculated by settingall skill and preference values of the groupmember to the lowest possible score of 1 andthen calculating (4),

Matchi = percentage of match.


149

Table 3. Task role ratings

Programming Emotional intelligenceTask role expertise required required Centrality

Project manager 6 10 10Substitute project manager 6 10 9Leader of a programming subteam 8 8 7Chief programmer 10 6 10Program designer 4 8 6Programmer 8 2 5Documenter 2 4 1


Table 5 displays the complete path of calcu-lations for the task role distribution of theempirical team, starting with input variables(preferences and skills) and running throughequations 1, 2, 4 and 7. Whenever results of anequation are reported in a column of the table,the number of the corresponding equation isspecified in the column title.

The assignment of 9 out of 15 task roles is pre-dicted correctly by the model. The assignmentof all three of the most central task roles (projectmanager, substitute project manager, chief pro-grammer) is predicted correctly by the model. Itis interesting to note that in contrast to themodel one of the students’ supervisors, whenasked before the beginning of the project whoshe believed the students would appoint asproject manager, did not succeed in predictingthe group’s decision.

As has been shown above, the number oftheoretically possible task role distributions is43,243,200 (43 millions). The position of theempirical task role distribution on the rankinglist of all possible task role distributions is145,655. Dividing the rank of the empirical taskrole distribution by the number of possible taskrole distributions results in a measure for theprobability of the empirical task role distri-

bution or a task role distribution of even higherfunctionality to develop randomly:

p OR e= , with (8)

p = probability measure,Re = rank of the empirical task role distri-

bution on the ranking list of all possible task roledistributions,

O = number of all possible task role distri-butions.

Equation 8 has been used to calculate theprobability of the task role distribution we haveobserved to develop randomly:

.p OR

43243200145655

003e= = =

This measure can be used as a test of signifi-cance for our hypothesis. With a probability of.003, the hypothesis that the self-organizinggroup would structure itself in a functional waycan therefore be accepted on a highly signifi-cant level.

Unfortunately, due to limitations of the casestudy data concerning the actual task perform-ance of the group could not be gathered in a sys-tematic manner. No other student group has


150

Table 4. Most functional task role distribution and empirical task role distribution

Most functional task role distribution Empirical task role distribution

Task role Occupied by Match (%) Task role Occupied by Match (%)

Project manager PERSON1 92 Project manager PERSON1 92Substitute project manager PERSON2 100 Substitute project manager PERSON2 100Subteam leader PERSON3 86 Subteam leader PERSON3 86Subteam leader PERSON4 73 Subteam leader PERSON11 74Subteam leader PERSON5 83 Subteam leader PERSON14 65Chief programmer PERSON6 100 Chief programmer PERSON6 100Program designer PERSON7 81 Program designer PERSON12 89Programmer PERSON8 35 Programmer PERSON8 35Programmer PERSON9 84 Programmer PERSON9 84Programmer PERSON10 61 Programmer PERSON10 61Programmer PERSON11 92 Programmer PERSON4 72Programmer PERSON12 97 Programmer PERSON5 76Programmer PERSON13 84 Programmer PERSON13 84Programmer PERSON14 84 Programmer PERSON7 28Documenter PERSON15 100 Documenter PERSON15 99


Table 5. Set of calculations for the empirical team

Task role Occupied Centrality Preference Program. Program. Eq Eq Environm. Stress Stress Stress Stress Overall Overall Match (7)by exp. requ. exp. required pressure comp. (1) comp. max pref. (2) pref. max stress (4) stress max (%)

