25
Leading Workers to Lead Themselves: The External Leadership of Self- Managing Work Teams Author(s): Charles C. Manz and Henry P. Sims, Jr. Source: Administrative Science Quarterly, Vol. 32, No. 1 (Mar., 1987), pp. 106-129 Published by: Sage Publications, Inc. on behalf of the Johnson Graduate School of Management, Cornell University Stable URL: http://www.jstor.org/stable/2392745 . Accessed: 12/06/2014 14:54 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Sage Publications, Inc. and Johnson Graduate School of Management, Cornell University are collaborating with JSTOR to digitize, preserve and extend access to Administrative Science Quarterly. http://www.jstor.org This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PM All use subject to JSTOR Terms and Conditions

Leading Workers to Lead Themselves: The External Leadership of Self- Managing Work Teams

  • Upload
    jr

  • View
    213

  • Download
    1

Embed Size (px)

Citation preview

Leading Workers to Lead Themselves: The External Leadership of Self- Managing Work TeamsAuthor(s): Charles C. Manz and Henry P. Sims, Jr.Source: Administrative Science Quarterly, Vol. 32, No. 1 (Mar., 1987), pp. 106-129Published by: Sage Publications, Inc. on behalf of the Johnson Graduate School of Management,Cornell UniversityStable URL: http://www.jstor.org/stable/2392745 .

Accessed: 12/06/2014 14:54

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Sage Publications, Inc. and Johnson Graduate School of Management, Cornell University are collaboratingwith JSTOR to digitize, preserve and extend access to Administrative Science Quarterly.

http://www.jstor.org

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Leading Workers to Lead Themselves: The Exter- nal Leadership of Self- Managing Work Teams

Charles C. Manz University of Minnesota Henry P. Sims, Jr. The Pennsylvania State University and George Mason University

(P 1987 by Cornell University. 0001-8392/87/3201-01 06/$1 .00.

We gratefully acknowledge the encourage- ment and support of Dr. Richard Cherry.

This paper explores the paradoxical role of the external leaders of self-managing workteams. Observation, inter- views, group elicitations, and a literature search were used to identify salient leader behaviors in a medium-sized man- ufacturing plant that had been operating for several years under a system of self-managing work teams. A self- management leadership questionnaire was developed to measure the 21 leader behaviors identified. Correlations with overall leadership-effectiveness ratings generally indi- cated that the external leaders' most important behaviors are those that facilitate the team's self-management through self-observation, self-evaluation, and self- reinforcement. The study suggests that there is a legitimate role for external leaders of self-managing work teams but that it differs from traditional and participative leadership roles.

According to an ancient Chinese saying, "The best of all leaders is the one who helps people so that, eventually, they don't need him" (Lao Tzu, taken from Maccoby, 1981). This state- ment seems to capture the idealistic essence of what external leadership should be like in contemporary self-managing work teams. Yet, some evidence seems to suggest that reality often falls short of the ideal. Indeed, leaders of these teams are a fact of life in real organizations and are considered necessary and functional by many (Lawler, 1986; Hackman, 1986; Manz and Sims, 1986). If self-managing teams are truly self-managing, then why should an external leader be required?

SELF-MANAGING TEAMS

Modern organizations in the U.S. face numerous problems, such as decreasing quantity and quality of production, worker dissatisfaction, high levels of turnover and absenteeism, and counterproductive employee behavior (Cummings and Malloy, 1977). New approaches are required to deal with the increasing interdependence, complexity, and uncertainty in the environ- ments of these organizations (Trist, 1977).

One particularly interesting approach to meeting these chal- lenges is the use of autonomous or self-managed work groups. Hackman has suggested that self-managing work groups usu- ally include "a relatively whole task; members who each pos- sess a variety of skills relevant to the group task; workers' discretion over such decisions as methods of work, task sched- ules, and assignment of members to different tasks; and com- pensation and feedback about performance for the group as a whole" (quoted in Cummings, 1978: 625). The use of self- managing groups involves a shift in focus from individual methods of performing work to group methods. The rationale for making this shift has been described as resulting from "the proposition that a group can more effectively allocate its re- sources when and where required to deal with its total variance in work conditions, than can an aggregate of individuals each of whom is assigned part of the variance" (Susman, 1976: 183). When employees become members of a self-managing group, they tend to define their work roles in terms of their value as contributors to the group's primary task rather than in relation to one specific job.

According to one estimate (Edward Lawler, informal communi-

1 06/Administrative Science Quarterly, 32 (1987): 1 06-1 28

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

cation), two to three hundred manufacturing plants in the United States seem to be using some derivative of a highly par- ticipative team approach. In addition, there are other, non- manufacturing organizations that rely on some variation of this approach, like the insurance firm studied by Manz and Angle (1986). These groups, called autonomous or self-managed work groups, are work innovations mainly characterized by the attempt to create a high degree of decision-making autonomy and behavioral control at the work-group level. Consequently, a much greater emphasis is placed on control within rather than from outside the group. There is some debate whether these teams are established to improve productivity or simply em- ployee quality of worklife, but it seems clear that management decision makers who take this approach at least have implicit goals of improved productivity, better quality, and/or reduced conflict.

Academics have had limited access to organizations that use the team approach, and sometimes when they have been given access it is with the proviso that there be no publicity or writing about it. With the exception of work by Walton (1977), Trist, Susman, and Brown (1977), Poza and Markus (1980), and some other projects reviewed by Cummings and Malloy (1977), relatively little scholarly work has been published about self-managing teams.

We were fortunate, for this study, to have open access to a small-parts manufacturing plant organized according to a self- managing-team work system that had been in place for several years. Our investigation focused mainly on the role of the exter- nal leader of these work groups. The main objective was to identify relevant leader behaviors, develop measures of these behaviors in this situation, and collect preliminary data on how these leader behaviors related to leader-effectiveness ratings.

External Leadership of Self-Managing Teams

One particularly puzzling aspect of studying a work system de- signed around self-managing teams is identifying the appropri- ate role and behaviors for external leaders of the group (Lawler, 1986; Manz and Sims, 1986). As one individual said to us, "If these teams are supposed to lead themselves, then what is a coordinator [external leader] supposed to do?" In other set- tings, terms such as "consultant," "facilitator," "communica- tor," and even "unleader" (Manz and Sims, 1984) have been used. In this paper, we use the term "coordinator" to indicate the external leader, which is the term used at the research site. Coordinators occupy a position in the hierarchy that would be analogous to a foreman or general foreman in a more traditional plant, although expected behaviors are quite different. The con- fusion surrounding the role of coordinator stems from the fact that he or she has responsibility for a team that is theoretically designed to be self-managing. What is the appropriate role for a coordinator when a group is supposed to lead itself?

While the philosophy of participation is relatively well-known, the question of what leaders actually do in fully committed par- ticipative situations remains obscure (Strauss, 1982; Lawler, 1986). To aid us in interpreting what we found in this system, we used several theoretical perspectives. First and foremost, we drew from psychologically based self-management theory

107/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

(Thoresen and Mahoney, 1974; Bandura, 1977; Mahoney and Arnkoff, 1978, 1979; Manz and Sims, 1980; Andrasik and Heimberg, 1982; Manz, 1986). In addition, other relevant per- spectives, especially sociotechnical systems theory (Emery and Trist, 1969; Susman, 1976; Cummings, 1978), leadership theory (Yukl and Nemeroff, 1979; Yukl, 1981), and reinforce- ment and goal theory (Luthans and Kreitner, 1975; Sims, 1977; Locke et al., 1981), were included to help develop a more com- prehensive set of leader behaviors as well as to help assess the unique effect of the self-management leader behaviors. Per- haps the most important aspect of this study, however, is that it reports direct observations of self-managing-team leader- ship. This qualitative data, while first serving as an important set of results in themselves, were also subsequently used to help develop and validate a self-management leadership ques- tionnaire. As Mills (1983: 45) pointed out, "It seems clear that self-management in organizations needs to be studied as a more complex concept than has been done to date." The multi- ple methods and perspectives relied on in this paper, together with the focus on leading self-managing groups, is consistent with Mills's challenge,

The primary purpose of this research was, first, to identify spe- cific self-management-team leader behaviors, specifically be- haviors that facilitate effective employee self-management, and, second, to develop preliminary measures of these be- haviors. The overall research question prior to observing the leaders in situ was: What behaviors are used by leaders (i.e., coordinators) within the self-managing work group system? Or in other words, what do they do and how does their behavior compare with what we know about leadership behavior in other contexts?

Secondarily we were interested in the question: What is the re- lationship between specific leader behaviors (what coordina- tors do) and overall leader effectiveness? We expected that leader behavior that encourages constructive employee self- management behaviors would be positively related to overall effectiveness ratings of coordinators.