Project manager Person1 1 9 6 8 10 9 0.5 0.5 7 1 9 0.30 3.95 92Chief programmer Person6 0.9 10 10 10 6 8 0.5 0 7 0 9 0.00 3.65 100Substitute proj. manag. Person2 1 10 8 10 8 9 0.5 0 7 0 9 0.00 3.95 100Subgroup leader Person3 0.7 7 8 8 8 8 0.5 0 7 3 9 0.42 2.98 86Subgroup leader Person11 0.7 8 8 8 8 4 0.5 2 7 2 9 0.77 2.98 74Subgroup leader Person14 0.7 6 8 8 8 4 0.5 2 7 4 9 1.05 2.98 65Designer Person12 0.6 9 4 7 8 7 0.5 0.5 5 1 9 0.24 2.25 89Programmer Person8 0.5 4 8 3 2 3 0.5 2.5 4 6 9 1.21 1.85 34Programmer Person4 0.5 7 8 7 2 7 0.5 0.5 4 3 9 0.51 1.85 72Programmer Person9 0.5 8 8 9 2 4 0.5 0 4 2 9 0.30 1.85 84Programmer Person10 0.5 6 8 6 2 5 0.5 1 4 4 9 0.73 1.85 61Programmer Person5 0.5 7 8 8 2 5 0.5 0 4 3 9 0.45 1.85 76Programmer Person13 0.5 8 8 8 2 5 0.5 0 4 2 9 0.30 1.85 84Programmer Person7 0.5 2 8 6 2 5 0.5 1 4 8 9 1.33 1.85 28Documenter Person15 0.1 10 2 4 4 3 0.5 0.5 2 0 9 0.00 0.46 99

5 - Stempfle (jk/d) 6

/3/01 10:26 am P

age 151

worked on the same project task, so a compara-tive rating between different groups could notbe obtained. An estimation of the effectivenessof the group is therefore possible only on thebasis of verbal accounts by the students’ super-visors and the simulated customer. All of themagreed in their evaluation of the students’ work,however, in stating that the students’ perform-ance was above average.

The influence of environmental pressureAs has been mentioned before, we were not ableto establish a reliable empirical measure forenvironmental pressure. Since the influence ofenvironmental pressure on task role distributionis of considerable interest, however, we have runa set of calculations with systematic variations inenvironmental pressure. We began calculationsby setting e to 0 (thus predicting the task roledistribution solely on the basis of the principle ofpreference); e was then adjusted in steps of .01until it reached the value of 1 (thus predictingthe task role distribution solely on the basis of theprinciple of competence). For each step, a com-plete set of calculations was executed, generatingresults on the most functional group structureunder the given condition along with the rank ofthe empirical group structure on the ranking listof all possible group structures. This procedureallows for a systematic test of the empirical taskrole distribution against multiple baselines.Employing this procedure, 100 different base-lines have been applied, ranging from a predic-tion based solely on preferences to a predictionbased solely on competencies. The interestingquestion that can be answered through this pro-cedure is whether competencies alone (e = 1) orpreferences alone (e = 0) suffice in order topredict the task role distribution in the observedgroup and, if this should not be the case, whatrelative weight of competencies vs. preferences(represented in e) best predicts the empiricaltask role distribution. This best prediction can berecognized by the fact that the rank of theempirical task role distribution on the rankinglist of all possible task role distributions will be ata minimum. If we assume that a correct esti-mation of e will result in an optimum prediction,this procedure allows for an estimation of the

strength of the environmental pressure theobserved team has experienced.

Figure 2 graphically displays the rank of theempirical task role distribution on the rankinglist of all possible task role distributions depend-ing on changes in e. As displayed in Figure 2, thebest prediction of the empirical task role distri-bution results from baselines with e approach-ing the value of 1, whereas the quality of theprediction decreases with decreasing values of e.However, due to the large scaling of the figure,small differences in the interval from e = .9 toe = 1 are not visible in the figure. In order tofurther clarify the results we have ‘zoomed in’on the critical area around e = .95, runninganother 100 calculations in the interval betweene = .9 and e = 1 in smaller steps of .001. Figure 3graphically displays the results of this calcu-lation. As shown in Figure 3, the quality of theprediction is not at a maximum when only theprinciple of competence is considered (e = 1),but increases when the principle of preferenceis considered with a small weighting. The bestprediction results from a baseline wheree = .952. Under this condition, the rank of theempirical team on the ranking list of all possibleteams is 7063. Given this baseline, the proba-bility, calculated with equation 8, for the empiri-cal team to establish the empirical task roledistribution or a task role distribution of higherfunctionality randomly is:

. .p OR

432432007063

0002e= = = (8)

Team effectivenessBesides explaining the emergence of the taskrole distribution in teams, the theory proposedin this paper may also be used to predict groupeffectiveness. The basic assumption is: the morefunctional a task role distribution according tothe theory, the more effective the group will bein accomplishing its task. Following Morgan,Glickman, Woodward, Blaiwes, and Salas(1986), group effectiveness can be separatedinto two different dimensions: A ‘task-work’dimension that relates to the degree of grouptask accomplishment and a ‘teamwork’ dimen-sion that relates to the quality of communi-cation, coordination and in our view most


152



153

Figure 3. Rank of empirical task role distribution depending on environmental pressure for e ≥ .9 ≤ 1.

Figure 2. Rank of empirical task role distribution depending on environmental pressure.


importantly to the degree of member satisfac-tion in the group (Morgan et al., 1986). Theeffectiveness hypothesis proposed above couldthus be divided into two separate hypotheses:

The more task roles are assigned according toskills (reflected in the task stress value), themore effective the group will be in accomplish-ing the group task. The more task roles areassigned according to preferences (reflected inthe people stress value), the more effective thegroup will be in ensuring high member satisfac-tion. (We do expect some kind of interactionbetween goal accomplishment and membersatisfaction, however.)

The computer program allows for two sepa-rate overall match scores to be calculated: a‘task’ match score resulting from the matchbetween task role demands and member skills,and a ‘member’ match score resulting from thematch between maximum possible task rolepreference and actual preference of groupmembers. The task match score can be calcu-lated with equation 1, the member stress valuecan be calculated with equation 2. Both scoreshave been calculated for the observed team. Thescores have been transformed into a percentagevalue by use of equation 9:

Match StressStress

100 1max

empi := -e o, with (9)

Stressemp = stress value for the empirical taskrole distribution calculated with equation 4,

Stressmax = stress value for the least functionaltask role distribution, i.e. the task role distri-bution where the stress value is at a maximum,

Match = percentage of match.

The task match score is 98 percent, the membermatch score is 89 percent. It seems that theobserved team has placed somewhat greaterweight on task accomplishment than onmember satisfaction. There is some support forthis finding in the sociometric data that havebeen obtained, which clearly reveals that not allgroup members were equally accepted by thegroup. Sociometry reveals several unpopulargroup members who were rejected by a numberof their peers. Several group members weredenied a preferred task role (e.g. one student

wanted to take on the role of project manager,but was assigned the role of a programmer). Onthe other hand, the performance of the team intask accomplishment was rated above average byall of the students’ supervisors.

Competence and preferenceSince we have gathered data on both prefer-ences and competencies of the students, theinterrelation between preferences and compe-tencies is of some interest. It seems reasonableto suspect that group members’ beliefs abouttheir own skills and competencies will have someinfluence on their formation of preferences forcertain task roles. It would be unwise for a groupmember to try to obtain a task role that he or sheknows he or she is not really qualified for. If thegroup member fails in his or her task role, thegroup is likely to react unfavorably with sanc-tions. Therefore, we expect group members toform their preferences partly on the basis oftheir beliefs about their own competencies. Wedo expect that competencies alone will not fullyexplain the formation of preferences, however,as motivational aspects such as individual goalsand the instrumentality of different task rolesfor the attainment of these goals will certainlyexert some influence.

In order to examine the above assumption, wehave correlated the students’ preferences fordifferent task roles with their actual competen-cies for these task roles. By running equation 1for all combinations of students and task roles, aset of ‘match’ scores has been obtained for eachstudent which express how well the student’scompetencies match the demands of each taskrole. Correlating each students’ preferences forthe different task roles with the match scorescalculated by equation 1 results in an averagecorrelation of .59, with a minimum correlationof .01 and a maximum correlation of .98. Formost students, preferences are moderately orhighly correlated with actual competencies,although competencies do not fully explain theformation of preferences. Note, however, that inthe observed team there are some studentswhose preferences do not match their actualcompetencies at all. For two students, prefer-ence and actual competence as expressed by the