METHOD

Procedure

The study had a qualitative phase and a quantitative phase, each consisting of distinct parts or steps, and was designed to provide multiple viewpoints, a triangulation view (Jick, 1979) of the leadership of self-managing teams. Extensive direct, in situ observation, interviews, group-elicitation, literature search, performance-rating procedures, and questionnaire administra- tion were combined to draw on the pragmatic viewpoints of those in the plant. These views were then interpreted with the aid of existing theory.

The qualitative phase addressed the question, "What do coor- dinators actually do?" and was divided into three parts. In part 1 an initial visit to the plant was made, to observe and develop preliminary ideas about the work system and the self- management-team leader behaviors. Multiple interviews and discussions were conducted with employees at all levels. Ex- tensive observations were made of the coordinators while they worked.

1 08/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

In part 2, between the first and second visits, information gathered during the first visit was used for preliminary identi- fication of relevant leader-behavior variables in this situation, and a literature search was conducted to identify relevant the- oretical perspectives and measurement approaches. The major theoretical perspectives were self-management theory, so- ciotechnical systems theory (STS), and the leadership, rein- forcement, and goal-setting theories. Especially for purposes of measurement, the prior work of Yukl and Nemeroff (1979) provided useful insight in developing these variables. A parallel research project by Gioia and Sims (1985) suggested measur- ing very specific behavioral items to reduce social construction effects. At the end of part 2, we had developed a preliminary categorization system of self-management-team leader behaviors.

In part 3, during the early part of the second visit, we conducted further interviews and made observations to reinterpret and qualitatively validate the preliminary leader-behavior variables. We especially looked for behavioral examples of the leader be- haviors to aid us in constructing a questionnaire. A group- elicitation procedure (Manz and Sims, 1984) was also used, specifically to draw out ideas about team leadership from em- ployees at all levels in the plant. This elicitation method, a varia- tion of the Delbecq, Van de Ven, and Gustafson (1975) nominal group method, asked employees what coordinators actually did. During this period, our preliminary judgment of the saliency of the leader behaviors was confirmed, some additional leader behaviors were identified, and we listed all the leader behaviors identified.

The quantitative phase focused on the relationship between the leader behaviors identified in the qualitative phase and leader effectiveness. A questionnaire, the Self-Management Leadership Questionnaire (SMLQ) (available upon request), was completed, typed, and reproduced on site. The specific items for the self-management variables are shown in Appendix A.

The questionnaire was administered to 276 of approximately 300 hourly employees, which included elected internal team leaders. The respondents were 92.7 percent male and 71.4 percent white, 26.7 percent black, and 1.9 percent other ethnic minorities. The mean number of years the employees had been employed with the organization was 4.4. The mean age of the respondents was 31.6 years old. The mean level of education was a high school diploma and some craft, technical, or college training, but without an associate or higher-level college de- gree. Questionnaires were administered to groups of 25 to 35 employees in an on-site conference room during regularly scheduled working hours. Employees were asked to respond to the questionnaire with their coordinator's leader behavior in mind. In addition, in a separate response column, they were asked to respond to each item with the behavior of their inter- nal elected team leader in mind. Data about internal team leaders were gathered as a secondary focus to provide a supplementary source of leadership information.

Overall Leadership Effectiveness

In this phase we also collected the coordinator overa II- effectiveness measures.

109/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

In the research literature, the concept of leadership effective- ness has differed from one study to another depending on the kinds of outcomes selected as criteria: e.g., subordinates' sat- isfaction with the leader, subordinates' commitment to group goals, group survival, etc. (Yukl, 1981). One commonly used measure of a leader's effectiveness is how well the leader's group performs its work and achieves its goals. But as Cart- wright and Zander (1963) indicated, a belief that groups can be made effective simply by providing good leaders is a rather na- ive view, given research findings. Indeed, the work environ- ment and the interaction among group members themselves, as well as with their leader, are important influences on group success.

A wide range of measures of leader effectiveness has been employed in past research. These include objective measures, such as profit growth, market share, and sales increase, and more subjective measures such as ratings of the group's suc- cess and ratings of the leader's effectiveness in performing his or her leadership role (Yukl, 1981). In this study we relied on the latter measure of a leader's effectiveness -that is, subjective ratings of the leader's effectiveness as a leader, collected from multiple sources within the plant.

We focused on perceptions of overall leader effectiveness especially employee and group aggregate views of the leader's level of effectiveness in performing the coordinator role. Our in- tention was to triangulate on the concept of effectiveness through those having a significant history in the plant. Thus, in- ternal team leaders in particular, whom we observed interacted the most frequently of all team members with coordinators, and regular team members, who also had significant but less contact, are particularly suited for this purpose.

Two primary measures were collected. First, the team mem- ber's evaluation of the coordinator consisted of four Likert- scale items on the questionnaire (e.g., "His/her overall effec- tiveness is very high"). The seven-point response scale ranged from "definitely not true" to "definitely true." The Cronbach al- pha for this scale was .92. For later analysis, this evaluation measure was separated into the employee's evaluation of the coordinator (not including team-leader responses) and the team leader's evaluation of the coordinator, depending on the source of the rating.

A second measure, support-group evaluation of the coordinator (superior ratings), was collected in a special ranking and rating procedure. Two types of measures were used: in the first, each coordinator was rated on a scale from 1 (marginal effec- tiveness) to 8 (excellent). In the second, all coordinators were ranked from "most effective" to "least effective." Six raters were selected from the management group, based on their awareness and closeness to production operations in the plant. The use of both ratings and rankings might be considered somewhat analogous to a multitrait/multirater procedure. The Pearson zero-order correlations between ratings and rankings obtained from the six support-group members were .94 (p < .001). Furthermore, interrater reliabilities were estimated by calculating Cronbach alpha coefficients, essentially by treat- ing raters as questions making up a scale and ratees (coordina- tbrs) as cases (Cronbach, 1951; Brown, 1976). The reliability of

1 1O/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

the ratings was .92 and the rankings .89 by this method. Since the ratings and rankings were so highly correlated, the final rat- ing score from the support group was calculated by averaging the standardized ratings and the standardized rankings for each coordinator.

In addition, a nonstatistical analysis of the relationships be- tween these three ratings was done by aggregating employee evaluations by coordinator, aggregating team-leader evaluations by coordinator, and computing Spearman rank- order correlations with a sample size of 10 (too small for statis- tical conclusions). These interrating correlations were .50, .50, and .58, providing some qualitative indications of convergence between the three rating sources.

In this portion of the analysis, the leader-behavior variables were related to the leader-effectiveness evaluations. A key is- sue that needed to be addressed was whether to focus the analysis at the individual level or at the aggregate (group) level. While a theoretical stimulus for interpreting this research came from an individual-level perspective (the self-management literature), the work system existed at the group level. Thus, preliminary analyses were performed to determine whether further analysis should be conducted at the group level or the individual level. This analysis is summarized in Appendix B. The main conclusion was that further analysis should mainly pro- ceed at the individual level.

The questionnaire results were analyzed with factor analysis. Qualitative and statistical results were later discussed with cor- porate personnel for purposes of feedback and to improve in- terpretation of the results.

Description of the Team System

The nonunionized small-parts plant is located in the southern part of the United States. It is owned by a large corporation and employs approximately 320 employees. The plant has been in existence since the early 1970s and was organized from the very beginning according to a self-managing team concept. The technology used in the plant can be described as small-parts production or assembly and it is generally sequentially interde- pendent. Each work group is assigned to a system of closely related tasks, many of which are small assembly-line operations. Teams are generally distinct from each other, both physically and by in-process inventory buffers.

The organizational structure has three distinct hierarchical levels. Upper plant management (called the support team) is the level at which many traditional plant-management responsibilities are handled (e.g., planning overall plant produc- tion schedules, dealing with client firms, etc.). The support team is also formally responsible for the supervision of coordi- nators. The support team generally plays a supportive rather than directive role in the plant's operation and attempts to oper- ate as a team. The work-team coordinators occupy the next hi- erarchical level and were identified, prior to undertaking this study, as the main focus of investigation. The final level is the self-managing work teams themselves. The size of the teams ranges from approximately three to nineteen members, al- though most included 8 to 1 2. Within each team, there is an elected team leader who also has leadership responsibilities

111 /ASQ, March 1 987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

and receives higher pay than other group members. For the most part, this individual does the same physical work as the other employees.