154


results of equation 1 are not correlated at all(correlations are .01 and .04). This mismatchcould result for two different reasons: eitherthese students’ perceptions of their own compe-tencies are inadequate, or the students prefertask roles for which they know they are not wellsuited. Since we also gathered data on students’self-perception of their competencies, we canpresent some results in order to clarify the abovequestion. Before the beginning of the project,we asked the students to rate their own compe-tence for each of the task roles on a 10-pointrating scale. Correlating students’ self-perception of competencies with their actualcompetencies as calculated by equation 1 resultsin an average correlation of .49, with aminimum of .00 and a maximum of .95. Thisshows that while for most of the students self-perception of their competencies for differenttask roles do match their competencies as ratedby observers and calculated with equation 1, forsome students this is not the case. Severalstudents expressed self-perceptions of their owncompetencies that did not match observer per-ceptions. Since observer perceptions have beenproven to be highly valid, it can be maintainedthat some students display self-perceptions oftheir own competencies that do not matchreality. It is interesting to note that the studentswho have shown inadequate self-perceptionswere not very popular in the group. Both of thestudents who have shown inadequate self-perceptions have received no nomination onthe sociometric ‘sympathy scale’, but severalnominations on the ‘antipathy scale’ (one ofthese students received 6, the other onereceived 3 nominations on the ‘antipathyscale’). The correctness of the self-perception ofcompetence as expressed by the above corre-lation between self-perception and observer per-ception correlates with sociometric ratings ofsympathy obtained after the project with anaverage of .65. Although a causal interpretationof this finding would be highly speculative fornow, it seems that the group highly values groupmembers who display adequate self-perceptionsof their competencies.

One could suspect at this point that the factthat some students obviously hold inadequate

perceptions about their own competencies fullyexplains why these students’ preferences do notmatch their actual competencies as rated byobservers. If this were the case, students’ prefer-ences should not be correlated with actual com-petencies as perceived by observers, but shouldbe highly correlated with self-perceived compe-tencies. This is only partly the case, however.Preferences for different task roles and self-perceived competencies for the task roles arecorrelated with .68. Although self-perceptionsof competence better explain the formation ofpreferences than observer perceptions of com-petence, there still is some variance whichcannot be accounted for by inadequate self-perceptions.

Recapitulating the above points, in the for-mation of preferences for different task rolesself-perceived competencies seem to play animportant role. The adequacy of self-percep-tions can greatly differ between individuals. Theformation of preferences cannot be explainedby self-perceived competencies alone, however.

Discussion

The theory proposed in this paper was devel-oped to answer a single question: ‘How will workgroups structure themselves in order to ‘survive’by successfully solving the problems that theyare being presented with?’ By pointing out twobasic principles, the principle of competenceand the principle of preference, we have pro-posed two basic demands that any work groupmust meet in order to survive: the principle ofcompetence requires the group to structureitself around a group task as an external require-ment, leading the group to assign task roles to itsmembers according to their individual skills.The principle of preference requires the groupto take into account individual preferencesresulting from individual needs as internalrequirements. These two demands reflect twobasic dimensions that have been proposed bydifferent researchers under different labels(task-performance pattern vs. interpersonalrelations pattern, concern for task vs. concernfor people, initiating structure vs. consideration,etc.) to be the basic components of group


155


interaction (McGrath, 1984; McGrath et al.,1999; Moreland & Levine, 1982) and the centralingredients of successful leadership in smallgroups (e.g. Fleishman, 1973; Katz, Macoby, &Morse, 1950). Thus, the functional theory oftask role distribution we propose seems to con-verge with a large body of past research. Thetheory transcends this research in a particulardirection, however, specifying exactly how thesetwo dimensions interact in the emergence of atask role distribution.

The study we have described has been con-ducted in a complex real-world environment.Although ecological validity can thus be con-sidered high, the results have to be regardedwith caution, given the apparent lack ofmethodological vigor in the study. The pro-posed skills and the measurement methods(ratings) have not been subject to prior in-depthanalysis. We would further like to point out thatthe results obviously cannot be regarded as arigorous test of the theory, as the study has beenconducted with a single group. Thus, we do notclaim that we have obtained conclusive evidencein favor of the theory. Further research must beconducted in controlled laboratory environ-ments in order to complement the results fromthe case study.

The calculations that have been reported fordifferent baselines with variations in environ-mental pressure show that adjusting the relativeweight of the principle of competence vs. theprinciple of preference has a substantial influ-ence on the predictive power of the model.Although we succeed quite well in predictingthe empirical task role distribution on the basisof an equal weighting of the principle of com-petence vs. the principle of preference, thequality of the prediction increases when theprinciple of competence is weighted substan-tially higher than the principle of preference. Inthe team we have observed, competenciesexplain a lot more of the team’s task role distri-bution than do preferences. The fact thatweighting the principle of competence higherthan the principle of preference results in abetter predictive power of the model for theobserved team should not be generalized at thispoint, however, since we could not yet compare

results between different teams. Our theoreticalassumptions suggest that the high weighting oncompetence in the observed team can beexplained by the fact that environmental pres-sure in the observed team was high. Furtherstudies must clarify whether different environ-mental conditions will actually lead to differ-ences in the strength of the influence ofcompetence vs. preference and whether inteams with low environmental pressure the pre-dictive power of the model actually increaseswith higher weights on the principle of prefer-ence. At this point we cannot clarify whether astronger weighting of the principle of compe-tence vs. the principle of preference is a generalrule that can be applied to all work teams orwhether the weight of the principle of compe-tence vs. the principle of preference is actuallydependent on environmental pressure, as sug-gested in the theory.

One important limitation of the theory shallbe discussed at this point. This limitation refersto the fact that the theory presented in thispaper is a ‘static’ theory. The theory allows forpredicting the end-product of a complex groupdecision-making process, but does not explainthe process itself. How are groups able to solve aproblem mathematically so complex that it takesa computer millions of calculations? From anevolutionary perspective, it seems logical andnecessary that nature has equipped our specieswith great ability to form highly functioningteams quickly. Humans are social beings, andnature cannot afford to produce a social beingwithout equipping it with the skills necessary tomanage task-focused social interaction. But howdo groups actually go about distributing taskroles in the group? As for now, our data do notprovide a conclusive answer to this question.Further research needs to be conducted, focus-ing not on the end-product, but on the processof task role distribution. Our unsystematicobservations of the observed team suggest,however, that the task role distribution in theteam has been established dynamically. Aninitial task role distribution was establishedmainly on the basis of preferences on the firstday of project work. Team members were simplyasked to sign up for a specific subgroup and pick


156


a task role. Task roles have been repeatedlyreassigned later on in the process, however, as areaction to problems and deficits in the team.For example, the role of chief programmer hadnot been assigned at all in the beginning ofproject work. Later on the group realized,however, that the lack of an overall programarchitecture and the lack of clearly definedinterfaces between different program partscreated major problems. In response to theseproblems, one of the group members took onthe role of chief programmer. The final task roledistribution in the observed group was notestablished until the third day of project work.

This points out the importance of a groupremaining flexible in establishing a functionaltask role distribution. Flexibility may berequired for continuous adaptation not onlyduring group formation, but also in laterperiods of group work, particularly in times ofchange and crisis. Since the task role distri-bution highly depends on the nature of thegroup task, the task role distribution may needto change dynamically over time as the grouptask changes. New task roles may need to bedefined according to the demands of a changedsituation, existing task roles may need to be reas-signed to different group members. Skills andpreferences of the group members are alsolikely to change over time, requiring the groupto continuously reorganize itself. When apply-ing the theory to long-term group development,the dynamic nature of groups interacting withan often dynamic environment must be takeninto account. A group’s flexibility even in laterperiods of the group’s existence may determinethe group’s effectiveness to a high degree.