The work system generally places more responsibility on em- ployees than do typical manufacturing environments. The gen- eral philosophy is self-control or self-management. Work teams are assigned a wide range of tasks and responsibilities, includ- ing preparation of an annual budget, keeping records of their hours worked, recording quality-control statistics (subject to au- dit), making within-group job assignments, and participating in assessment of the job performance of fellow group members. Teams engage in various problem-solving activities that include handling scheduling, equipment, and process problems as well as group-member problem behaviors such as absenteeism. Weekly scheduled and ad hoc meetings serve as problem- solving forums in which such issues are addressed.

The scheduled weekly meetings are typically held in an area of the plant away from the production area. Teams have been trained in conducting meetings and group problem solving. Problems are frequently raised for open discussion during these meetings. Usually, the elected team leader will organize and conduct the meeting, but other team members are provided with the opportunity to speak freely. The external co- ordinator or members of the support team are often invited to work with the self-managing team in dealing with specific is- sues and problems but do not attend on a routine basis. A coor- dinator might attend a team meeting, for example, to help members work out a particularly difficult quality-control problem. Observations of a number of these meetings re- vealed a relatively sophisticated level of discussion and prob- lem solving (i.e., in terms of the technical nature of the discussions and persistent focus on reaching a solution) that focused on improving work performance as well as on various concerns of individual team members.

The pay system at the plant is based on the expertise level of employees. The level of pay for an individual employee is based on the number of tasks he or she can competently perform. When an employee feels he or she has mastered a given job, he or she is tested on that job. In order to reach the highest pay level, an employee must learn all of the jobs of two work groups. This pay system is similar to that used in a pet-foods plant studied by Walton (1977) and in a paint production plant studied by Poza and Markus (1980).

Another distinguishing characteristic of the work system is a noticeable lack of status symbols. The plant manager's office, for example, can be and frequently is used for group meetings without advance permission. There is no assigned parking, and there is a single cafeteria for all employees.

External Team Leadership. Because of the inherent contra- diction of placing an external leader over self-managing teams, it was the coordinator role that was most puzzling. Although the coordinator's role was originally ill-defined, it emerged largely through trial and error during the history of the plant. At the time of the study, the system had generally stabilized, but employees at all levels of the organization, including coordina- tors themselves, still had some degree of ambiguity about what coordinators were actually contributing and ought to be

1 12/ASQ, Ma rch 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

doing. Management was therefore interested in understanding the leader behaviors that facilitated group-member self- management.

The following behavioral instances, a small sample of the ex- tensive qualitative data we collected, illustrate the kinds of co- ordinator behaviors we observed. Coordinator leadership was manifested primarily through minute-by-minute verbal and non- verbal behavior, which we observed at the site. For example, we observed coordinators encouraging teams to engage in the self-management behaviors, self-observation and self- evaluation. Late one afternoon, a young male employee en- tered the quality-control laboratory. He was a production worker from one of the small assembly teams. He went to a chemical work bench and quickly performed a spot test on a piece of material. At the end of the test, he came to the coordi- nator and said, "I think we've got trouble. The chemistry seems to be wrong on X part." With encouragement from the coordi- nator, several other pieces were randomly selected from the production area and were also tested. The problem seemed to be pervasive, and a significant portion of the day's production had to be scrapped. Immediately, a "fix" was initiated to bring the chemistry back within tolerance.

In itself, there is nothing unusual about an employee making a quality check. But in this case, the individual was an hourly pro- duction worker, not a quality-control technician, who "sensed" something was wrong because "the color wasn't just right." He had the initiative and the self-developed skill to conduct the test. It was an example of team members observing their own output and making judgments about the value of that output. It saved the company a small but significant amount of money because the problem was corrected before a great quantity had been produced and the defective material was removed from subsequent production operations. During the entire episode, the coordinator encouraged the team members (and especially the employee who performed the test) and reinforced their alertness in catching this problem. Interestingly, while the coor- dinator was clearly distressed at the loss of production, he re- frained from laying blame and, instead, concentrated on reinforcing the team's behavior in spotting the problem.

We also observed coordinators conducting role-play exercises with elected team leaders - a form of rehearsal - and encour- aging groups to evaluate themselves and to give both positive and negative feedback within the team. In one instance, a team conducted a formal reprimand of a team member during a team meeting for excessive absenteeism. The coordinator had en- couraged this confrontation and worked with team members in rehearsing approaches for the meeting. Our subsequent direct observation revealed a very effective meeting.

In general, (consistent with the self-management leader per- spective) we observed a notable absence of direct commands or instructions from the coordinators to the team. However, questions were used with great frequency. "What is a reason- able scrap rate to shoot for?" (facilitating self-goal-setting); "What will you say to Bill about his absenteeism?" (encourag- ing rehearsal); "Do you think you can do it?" (a way of eliciting high self-expectation).

Other coordinator behaviors also had self-management

11 3/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

implications and were important. For example, we frequently observed coordinators' efforts to encourage group problem solving. Sometimes a special meeting might be suggested to solve a difficult quality or process problem, but the meeting would be carried out by the team itself. In one instance a coor- dinator used questions to stimulate independent problem solv- ing. An employee informed his coordinator that a guard rail on a ramp had been damaged and asked the coordinator what he should do. The coordinator responded by asking, "What do you think should be done?" After a moment of reflection the em- ployee indicated what he thought would be appropriate and, with the coordinator's encouragement and reinforcement, proceeded to fix the guard rail according to his own plan. Coordinators also used more traditional behaviors, such as communication to and from management and positive verbal reward, in some cases.

Overall, we observed a general pattern of behavior that was quite different from our previous experiences in more tradi- tional production plants. The underlying theme of leadership practice was to influence the team and team members to be able to do it themselves, rather than for the coordinators to exercise direct control or do it for the team. There was an abundance of deliberate and calculated efforts to foster inde- pendence rather than allow the dependence of more traditional work groups. In the results that follow, a more specific list of relevant coordinator behaviors is delineated, and the relation- ship of these leader behaviors with overall effectiveness rat- ings is assessed.

RESULTS

Leader Behavior Variables

The final list of variables synthesized from the qualitative phase of this research is given, with a description of each variable, in Table 1. The variables are grouped according to the most closely related existing theory. Each variable that survived to this final list was qualitatively validated through specific on-site observations, interview responses, or group-elicitation responses. These variables were the conceptual input for con- struction of the Self-Management Leadership Questionnaire

Table 1 (SMLQ).

Description of Leader-Behavior Variables, Grouped by Relevant Theory

A priori variables Final variable name* Variable description Mean S.D.t

I. Self-Management Leader Behaviors Self-Management Theory Encourages self-reinforcement Encourages self-reinforcement Leader encourages work group to be self-

(.89) reinforcing of high group performance. 4.64 1.46 Encourages self-criticism Encourages self-criticism Leader encourages work group to be self-

(.80) critical of low group performance. 4.70 1.35 Encourages self-goal-setting Encourages self-goal-setting Leader encourages work group to set

(.85) performance goals. 5.06 1.30 Encourage self-observation Encourage self-observation/evaluation Leader encourages work group to monitor, be Encourage self-evaluation 1 (.81) aware of, and to evaluate level of

performance. 5.16 1.22 Encourage self-expectation Encourage self-expectation Leader encourages work group to have high

(.83) expectations for group performance. 5.33 1.27 Encourage rehearsal Encourage rehearsal Leader encourages work group to go over an

(.85) activity and "think it through" before actually performing the activity. 4.43 1.41

II. "Other" Leader Behaviors Sociotechnical system/autonomous-work- group literature Communicates to management \ Communicates to/from management Leader communicates group views to upper Communicates from management J (.91) management (i.e., support group) and man-

agement views to the group. 5.06 1.14

114/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

Facilitates equipment/supplies Facilitates equipment/supplies Leader facilitates obtaining equipment and Facilitates production flow (.89) supplies for the work group. 5.12 1.23 Communicates between groups Communicates between groups Leader communicates group views to and

(.89) from othergroups. 5.19 1.16 Encourages within-group communication Encourages within-group communication Leader encourages open communication

(.88) among group members, including the exchange of information for learning new jobs. 5.43 1.11

Encourages group training of inexperienced employees Trains inexperienced employees Leader trains group members on various

Trains inexperienced employees (.90) group jobs. 4.36 1.54 Encourages group problem solving Encourages group problem solving Leader encourages the group to solve its

(.74) own problems. 5.42 1.12 Encourages within-group job assignments Encourages within-group job assignments Leader encourages the group to assign tasks

(.85) to its members on its own. 4.84 1.48 Encourages flexible task boundaries Encourages flexible task boundaries Leader encourages group to be flexible in its

(.78) work (i.e., to do whatever needs to be done that the work group is capable of). 5.83 1.02

Operant (reinforcement) literature Positive verbal reward Positive verbal reward Leader verbally rewards (i.e., praises) group

(.90) for desirable performance. 4.90 1.47 Punitive/corrective behavior Punitive/corrective behavior Leader verbally reprimands group for poor

(.86) performance. 5.57 1.08 Goal setting Goal setting Leader assigns performance goals to the

(.84) work group. 4.84 1.28 Expectation of group performance Expectation of group performance Leader expects high group performance. 5.95 .73

(.76)

Information obtained at the research site Communicates production schedule Communicates production schedule Leader communicates to the group about

(.95) plant production schedule (including any changes). 5.21 1.39

Works alongside employees Works alongside employees Leader physically works with group members (.92) to help them do their work. 4.52 1.59

Truthfulness Truthfulness Leader communicates in a way that is truthful (.92) and believable to group members. 5.02 1.37

Encourages group planning (not supported by factor analysis) Communication with other coordinators (not supported by factor analysis)

*Alpha reliability coefficients are given in parentheses. tMeans and standard deviations are of employees' responses about the coordinators.