In most organizations, self-organizing teamsare still rare. Task roles, especially leadershiproles, are mostly being assigned to groupmembers by superiors who oftentimes do nothave enough information either about the skillsor about the preferences of group members. Forthis reason, the decisions taken often result in asuboptimum task role distribution, with aninformal task role distribution developing as thegroup tries to correct deficits resulting from dys-functional task role assignments. Imposing adysfunctional task role distribution on a group

in terms of our theory adds to the group ‘stressvalue’, making it more difficult for the group toachieve its goals. It is our conviction that self-organizing teams bear a high potential for on-going success in organizations with improvingthe well-being of the group members at thesame time. Research conducted in Europe onself-organizing teams (Alioth & Ulich, 1981;Emery & Thorsrud, 1982; Ulich, 1991) providesresults indicating that self-organizing teams aremore effective in accomplishing their task. Pro-viding teams with the opportunity to organizethemselves also helps to enhance personal andprofessional qualifications of the team members(Ulich, 1991; Weber, 1997). Implementing self-organization in teams requires careful prep-aration, however. Hackman (1990) points out anumber of prerequisites for successfully imple-menting self-organization in organizations (e.g.establishing team rewards instead of individualrewards in the team).These prerequisites shouldbe closely considered before self-organizingteams are established in organizational practice.

AcknowledgmentThe computer program that has been developed inconjunction with this paper can be obtained fromthe authors.

ReferencesAlioth, A., & Ulich, E. (1981). Gruppenarbeit und

Mitbestimmung [Teamwork and participation]. InF. Stoll (Ed.), Die Psychologie des 20. Jahrhunderts(pp. 863–885). Zürich: Kindler.

Argyle, M. (1983). Five kinds of small social group. InH. H. Blumberg, A. P. Hare, V. Kent, & M. F.Davies (Eds.), Small groups and social interaction(Vol. I). Chichester: Wiley.

Barker, R. D. (1968). Ecological psychology. Stanford,CA: Stanford University Press.

Barkow, J. H., Cosmides, L., & Tooby, J. (Eds.).(1992). The adapted mind: Evolutionary psychologyand the generation of culture. Oxford: OxfordUniversity Press.

Barrick, M. R., Stewart, G. L., Neubert, M. J., &Mount, M. K. (1998). Relating member ability andpersonality to work-team processes and teameffectiveness. Journal of Applied Psychology, 83,377–391.


157


Beach, L. R. (1993). Four revolutions in behavioraldecision theory. In M. M. Chemers & R. Ayman(Eds.), Leadership theory and research. Perspectives anddirections. San Diego, CA. Academic Press.

Bischof, N. (1985). Das Rätsel Ödipus. Die biologischenWurzeln des Urkonflikts von Intimität und Autonomie[The mystery of Oedipus. The biological roots ofthe basic conflict between intimacy andautonomy]. München: Piper.

Blumberg, H. H., Hare, A. P., Kent, V., & Davies, M.F. (Eds.). (1983). Small groups and social interaction(Vol. I). Chichester: Wiley.

Cranach, M. v. (1993). Toward a theory of the actinggroup. In E. H. Witte & J. H. Davis (Eds.),Understanding group behavior. Small group processes andinterpersonal relations (Vol. 2). Mahwah, NJ: Erlbaum.

Dakin, R. (1997). The travelling salesman problem.Introduction [On-line]. Available:http://www.pcug.org.au/~dakin/tsp.htm.

Dawkins, R. (1990). The selfish gene. Oxford: OxfordUniversity Press.

Dawkins, R. (1995). River out of Eden: A Darwinianview of life. London: Weidenfeld & Nicolson.

Dennett, D. C. (1995). Darwin’s dangerous idea:Evolution and the meanings of life. New York: Simon& Schuster.

Dörner, D. (1999). Blauplan für eine Seele [A blueprintof the mind]. Reinbek: Rowohlt.

Emery, F. E. (1978). Characteristics of socio-technical systems. In F. E. Emery (Ed.), Theemergence of a new paradigm of work (pp. 38–86).Canberra: Australian National University, Centerfor Continuing Education.

Emery, F. E., & Thorsrud, E. (1982). IndustrielleDemokratie [Industrial democracy]. In E. Ulich(Ed.), Schriften zur Arbeitspsychologie (Vol. 25). Bern:Huber.

Fleishman, E. A. (1973). Twenty years ofconsideration and structure. In E. A. Fleishman &J. G. Hunt (Eds.), Current developments in the study ofleadership. Carbondale, IL: Southern IllinoisUniversity Press.

French, J. R. P., & Raven, B. H. (1959). The bases ofsocial power. In D. Cartwright (Ed.), Studies insocial power (pp. 607–623). Ann Arbor, MI:University of Michigan Press.