To aid in our interpretation, we also conducted a factor analysis of the SMLQ to evaluate construct validity. The final factor structure for the self-management leader part of the question- naire is shown in Appendix A. Generally, the variables from the qualitative phase were supported by this analysis, although the variables "encourage self-evaluation" and "encourage self- observation" collapsed into a single variable. Also, a seventh factor was uninterpretable. Finally, Cronbach-alpha coefficients were calculated for each factor and ranged from .80 to .89 (see Table 1).

A similar factor analysis was performed for the "other" leader variables. A fifteen-factor solution was selected from this anal- ysis. Again, the factor analysis largely mirrored the variables de- rived from the qualitative phase. These variables are also listed and defined in Table 1.

Relationships with Leader Effectiveness

Table 2 presents the relationships between the self- management leader variables and effectiveness evaluations of coordinators. Employee evaluations of coordinators were ex- amined two different ways: the team leaders' evaluation and team members' (not including team leader's) evaluations. At first glance, it can be seen that all but one of the Pearson zero- order correlations are significant at p < .01. Coordinator en- couragement of group self-reinforcement, self-observation/ evaluation, and rehearsal display the highest zero-order correla- tions. The largest two zero-order correlations are between the variable "evaluation of the coordinator" by the team leader and the variables "encourages self-reinforcement" (.78) and "en- courages self-observation/evaluation" (.81), respectively.

11 5/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Table 2

Zero-Order Correlations and Fifteenth-Order Partial Correlations between Self-Management Leader Behavior Variables and Evaluations of Coordinators

Evaluation of Coordinator

Self-Management Leader By team leader By team member By support group Behaviors (N = 31)* (N = 234)* (N = 10)t

Coordinator encourages:

Self-reinforcement .780 (.66)0 .550 (-.02) .45 Self-criticism .09g (.06) .360 (.09) -.01

Self-goal-setting .650 (.49) .530 (.05) .16

Self-observation/self-evaluation .810ar (.68)0 .580 (.10) .43 Self-expectation .740** (.59)*** .470** (-.06) .43

Rehearsal .690 (.55)0 .61 are (.10) .540

* p < . 10; * p < .05; *g p < .01. *Correlations to the left a re Pearson zero-order correlations and those i n parentheses are fifteenth-order partial correlations, controlling for 15 "other" leader-behavior variables. tSpearman rank-order correlations.

A partial correlation analysis was conducted by removing from the evaluation criteria variance explained by the 1 5 "other" leader variables. In the case of team-member evaluations of co- ordinators, removal of the partial variance resulted in very little remaining variance explained by the self-management leader variables. However, in the case of the team-leader evaluations, removal of variance due to the "other" leader behaviors still left very significant variance explained by the self-management leader variables. Overall, the analysis indicates that when team leaders do the evaluation, the self-management leader be- haviors play an additional important role, beyond the variance explained by other, more traditional leader behaviors. This par- tial correlation analysis is a very conservative test, since (1) any multicolinearity between the self-management variables and the "other" leader behavior variables will artificially reduce the explained variance of the self-management variables, and (2) many of the "other" variables based on sociotechnical systems theory also relate conceptually to group self-management (e.g., encourages within-group communication).

The strongest relationship with the evaluation variables was "encourages self-observation/self-evaluation." Note also that "encourages self-criticism" was generally not significant. Not only are there potential drawbacks to self-criticism in theory- it focuses on what not to do rather than what to do, it can be de- moralizing, it is difficult to apply consistently (Manz and Sims, 1980; Manz, 1983)- but our interpretation of the variables as a set, combined with our qualitative observations, helps to ex- plain the lack of significant correlations with encouraging self- criticism. We saw numerous instances of coordinators encour- aging a significant amount of self-observation/self-evaluation. Typically, however, this was done in a manner intended to facil- itate future improvement, using a constructive rather than a blaming approach. It was, for example, frequently combined with encouraging self-expectation ("We should try to do bet- ter") or self-goal-setting. Most of all, the coordinators seemed to avoid encouraging self-evaluation for purposes of recrimina- tion or inducing guilt (i.e., self-criticism).

These strong relationships can also be understood better when

116/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

considering our observations of the relationship between coor- dinators and team leaders. Our observations revealed that communication of coordinators with team leaders tended to be more frequent than with other group members. Consequently, team leaders appeared to be in a better position than anyone else in the plant to be aware of the coordinator's behavior. Therefore, the team leaders' evaluations of coordinators were of major interest to us in exploring the results. Despite the modest sample size for the team-leader analysis, many of the correlations clearly met standards of statistical significance. These results suggest that team leaders can make strong dis- tinctions between the coordinators' use of different classes of leader behaviors. Self-management coordinator behaviors were clearly correlated with team-leader ratings of overall coor- dinator effectiveness.

Finally, we have also included in Table 2 the correlations be- tween the self-management leader behaviors and evaluations by the support team. These correlations are presented for sup- plementary information only, since the leader-behavior

-variables have been aggregated by coordinator and are subject to the criticism brought out by Firebaugh (1978) (see Appendix B). Also, the sample size (N = 10) is too small for meaningful tests of significance. Nevertheless, it is useful to compare the pattern and magnitude of these correlations with the other rela- tionships in the table as one additional qualitative source of in- formation. Note that a convergent pattern of relationships emerges: the managers assess the components of coordinator effectiveness in a manner similar to team leaders. Overall, this nonstatistical analysis adds some additional confidence to the previous results. Patterns of correlations with effectiveness ratings generally converge from sources both above and below the coordinator level.

Comparing Coordinators and Team Leaders

Collection of the questionnaire data also provided an oppor- tunity to contrast statistically the role of the coordinator with the role of the elected team leader. In Table 3, a significant t-value indicates that for the entire sample, an overall dif- ference was perceived between team leader behavior and co- ordinator behavior on a given leader-behavior variable.

First, all of the self-management leader behaviors demonstrate significant differences except encourages rehearsal." The di- rection of these differences indicates that coordinators were perceived to be significantly higher on encouraging self- management than were the team leaders.

In addition, some significant differences in the opposite direc- tion were revealed on the "other" leader behaviors: facilitates equipment/supplies, trains inexperienced employees, and works alongside employees. These differences reflect the role of team leaders, who work alongside other team members. In fact, it was fairly typical for team members to engage in these same kinds of behaviors to assist one another.

We have also included the statistics from the employees' eval- uation of the team leader and the coordinator and satisfaction with the team leader and the coordinator, which showed no significant difference. This finding serves as a type of baseline and seems to indicate that the differences do not stem from an

1 17/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Table 3

Differences between SMLQ Scores of Team Leader and Coordinator Behaviors, Employee Evaluation, and Satisfaction*

Mean Score Variable Team leader Coordinator T-value

Self-management leader variables Encourages self-reinforcement 4.32 4.59 -3.77 (p < .001) Encourages self-criticism 4.37 4.67 -5.20 (p < .001) Encourages self-goal setting 4.66 5.02 -5.22 (p < .001) Encourages self-observation/evaluation 4.85 5.17 -5.29 (p <.001) Encourages self-expectation 4.93 5.34 -6.36 (p < .001) Encourages rehearsal 4.31 4.40 -1.31 (p <.19)

Other leader behavior variables Communication to/from management 5.02 5.06 -.80 (p < .43) Facilitates equipment/supplies 5.39 5.11 4.45 (p < .001) Communicates between groups 5.16 5.18 -.27 (p < .78) Communicates production schedule 5.18 5.20 -.35 (p < .73) Encourages within-group communication 5.39 5.43 -.72 (p < .47) Trains inexperienced employees 5.42 4.43 10.94 (p <.001) Encourages group problem solving 5.17 5.38 -3.46 (p < .001) Encourages within-group job assignments 4.78 4.76 .18 (p < .86) Encourages flexible task boundaries 5.62 5.81 -3.65 (p < .001) Positive verbal reward 4.65 4.86 -2.48 (p < .014) Punitive/corrective behavior 5.17 5.59 -7.04 (p < .001) Goal setting 4.54 4.89 -5.67 (p <?.001) Expectation of group performance 5.70 5.97 -6.93 (p < .001) Works alongside employees 5.71 4.53 11.29 (p <.001) Truthfulness 5.08 4.99 1.15 (p <.25)

Employee evaluation and satisfaction Evaluation of team leader and coordinator 5.09 5.02 .95 (p < .345) Satisfaction with team leader and coordinator 5.09 5.18 -.99 (p < .324)

*Approximate N for each paired comparison is 240.

overall affective response bias in favor of the coordinator.