Gergen, K. J., Greenberg, M. S., & Willis, R. H.(Eds.). (1980). Social exchange: Advances in theoryand research. New York: Plenum Press.

Gladstein, D. L. (1984). Groups in context: A modelof task group effectiveness. Administrative ScienceQuarterly, 29, 499–517.

Goleman, D. (1995). Emotional intelligence. New York:Bantam Books.

Hackman, J. R. (1968). Effects of task characteristics

on group products. Journal of Experimental SocialPsychology, 4, 162–187.

Hackman, J. R. (1983). A normative model of work teameffectiveness. (Technical Report No. 2. ResearchProgram on Group Effectiveness). Yale School ofOrganization and Management, New Haven, CT.

Hackman, J. R. (Ed.). (1990). Groups that work (andthose that don’t). San Francisco, CA: Jossey-Bass.

Haken, H. (1987). Die Selbstorganisation derInformation in biologischen Systemen aus der Sichtder Synergetik [Self-organization of information inbiological systems from the viewpoint of the theoryof synergetics]. In B.-O. Küppers (Ed.), Ordnung ausdem Chaos. München: Piper.

Haken, H. (1995). Erfolgsgeheimnisse der Natur.Synergetik: Die Lehre vom Zusammenwirken [Secrets ofnature’s success. Synergetics: the theory ofcombined effect]. Reinbek: Rowohlt.

Helmreich, R. L., & Foushee, H. C. (1993). Why crewresource management? Empirical and theoreticalbases of human factors training in aviation. In E.L. Wiener, B. G. Kanki, & R. L. Helmreich (Eds.),Cockpit resource management. San Diego, CA:Academic Press.

Helmreich, R. L., & Schaefer, H.-G. (1994). Teamperformance in the operating room. In M. S.Bogner (Ed.), Human error in medicine. Hillsdale,NJ: Erlbaum.

Herzberg, F., Mausner, B., & Snyderman, B. (1959).The motivation to work. New York: Wiley.

Holland, J. H. (1992). Adaptation in natural andartificial systems: An introductory analysis withapplications to biology, control, and artificialintelligence. Cambridge, MA: MIT Press.

Ilgen, D. R. (1999). Teams embedded inorganizations. Some implications. AmericanPsychologist, 54, 129–139.

Katz, D., Macoby, N., & Morse, N. C. (1950).Productivity, supervision and morale in an officesituation. Ann Arbor, MI: University of Michigan,Institute for Social Research.

Krech, D., Crutchfield, R., & Ballachey, E. G. (1962).Individual in society. New York: McGraw-Hill.

Küppers, B.-O. (Ed.). (1987). Ordnung aus dem Chaos[Order out of chaos]. München: Piper.

Laughlin, P. R. (1980). Social combination processesof cooperative problem-solving groups as verbalintellective tasks. In M. Fishbein (Ed.), Progress insocial psychology (Vol. 1). Hillsdale, NJ: Erlbaum.

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

McGrath, J. E., Arrow, H., & Berdahl, J. L. (1999).Cooperation and conflict as manifestations ofcoordination in small groups. Polish PsychologicalBulletin, 30, 15–45.


158



159

Moreland, R. L., & Levine, J. M. (1982). Socializationin small groups: Temporal changes in individual–group relations. In L. Berkowitz, (Ed.), Advances inexperimental social psychology (pp. 137–192). NewYork: Academic Press.

Moreland, R. L., Levine, J. M., & Cini, M. A. (1993).Group socialization: The role of commitment. InM. A. Hogg & D. Abrams (Eds.), Group motivation.Social psychological perspectives. New York: HarvesterWheatsheaf.

Moreland, R. L., Levine, J. M., & Wingert, M. L.(1996). Creating the ideal group: Compositioneffects at work. In E. H. Witte & J. H. Davis (Eds.),Understanding group behavior. Small group processes andinterpersonal relations (Vol. 2). Mahwah, NJ: Erlbaum.