These findings, in conjunction with our qualitative observa- tions, reflect fundamental differences in the roles of the two types of leaders. The team leader, within the group, serves as an additional team member who facilitates the group's organiz- ing itself and coordinating job assignments and who makes sure materials are available. The coordinator, outside the group, has a fundamental responsibility to get the group to manage it- self, and he or she employs various self-management leader behaviors to accomplish that end.

Overall, the statistical results of this study indicate that several self-management leader behaviors were significantly related to overall leadership effectiveness and that these relationships exist even when the effects of more traditional leader-behavior variables are removed.

Finally, while we do not have specific quantitative data to ascer- tain whether this system as a whole was effective or not, infor- mal information acquired from conversations with corporate officials indicated productivity gains "significantly greater than 20 percent" when compared to other plants of the same tech- nology using more traditional management methods. Our dis- cussions with management also revealed similar positive comparisons regarding quality, turnover, and satisfaction. In the case of turnover, for example, in response to our inquiry, a manager in the plant counted on the fingers of one hand the employees who had chosen to leave. The corporation, in fact, is

118/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

currently attempting to diffuse similar self-management systems throughout the organization.

DISCUSSION

Some of the leader-behavior variables articulated in this re- search are similar to those found in the existing leadership liter- ature - (e.g., communicating to and from management and between work groups, helping to assure that work groups have the equipment and supplies they need, training inexperienced employees, etc.). However, a fundamental difference does ex- ist in how leader functions are carried out, especially in terms of the shift in source of control from the leader to the follower. The uniqueness of the self-management leader's role lies in the commitment to the philosophy that the teams should suc- cessfully complete necessary leadership functions for them- selves. The dominant role of the external leader then, is to lead others to lead themselves. This is quite different from the tradi- tional role of the leader.

Contrast with Traditional and Participative Leader Behavior

Ideally, in this study we would have preferred to conduct a di- rect experimental contrast with more traditional plants, but this was not politically feasible in this organization. However, we did conduct a retrospective qualitative comparison, addressing two essential questions. First, how does a leader of a self- managing team differ from a traditional foreman or supervisor? Second, how does leadership of self-managing teams compare with leadership in other participative situations? To address the first question, we drew on our own practical field experience, our own experience in leadership research, and especially the rather extensive literature on leadership research (e.g., Stogdill, 1974, Bass, 1981). Two published works seem especially rep- resentative in capturing the essence of what traditional leaders really do.

In the 1970s, Schriesheim and various colleagues conducted extensive analysis of original leader-behavior data and the lead- ership literature, with a special focus on the very dominant Ohio State Leadership scales. Much of this work was syn- thesized and summarized in an article by Schriesheim, House, and Kerr (1976), in which they attempted to integrate the the- oretical and empirical findings on the leadership scale of initiat- ing structure. To some extent, their summary represents the "conventional wisdom" or what is commonly known about the leadership of traditional groups.

Schriesheim and his colleagues drew especially from three widely used instruments to measure leader behavior: the Supervisory Behavior Description Questionnaire (SBDQ) (Fleishman, 1957), the early Leader Behavior Description Questionnaire (LBDQ) (Halpin, 1957), and the revised LBDQ (Stogdill, 1963). The SBDQ is especially appropriate for pur- poses of comparison, because it mainly refers to shop-floor leadership situations. Furthermore, generation of the original item pools of the Ohio State leadership questionnaires was deeply rooted in empirical field investigations.

Schriesheim and his colleagues articulated what is now widely accepted: that the initiating-structure dimension is really a rather complex abstract construct that is composed of several

119/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

subdimensions. They identified and listed these subdimen- sions - initiation, organization, domination, production, recog- nition, integration, and communication -which are described in Table 4.

Table 4

Comparison of Leader-Behavior Dimensions from the Literature with Self-Management Leader Dimensions

Traditional Leader Behaviors Initiating Structure Dimensions* Leader Prototype Elicitationt Self-Management Leader Behaviors

Initiation: Originates, facilitates, or Emphasizes goals. Encourages self-reinforcement. rejects new ideas.

Organization: Defines and structures his Coordinates groups. Encourages self-observation/evaluation. or her own work, the work of group members, and the relationships Provides information. Encourages self-expectation. among group members in the performance of work. Proposes solutions. Encourages self-goal-setting.

Domination: Restricts the behavior of individuals or the group in action, Specifies problems. Encourages rehearsal. decision making, or freedom of expression. Exercises influence. Encourages self-criticism.

Production: Sets levels of effort or achievement or prods members for Talks frequently. greater effort.

Recognition: Expresses approval or disapproval of group-member behavior.

Integration: Subordinates individual behavior, encourages pleasant group atmosphere, reduces conflict among members, and promotes individual adjustment to the group.

Communication: Provides information to subordinates, seeks information from them, facilitates information exchange, shows awareness of affairs pertaining to the group.

*Taken from Schriesheim, House, and Kerr (1976). tTaken from Lord, Foti, and De Vader (1984).

As a second and more recent description of traditional leader behavior, we drew upon the line of research conducted by Lord and his associates on cognitive representations of leadership prototypes. Lord, Foti, and De Vader (1984, and personal com- munication), for example, conducted research in which they used cognitive elicitation techniques to identify the attributes typically associated with leadership-category prototypes. Like the early roots of the Ohio State research, this is especially use- ful for our purposes because it identifies and articulates, through empirical investigation, the cognitive content of what people actually bring to mind when the category of "leader" is stimulated. The results of this cognitive elicitation are also shown in Table 4. Lord, Foti, and De Vader (1984) reported that these leader attributes all empirically demonstrate high leader prototypical ratings in subsequent research, as well as a very high correlation with overall leadership perceptions.

As one compares the findings of Schriesheim and his associ- ates with those of Lord and his associates, a striking similarity is evident, despite the differences in methods and time: there is an implicit presumption that influence flows from the top down. That is, widely held traditional notions of leadership

120/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

seem to revolve around the idea that the leader is someone who does something to influence someone else directly. The assumption is that the power and initiation lie virtually entirely with the leader.

Table 4 also shows the self-management leader dimensions from our own research, for purposes of contrast. The order has been rearranged so that the dimensions relating most strongly to leader effectiveness have been listed first. The implicit as- sumptions underlying the set of self-management dimensions is quite different from those of the earlier, more traditional be- haviors. Instead of a top-down philosophy of control, these di- mensions imply a bottom-up perspective. Even the labels of the dimensions indicate that subordinates can perform leader- ship functions for themselves and the leader's job is to teach and encourage subordinates to lead themselves effectively. As we observed the leadership practices within the self-managed work group system, this fundamental philosophical difference was strongly apparent. In the self-management system, organizing, directing, and monitoring functions - all part of tra- ditional notions of leadership -are largely centered within the group.

It is also useful to compare self-management leader behaviors with more generic participative-leadership behaviors, that is, those intended to invoke participation of workers but not nec- essarily in self-managing teams.

The concept of participation has been addressed in the litera- ture for many years, certainly predating self-managing work teams. Burns and Stalker (1959: 233), for example, argued that in organic forms of organization, workers, when they are oper- ating effectively, tend to engage in behaviors through which they "openly or tacitly reject subordination" - behaviors such as those found at our self-managing research site. In addition, they pointed out that a movement from mechanistic to organic forms tends to result in communication that was formerly based on vertical upward flows of information and downward movement of decisions and instructions giving way to lateral exchanges between equals and members of teams. Thus, they suggested that the content of the communication in the organic firm will consist of advice and information, as opposed to decisions and instructions.

At the philosophical level, the influence processes suggested in the organic form of organization are distinctly similar to those found in self-managing teams. For example, the departure from the clarity of a hierarchic structure in organic forms can cause managers to seek more information (e.g., through the use of questions, as was suggested by our data for coordinators) rather than to rely on managing by command. Burns and Stalker (1959) also suggested some general ingredients in organic sys- tems at the level of the "managing director," e.g., direct leader- ship over subordinates becomes less important; persons in the system must work out their own actions within temporary deci- sion frames established by people around them.

While extensive reviews of participation in decision making are available (e.g., Locke and Schweiger, 1979), specific recom- mendations of elements for leading employees under participa- tive conditions are scarce in the literature. Vroom and Yetton ( 1973), for example, delineated conditions under which pa r-

12 1/ASQ, Ma rch 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

ticipative decision making might be more appropriate, but they offered little guidance on how such participation should be car- ried out by specific day-to-day leader behaviors. As recently as 1982, Strauss completed an international review of worker par- ticipation and concluded that there has been almost no re- search on how managers actually should behave in such situations. He pointed out that the conventional wisdom re- garding such organizations is that managers should "be pa- tient, be good listeners and politicians, make decisions by consensus rather than by edict, and attend a lot of meetings. Beyond this, what should they do?" (Strauss, 1982: 234).

Some contemporary writers offer more specific suggestions on participative leadership, as has Bradford (1976), under his la- bel of group-centered leadership. Yukl (1981: 246-247) sum- marized Bradford's prescriptions: the leader should use attentive listening and careful observation of nonverbal cues of member needs, feelings, etc.; serve as a consultant, advisor, teacher, and facilitator; model and encourage appropriate leader behaviors; establish a climate that is conducive to ex- pression of both feelings and ideas; encourage the group to ad- dress its maintenance needs and process problems in its regular group meetings; and relinquish control, allowing the group to make final decisions on appropriate issues. Indeed, these role prescriptions are consistent with the self- management behaviors found in our research.

Similarly, Meadows (1980), drawing heavily on the original work of Burns and Stalker to operationalize "organicity," de- veloped dimensions such as blurred role boundaries, suppor- tive internal communications, and freedom of access to group decision processes. Again, these dimensions suggest parallel logic, but without the specificity of the behaviors identified in our study. He found further evidence that organicity was positively correlated with innovativeness in 28 research and de- velopment groups.

More specific to employee self-management, Hackman (1986), inspired by the early work of McGrath (1962), suggested that if the leader ensures that all critical functions for group mainte- nance and task completion are achieved within the group itself, the leader has done the job well. Hackman proposed that the role of leaders of self-managing units should include the two primary functions of monitoring and taking action to facilitate favorable performance conditions. A strength of this functional approach, he argued, is that it avoids the trap of trying to spec- ify specific actions that should always be applied in given situa- tions by allowing for a multitude of ways for accomplishing a given function.

Overall, the participation literature reviewed above makes a sig- nificant contribution by suggesting a guiding philosophy and overall roles for leaders in highly participative situations. Nev- ertheless, specific leader behaviors, particularly those that serve to equip employees with self-management tools for suc- cessfully coping with significant autonomy, are scarce. This point is especially relevant when addressing the leadership of self-managing groups. Lawler (1986), for example, pointed out that a great deal of confusion and frustration has sur- rounded the leadership of self-managing groups in almost every case (cf. Walton and Schlesinger, 1979). Leaders com-

122/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

plain of confusion in their role in terms of what decisions they should and should not make. As Lawler (1986: 184) stated, "the problem is that there is no clear cut description of the cor- rect behavior for the first line supervisor. . . . Several organiza- tions are trying to solve this problem by developing appropriate training programs, but to the best of my knowledge, no exem- plary program exists." Hackman (1986) helped to clarify some of this confusion by applying the functions of monitoring and action taking across five enabling conditions: direction, struc- ture, context, coaching, and resources. Nevertheless, specific behaviors are not prescribed.

The leader behaviors addressed in this paper do not represent entirely new conceptual and philosophical perspectives about the leader's role. Rather, they help to extend previous view- points of participative leadership by providing more specific be- havioral descriptions, based largely on direct observations of self-managing group leaders engaged in the act of leading.

Before concluding this discussion, it is important to address di- rectly some of the significant limitations of this study. The first and most obvious weakness is the absence of a control or con- trast group that would provide a direct comparison with leader- ship practices in more traditional plants. From a classical experimental viewpoint, the real "treatment" in this plant was the implementation of the self-managing team concept, which was originally launched in 1974. To be realistic, our research, in terms of studying the work system, is essentially a "post-only" measurement, without control. That is, measures were taken only after the team-system implementation, with no measures taken before the implementation or from a control group. Campbell and Stanley (1966) clearly viewed this design as vir- tually the weakest possible. Unfortunately, a stronger design was not politically feasible.

Another weakness is the dependence of the quantitative part of the research on leadership questionnaire methods, which have been attacked for questionable validity. Salancik and Pfeffer (1978), for example, in the job characteristics area, pre- sented a social-information-processing perspective that sug- gests there is a social cognitive construction explanation of job- characteristic measures. Research has supported this view- point to some degree, although some research has shown job- characteristic measures to be affected by some combination of social construction and the external environment (cf. Griffin, 1983).

Leadership measures based on questionnaire methods are subject to the same criticisms. Do these leader-behavior measures represent some underlying schema (cf. Lord, 1985), as opposed to specific behavior of the leader? Gioia and Sims (1985) directly investigated this issue and concluded that broad global items were indeed subject to social construction, while more narrow specific items were more likely to reflect actual behavior. Yukl (1981) also pointed out the need for greater pre- cision in our measures of leader behaviors to study what leaders do (as well as the need to use multiple methods, includ- ing direct observation). Further, it is difficult to know whether each employee has sufficient knowledge about each specific leader behavior, for example, how a leader communicates to management. This potential weakness was addressed, to

123/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

some extent, by relying on information gained from the re- search site to help generate and qualitatively validate the mea- sures used. Questionnaire measures also have a tendency to inflate correlations when the criterion variable is taken from the same source (although the partial-correlation processes used in this research help to diminish this problem). Further, as dis- cussed before, what are overall ratings of leader effectiveness really measuring?

Despite these epistemological limitations to the leadership- questionnaire measures, we believe they serve a useful func- tion, especially if carefully constructed from behavioral-specific items (as recommended by Gioia and Sims, 1985, and Yukl, 1981) and when triangulated by careful observational methods. The study itself afforded the opportunity to observe intensively a rarely seen organizational system based on self-managing teams and especially the leadership of those teams.

Overall, this study suggests that there is a legitimate role for external leaders of self-managing teams and that it differs con- siderably from the traditional leadership role. In general, both the qualitative data and our quantitative analyses suggest (1) the self-management leader behaviors (derived from self- management theory) generally represent important indicators of external-leader effectiveness in team situations, (2) there is some evidence that these leader behaviors have a significant additional importance over and above the more traditional lead- ership or sociotechnical system leader behaviors, (3) an impor- tant means by which these self-management leader behaviors are expressed is through one-on-one individual communication with team members, especially internal team leaders, and (4) the most important self-management leader behaviors were "encourage self-reinforcement" and "encourage self- observation/evaluation."

The self-management leader behaviors and "other" relevant leader behaviors identified in this research appear to provide a beginning for exploring and understanding what external leaders actually do in a context characterized by a strong self- management philosophy. Also, we believe these behaviors have potential for allowing us to gain insight into other situa- tions, perhaps even those in traditional organizations where a self-management philosophy is not overtly embraced. Clearly, additional research is needed to further validate and better un- derstand this list of behaviors. Ultimately, such a line of re- search could uncover the insights needed for meeting the challenge of leading those who are supposed to be self- managing and for providing badly needed guidelines for select- ing and training external leaders of self-managing teams.

REFERENCES

Andrasik, Frank, and Judy S. Heimberg 1982 "Self-management pro-

cedures." In L. W. Frederiksen (ed.), Handbook of Organiza- tional Behavior Management: 219-247. New York: Wiley.

Bandura, Albert 1977 Social Learning Theory. En-

glewood Cliffs, NJ: Prentice- Hall.

Bass, Bernard M. 1981 Stogdill's Handbook of Leader-

ship: A Survey of Theory and Research. New York: Free Press.

Blalock, Hubert M., Jr. 1979 "The presidential address:

Measurement and conceptual- ization problems: The major obstacle to integrating theory and research." American So- ciological Review, 44: 881-894.

Bradford, Leland P. 1976 Making Meetings Work. La-

Jolla, CA: University Associates.

Brown, Frederick G. 1976 Principles of Educational and

Psychological Testing. New York: Holt, Rinehart and Winston.

124/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

Burns, Tom R., and G. M. Stalker 1959 The Management of Innova-

tion. Chicago: Quadrangle Books.

Campbell, Donald T., and Julian C. Stanley 1966 Experimental and Quasi-

Experimental Designs for Re- search. Chicago: Rand McNally.