Morgan, B. B., Jr., Glickman, A. S., Woodward, E. A.,Blaiwes, A., & Salas, E. (1986). Measurement of teambehaviors in a Navy environment (NTSC Report No.86–014). Orlando, FL: Naval Training SystemCenter.

O’Neil, H. F., Jr., Allred, K., & Baker, E. L. (1997).Review of workforce readiness. Theoreticalframeworks. In H. F. O’Neil, Jr. (Ed.), Workforcereadiness. Competencies and assessment. Mahwah, NJ:Erlbaum.

Prigogine, I. (1985). Vom Sein zum Werden [Frombeing to becoming]. München. Piper.

Rosenstiel, L. v. (1989). Selektions- undSozialisationseffekte beim Übergang vomBildungs- ins Beschäftigungssystem [Selection- andsocialization-effects in the transference phase fromthe educational to the occupational system].Zeitschrift für Arbeits- und Organisationspsychologie, 33,21–32.

Rosenstiel, L. v., Molt, W., & Rüttinger, B. (1995).Organisationspsychologie [Organizationalpsychology] (8th ed.). Stuttgart: Kohlhammer.

Salovey, P., & Mayer, J. D. (1990). Emotionalintelligence. Imagination, cognition and personality, 9,185–211.

Schuster, P. (1987). Molekulare Evolution undUrsprung des Lebens [Molecular evolution andthe origin of life]. In B.-O. Küppers (Ed.), Ordnungaus dem Chaos. München: Piper.

Shaw, M. E. (1973). Scaling group tasks: A methodfor dimensional analysis. JSAS Catalog of SelectedDocuments in Psychology, 3, 8.

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

Stempfle, J. (1998a). Anatomie einer Arbeitsgruppe.Entwurf eines allgemeinen psychologischen Modells einerArbeitsgruppe und empirische Untersuchung einerProjektgruppe im Bereich der Softwareentwicklung[Anatomy of a work group. A generalpsychological model of work groups and an

empirical study of a project group in softwaredevelopment]. Unpublished manuscript,University of Bamberg/Germany.

Stempfle, J. (1998b). Teamarbeit erfolgreich gestalten[Implementing teamwork successfully].Unpublished manuscript, University ofBamberg/Germany.

Stempfle, J. (1999). Herr der Fliegen oder wieArbeitsgruppen sich organisieren [Lord of theflies or how work groups structure themselves]. InG. Krampen, H. Zayer, W. Schönpflug, & G.Richardt (Eds.), Beiträge zur AngewandtenPsychologie. Proceedings of the 5. DeutscherPsychologentag, October 7–10, (pp. 49–53). Bonn:Deutscher Psychologen Verlag.

Stewart, D. D., & Stasser, G. (1995). Expert roleassignment and information sampling duringcollective recall and decision making. Journal ofPersonality and Social Psychology, 69, 619–628.

Thibaut, J. W., & Kelley, H. H. (1959). The socialpsychology of groups. New York: Wiley.

Ulich, E. (1991). Arbeitspsychologie [Work psychology].Zürich: Verlag der Fachvereine.

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

Walster, E., Walster, G. W., & Berscheid, E. (1978).Equity: Theory and research. Boston, MA: Allyn &Bacon.

Weber, W. G. (1997). Analyse von Gruppenarbeit:Kollektive Handlungsregulation in soziotechnischenSystemen [Analyzing group work: Collective actionregulation in socio-technical systems]. Bern: Huber.

Paper received 9 February 2000; revised version accepted17 October 2000.

Biographical notesjoachim stempfle is a faculty member of the

research staff at the Institut für TheoretischePsychologie at the University ofBamberg/Germany. His research interests are inthe areas of team and leadership research.

oliver hübner is a freelance software developer.His research interests are in the area of computersimulations of psychological processes.

dr. petra badke-schaub is a faculty member ofthe research and lecturer staff at the Institute ofTheoretical Psychology at the University ofBamberg/Germany. She also heads a researchproject about designing in teams. Her researcharea comprises critical situations and complexproblem solving in groups.


A Functional Theory of Task Role Distribution in Work Groups

Documents

distribution of task

project task

questions of group structure

study of work groups

task forces

group composition

selforganizing work

small groups