Cartwright, D., and A. Zander 1963 Group Dynamics: Research

and Theory. New York: Harper & Row.

Cronbach, Lee J. 1951 "Coefficient alpha and the in-

ternal structure of tests." Psy- chometrika, 16: 297-334.

Cummings, Thomas 1978 "Self-regulated work groups: A

socio-technical synthesis." Academy of Management Re- view, 3: 625-634.

Cummings, Thomas, and Edmund S. Malloy 1977 Improving Productivity and the

Quality of Work Life. New York: Praeger.

Delbecq, Andr6 L., Andrew H. Van de Ven, and David H. Gustafson 1975 Group Techniques for Program

Planning: A Guide to Nominal and Delphi Processes. Glen- view, IL: Scott, Foresman.

Emery, Fred E., and Eric L. Trist 1969 "Socio-technical systems." In

F. E. Emery (ed.), Systems Thinking: 281-296. London: Penguin Books.

Firebaugh, Glenn 1978 "A rule for inferring individual-

level relationships from aggre- gate data." American Sociologi- cal Review, 43: 557-572.

Fleishman, Edwin A. 1957 "A leader behavior description

for industry." In R. M. Stogdill and A. E. Coons (eds.), Leader Behavior: Its Description and Measurement. Columbus, OH: Bureau of Business Research, Ohio State University.

Gioia, Dennis, and Henry P. Sims, Jr. 1985 "On avoiding the influence of

implicit leadership theories in leader behavior descriptions." Educational and Psychological Measurement, 45: 217-237.

Gove, Walter R., and Michael Hughes 1980 "Reexamining the ecological

fallacy: A study in which aggre- gate data are critical in inves- tigating the pathological effects of living alone." Social Forces, 58: 1157-1177.

Griffin, Ricky W. 1983 "Objective and social sources

of information in task redesign: A field experiment." Admin- istrative Science Quarterly, 28: 184-200.

Hackman, J. Richard 1986 "The psychology of self-

management in organizations." In M. S. Pollack and R. 0. Per- loff (eds.), Psychology and Work: Productivity Change and Employment: 85-1 36. Wash- ington, DC: American Psycho- logical Association.

Halpin, A. W. 1957 Manual for the Leader Behavior

Description Questionnaire. Co- lumbus, OH: Bureau of Busi- ness Research, Ohio State University.

Jick, Todd D. 1979 "Mixing qualitative and quan-

titative methods: Triangulation in action." Administrative Sci- ence Quarterly, 24: 602-61 1.

Lawler, Edward E. 1986 High Involvement Manage-

ment. San Francisco: Jossey- Bass.

Lazerfeld, Paul F., and Herbert Menzel 1972 "On the relation between indi-

vidual and collective proper- ties." In P. F. Lazerfeld, A. K. Pasanella, and M. Rosenberg (eds.), Continuities in the Lan- guage of Social Research: 225- 248. New York: Free Press.

Locke, Edwin A., and David M. Schweiger 1979 "Participation in decision mak-

ing: One more look." In L. L. Cummings and Barry M. Staw (eds.), Research in Organiza- tional Behavior, 1: 265-339. Greenwich, CT: JAI Press.

Locke, Edwin A., Karyll N. Shaw, Lise M. Saari, and Gary P. Latham 1981 "Goal setting and task perform-

ance." Psychological Bulletin, 90:125-152.

Lord, Robert G. 1985 "An information processing ap-

proach to social perceptions, leadership, and behavioral mea- surement in organizations." In L. L. Cummings and B. M. Staw (eds.), Research in Organiza- tional Behavior, 7: 87-1 28. Greenwich, CT: JAI Press.

Lord, Robert G., R. J. Foti, and C. L. De Vader 1984 "A test of leadership categori-

zation theory: Internal struc- ture, information processing, and leadership perceptions." Organizational Behavior and Human Performance, 34: 343- 378.

Luthans, Fred, and Robert Kreitner 1975 Organizational Behavior Modi-

fication. Glenview, IL: Scott, Foresman.

Maccoby, Michael 1981 The Leader. New York: Ballan-

tine Books.

Mahoney, Michael J., and Diane B. Arnkoff 1978 "Cognitive and self-control

therapies." In S. L. Garfield and A. E. Borgin (eds.), Handbook of Psychotherapy and Behavior Change: 689-722. New York: Wiley.

1979 "Self-management: Theory, re- search, and application." In J. P. Brady and D. Pomerleau (eds.), Behavioral Medicine: Theory and Practice: 75-96. Baltimore: Williams & Williams.

Manz, Charles C. 1983 The Art of Self-Leadership.

Englewood Cliffs, NJ: Prentice- Hall.

1986 "Self-leadership: Toward an expanded theory of self- influence processes in organi- zations." Academy of Manage- ment Review, 11: 585-600.

Manz, Charles C., and Harold Angle 1986 "Can group self-management

mean a loss of personal con- trol: Triangulating on a para- dox." Group and Organization Studies: 11: 309-334.

Manz, Charles C., and Henry P. Sims, Jr. 1980 "Self-management as a sub-

stitute for leadership: A social learning theory perspective." Academy of Management Re- view, 5: 361-367.

1984 "Searching for the 'unleader': Organizational member views on leading self-managed groups." Human Relations, 37: 409-424.

1978 "Leading self-managed groups: A conceptual analysis of a paradox." Economic and Industrial Democracy, 7: 141-165.

McGrath, Joseph E. 1962 Leadership Behavior: Some

Requirements for Leadership Training. Washington, DC: U.S. Civil Service Commission.

125/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Meadows, Ian S. G. 1980 "Organic structure and innova-

tion in small work groups." Hu- man Relations, 33: 369-382.

Mills, Peter K. 1983 "Self-management: Its control

and relationship to other organi- zation properties." Academy of Management Review, 8: 445-453.

Poza, Ernesto J., and M. Lynne Markus 1980 "Success story: The team ap-

proach to work restructuring." Organizational Dynamics, Win- ter: 3-25.

Roberts, Karlene H., and L. Bernstein (eds.) 1980 New Directions for Methodol-

ogy of Social and Behavioral Science: Issues in Aggrega- tion. San Francisco: Jossey- Bass.

Roberts, Karlene H., Charles L. Hulin, and Denise M. Rousseau 1978 Developing an Interdisciplinary

Science of Organizations. San Francisco: Jossey-Bass.

Rummel, Rudolph J. 1970 Applied FactorAnalysis. Evan-

ston, IL: Northwestern Univer- sity Press.

Salancik, Gerald R., and Jeffrey Pfeffer 1978 "A social information process-

ing approach to job attitudes and task design." Administra- tive Science Quarterly, 23: 224-253.

Schriesheim, Chester A., Robert J. House, and Steven Kerr 1976 "Leader initiating structure: A

reconciliation of discrepant re- search results and some em- pirical tests." Organizational Behavior and Human Perform- ance, 15: 297-321.

Sims, Henry P., Jr. 1977 "The leader as a manager of re-

inforcement contingencies: An empirical example and a model." In J. G. Hunt and L. L. Larson (eds.), Leadership: The Cutting Edge: 121-137. Car- bondale, IL: Southern Illinois University Press.

Stogdill, Ralph M. 1963 Manual for the Leader Behavior

Description Questionnaire Form XII. Columbus, OH: Bu- reau of Business Research, Ohio State University.

1974 Handbook of Leadership: A Survey of Theory and Re- search. New York: Free Press.

Strauss, George 1982 "Workers' participation in man-

agement: An international per- spective." In L. L. Cummings and B. M. Staw (eds.), Re- search in Organizational Be- havior, 4: 173-265. Greenwich, CT: JAI Press.

Susman, Gerald I. 1976 Autonomy at Work: A Socio-

technical Analysis of Participa- tive Management. New York: Praeger.

Thoreson, Carl E., and Michael J. Mahoney 1974 Behavioral Self-Control. New

York: Holt, Rinehart & Winston.

Thorndike, Robert M. 1978 Correlational Procedures for

Research. New York: Gardner Press.

Trist, Eric 1977 "Collaboration in work set-

tings: A personal perspective." Journal of Applied Behavioral Science, 13: 268-278.

Trist, Eric L., Gerald I. Susman, and G. R. Brown 1977 "An experiment in autonomous

working in an American under- ground coal mine." Human Relations, 30: 201-236.

Vroom, Victor H., and Phillip W. Yetton 1973 Leadership and Decision Mak-

ing. Pittsburgh: University of Pittsburgh Press.

Walton, Richard E. 1977 "Work innovations at Topeka:

After six years." Journal of Ap- plied Behavioral Science, 13: 422-433.

Walton, Richard E., and Leonard A. Schlesinger 1979 "Do supervisors thrive in par-

ticipative work systems?" Or- ganizational Dynamics, 8(3): 24-39.

Yukl, Gary A. 1979 Leadership in Organizations.

Englewood Cliffs, NJ: Prentice- Hall.

Yukl, Gary A., and Wayne Nemeroff 1986 "Identification and measure-

ment of specific categories of leadership behavior: A prog- ress report." In J. G. Hunt and L. L. Larson (eds.), Crosscur- rents in Leadership: 164-200. Carbondale, IL: Southern Il- linois University Press.

APPENDIX A: Factor Analysis of the Self-Management Leader Behavior Items

The factor analysis of the SMLQ is presented in Table Al. In this initial analysis of the SMLQ, no simple criteria were used in determining the number of fac- tors adopted from the analysis. The best description of the criterion we actually used in determining the number of factors is what Rummel (1970: 357) has re- ferred to as a "research sense" of the data. This approach was considered more acceptable than deciding on the number of factors according to a more ar- bitrary decision, such as an eigenvalue greater than 1 .0, which has been se- verely questioned elsewhere (e.g., Thorndike, 1978: 267). Consequently, a number of factor solutions were examined for each set of variables in an at- tempt to identify the solution that made the most sense.

126/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

Table Al

Factor Analysis of the Self-Management Leader Behavior Items*

Encouraging self-management leader Commu- behavior items I II IlIl IV V VI VIl nality

I. Encourage rehearsal He/she encourages us to go over an ac- tivity before we attempt it. .66 .26 .16 .15 .13 .13 .15 .60 He/she encourages us to practice (either physically or mentally) a new task before we do it the first time. .57 .25 .22 .26 .18 .15 .03 .56 He/she helps us to go over a new task be- fore we actually begin the task. .59 .22 .12 .23 .17 .19 -.01 .52 He/she encourages us to think about how we are going to do a job before we begin the job. .68 .21 .20 .21 .28 .13 .06 .69

II. Encourage self-goal-setting He/she prompts us to define the goals for our own team. .22 .63 .23 .16 .05 .13 .19 .58 He/she urges us to define our own goals for our group. .33 .57 .10 .15 .11 .32 .06 .59 He/she encourages us to establish our own task goals. .27 .58 .18 .28 .22 .10 -.01 .58 He/she encourages us to set goals for our group performance. .24 .64 .11 .22 .38 .21 -.06 .71

Ill. Encourage self-criticism If our performance on a job is below par, he/she encourages us to be critical of ourselves. .15 .26 .57 .07 .01 .16 .21 .49 He/she expects us to be tough on our- selves when our performance is not up to standard. .13 .10 .56 .08 .30 .12 .41 .62 He/she urges us to be self-critical if our performance is not up to par. .19 .17 .68 .30 .12 .11 -.07 .64 He/she encourages us to be critical of our- selves when we do poorly. .13 .05 .69 .15 .22 .04 -.04 .57

IV. Encourage self-reinforcement He/she encourages us to praise each other if we have done a job well. .25 .31 .31 .56 .05 .17 .09 .60 If we do an assignment especially well, then he/she encourages us to feel positive about ourselves. .40 .22 .17 .52 .19 .10 .44 .75 He/she encourages us to praise each other for doing a good job. .34 .24 .28 .68 .23 .10 -.02 .78 He/she encourages us to feel good about ourselves if we do a job well. .46 .27 .19 .50 .29 .21 .09 .71

V. Encourage self-expectation He/she encourages our group to think we can do very well in our work. .33 .32 .14 .34 .47 .32 .09 .67 He/she encourages us to expect high per- formance from our group. .28 .30 .25 .20 .58 .21 .16 .68 He/she encourages us to expect a lot from ourselves. .32 .13 .37 .12 .60 .09 .05 .64

VI. Encourages self-observation/evaluation He/she encourages us to be aware of our level of performance. .27 .35 .21 .13 .18 .67 .10 .74 He/she encourages us to know how our performance stands. .26 .33 .24 .25 .34 .38 -.04 .56 He/she encourages us to judge how well we are performing. .34 .29 .19 .34 .30 .42 .03 .61

Eigenvalues 10.71 1.52 1.07 .94 .82 .71 .67 Variance explained (%) 48.7 6.9 4.9 4.3 3.7 3.2 3.0 Cumulative variance explained(%) 48.7 55.6 60.5 64.8 68.5 71.8 74.8 Reliability (alpha) .85 .85 .80 .89 .83 .81 -

*Factor structure is shown after principal components analysis with varimax rotation.

127/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

APPENDIX B: Mixed-Level Analysis

Concern has been expressed in the literature regarding extrapolation of data obtained at one analysis level (e.g., the individual-respondent level) to another level (e.g., the group level) (Firebaugh, 1978; Blalock, 1979; Roberts, Hulin, and Rousseau, 1978; Gove and Hughes, 1980; Roberts and Bernstein, 1980). More specific to this research, the practice of calculating group means from individ- ual responses and using these means for group-level analysis has been ques- tioned. A typical result of such an approach is an exaggeration of relationships found at the group level as a result of the influence of an external variable or a set of variables associated with group membership (for a more thorough dis- cussion of this problem, see Lazerfeld and Menzel, 1972; Firebaugh, 1978; Blalock, 1979; Gove and Hughes, 1980). Consequently, aggregate-level analyses of this type should be viewed with caution.

Initially, we questioned whether our analysis should probe for group-level effects. First, an analysis of variance was performed on all leader-behavior variables, with group membership as the independent variable. The F statistics for four of the six self-management leader variables and six of the fifteen "other" leader-behavior variables were significant at the p < .05 level. This re- suit provided an indication of interrater agreement among group members about coordinator behavior and some justification for pooling behavior descrip- tions into a group average in order to probe for group-level effects.

In further analysis, in order to evaluate the relative effect at the individual level versus the relative effect at the group level, we used a mixed-level analysis, proposed by Firebaugh (1978). This method involves combining both individual- level responses and aggregate data (i.e., group means) into one data set for fur- ther analyses. Thus, in the new data set, each case includes both the individual responses and the group means for each of the research variables, hence the label, mixed-level analysis. Such an approach enables study of both individual and group effects while maintaining the larger individual-level sample size.

Multiple regressions were performed for selected dependent variables re- gressed on individual- and group-level data for each of the independent leader- behavior variables. Separate multiple regressions were performed using the in- dividual responses and group means for each of the independent variables, rather than entering many or all of the independent variables into a single (or few) multiple regression(s), so as to assess the independent predictive power of each of the leader-behavior variables. Each regression can be viewed as tan- tamount to a bivariate correlation in which the group effect of the independent variable has been separated from the individual effect.

The mixed-level analysis indicated that the leader-behavior variables generally demonstrated predictive power for the dependent variables involved (overall F ratings were significant at p < .01 in nearly every case) and that individual-level effects were much stronger than group effects. For example, standardized beta weights at the individual-respondent level for coordinators were higher than .70 in many cases, while standardized beta weights for group-level effects were often negative and uniformly much smaller than the corresponding indi- vidual effect for each of the leader-behavior variables. Table B1 shows the beta weights and overall F-ratios for the self-management leader variables. Results were similar for the "other" leader-behavior variables.

One speculative explanation for these findings is that, despite the formal group structure, self-management principles ultimately operate at the individual level.

128/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions

Self-Managing Work Teams

Table B1

Regressions of Leader-Effectiveness Evaluations on Self-Management Leader Variables, Including Group Means from the Coordinator-Aggregate Level*

Evaluation of Coordinator By team leader By team member

Self-Management Leader Individual effect (Bi)/ Overall Individual effect (B1)/ Overall Behaviors group effect (B2) F-ratio group effect (B2) F-ratio

Encourage:

Self-reinforcement .80/-.11 22.59 .52/.14 54.12

Self-criticism .15/-.26 1.10t .33/.12 18.71

Self-goal-setting .70/-.20 11.50 .50/.06 43.35

Self-observation/self-evaluation .82/-.05 26.29 .54/.14 59.96

Self-expectation .79/-.19 19.42 .42/.17 36.62

Rehearsal .73/-.12 13.57 .57/.13 72.45

d.f. = 2, 28 d.f. = 2, 227

*Group effects were obtained by using group means aggregated at the coordinator level. Numbers presented are the standardized Beta weights from a series of independent regression analyses. tNot significant at p < .05; all other F-ratios are significant at p < .01.

129/ASQ, March 1987

This content downloaded from 195.34.79.20 on Thu, 12 Jun 2014 14:54:37 PMAll use subject to JSTOR Terms and Conditions