WHAT MAKES TEAMS TAKE? EMPLOYEE REACTIONS TO WORK …

448

Industrial and Labor Relations Review, Vol. 55, No. 3 (April 2002). © by Cornell University.0019-7939/00/5503 $01.00

T

WHAT MAKES TEAMS TAKE?

EMPLOYEE REACTIONS TO WORK REFORMS

LARRY W. HUNTER, JOHN PAUL MACDUFFIE, and LORNA DOUCET*

This paper examines employee reactions to the introduction of work teams,reduced job classifications, and skill-based pay as established through theModern Operating Agreement (MOA) between Chrysler Corporation and theUnited Auto Workers. Survey data suggest that workers responded favorably tothe MOA across six diverse manufacturing plants, despite variation in foundingconditions. The authors draw on field research to assess differences in effectsacross individual plants. Individual attitudes were more negative in plantsfacing the threat of sell-off, although individuals in those plants also reportedengaging in more of the team-based behaviors required by the MOA. Individualresponses to the MOA also varied by demographic characteristics, and byperceptions of the MOA’s impact on various individual, group, and organiza-tion-level outcomes.

*Larry W. Hunter is Assistant Professor of Manage-ment at the Wharton School, University of Pennsylva-nia. John Paul MacDuffie is Associate Professor ofManagement, also at the Wharton School. LornaDoucet is Assistant Professor of Business Administra-tion at the College of Commerce and Business Ad-ministration, University of Illinois at Champaign-Ur-bana.

The authors thank the Chrysler Corporation, whichprovided funding for the employee attitude surveyused here, and Chrysler and United Auto Workersrepresentatives at the National Training Center andat the six Modern Operating Agreement (MOA) plantsfor their assistance with this research. They alsothank their colleagues at M.I.T. and George Washing-ton University who were part of the broader researchteam studying the MOA (Ann Frost, Susan Galloway

Goldberg, Thomas Kochan, Malcolm Lovell, AndrewMartin, and Robert McKersie) with funding from theDepartment of Labor. Finally, they appreciate help-ful comments from Peter Cappelli, Timothy Chan-dler, Robert House, Thomas Kochan, RobertMcKersie, Libby Scott, Susan Schwochau, RebeccaThacker, Adrian Tschoegl, and participants in semi-nars at M.I.T. and Wharton. The International MotorVehicle Program at M.I.T. and the Jones Center atWharton provided generous support.

The survey data in this study were collected withthe support of Chrysler and the United Auto Work-ers. The authors expect to be able to make the dataavailable to other interested researchers, but willneed approval of Chrysler and the UAW before theydo so. All statistical results reported in this study wereproduced with the software Stata.

he introduction of new forms of workorganization became a commonplace

occurrence in the 1980s and 1990s for manyAmerican unionized firms, particularly inthe manufacturing sector. Evidence sug-gests that the implementation of teams andother flexible work practices led to im-

provements in productivity and quality inmany settings. Relatively less, however, isknown about how these changes affectedworkers.

In this paper, we directly examine em-ployees’ reactions to work reforms. Thereforms were embedded in a broader effort

EMPLOYEE REACTIONS TO WORK REFORMS 449

to transform industrial relations at diversemanufacturing plants within the samehighly unionized company. Our goal is tounderstand why some workers and someworkplaces reacted more positively (or nega-tively) than others to these new ways ofworking.

Specifically, we study the views of em-ployees who worked under the ModernOperating Agreement (MOA) between theChrysler Corporation and the United Auto-mobile Workers (UAW). Chrysler, likemuch of the U.S. automobile industry, hadtypified the traditional, New Deal approachto labor-management relationships: mul-tiple job classifications, wages tied closelyto jobs, an “obey-now-and-grieve-later” ap-proach to shop floor conflict, and the con-centration of business strategy and processimprovement decisions in the hands ofmanagement. Yet in the late 1980s, Chryslerand the UAW attempted to reform indus-trial relations at six of their plants throughthe collective bargaining contract. TheMOA reduced job classifications, tied payto skills within those classifications, estab-lished joint consultation committees, and,most significantly, reorganized work intoshop-floor teams.

Understanding the reactions of theChrysler work force helps to shed light onimportant debates in contemporary indus-trial relations regarding the diffusion ofnew work practices and their effects onworkers. Did workers prefer the MOA tothe old system? Was the MOA better re-ceived in some plants than others, and if so,why? What factors influenced workers’ re-actions to these sweeping changes? Whichworkers and plants were more likely to sup-port the new model, and which less likely todo so? We address these questions with asurvey of workers who had experience un-der both systems.

Workers’ Responsesto New Work Practices

Studies demonstrating associations be-tween new work practices and superior firmperformance proliferated in the 1990s.(MacDuffie 1995; Ichniowski, Shaw, and

Prennushi 1997; Huselid 1995; Cooke 1992;Arthur 1992. For reviews, see Ichniowski etal. 1996; Becker and Gerhart 1996.) Dur-ing this same period, as Godard and Delaney(2000) noted, relatively scant research at-tention was devoted to the question ofwhether changes toward the high-perfor-mance workplace paradigm truly bringabout the mutual gains touted by theparadigm’s proponents (for example,Kochan and Osterman 1994). In short,despite continued diffusion of these prac-tices, we know little about how workersresponded to them.

Critics of the new industrial relationsand, more generally, of cooperative labor-management efforts and the team concept(Babson 1995; Compa and Riesman 1985;Parker and Slaughter 1988) argue thatworker reactions are often more negativethan management rhetoric would suggest.The slowness of the diffusion of new workpractices across the American economyfurther suggests that workers, particularlyunionized workers, could be skeptical as towhether changes in work practices havemuch to offer them (Godard and Delaney2000). From this perspective, workers op-pose work reorganization after a realisticappraisal of what is potentially lost from thetraditional system—for example, seniorityrights, work rule protections against arbi-trary management action, clear guidelinesfor regulating work pace, and a vigorousgrievance process. Even where new workpractices are implemented, what appearsto be willing adoption by workers may sim-ply reflect compliance born of fear aboutjob loss or punitive curtailment of pay andpromotion opportunities.

On the other hand, there is some evi-dence that American workers are favorablyinclined toward teams, joint decision-mak-ing processes, and other new work prac-tices. Many studies at the firm level haveshown that workers respond favorably toopportunities for involvement (for reviews,see Cotton 1993 or Pearce and Ravlin 1987),although such opportunities are not alwayslinked to systemic work reforms. The 1994Worker Representation and ParticipationStudy (WRPS) probed the reactions of a

450 INDUSTRIAL AND LABOR RELATIONS REVIEW

national sample of American workers tomore fundamental changes in work organi-zation and decision processes and foundstrong support for greater involvement inmanagement decisions affecting their dailywork experience (Freeman and Rogers1999). Significantly, unionized workers inthe WRPS sample strongly supported ef-forts to improve labor-management coop-eration, while also expressing a strong pref-erence to retain union representation.Union members saw these two avenues toemployee “voice” as complementary andnot contradictory. Appelbaum et al. (2000)also reported evidence that workers in jobswith more opportunities to participate indecision-making find those jobs more in-trinsically rewarding and have higher levelsof job satisfaction and organizational com-mitment, without higher stress.

Thus the issue of worker responses towork reorganization goes to the heart ofdebates about the merits, flaws, andsustainability of these transformation ini-tiatives. The mixed findings of earlier re-search further underscore the necessity ofunderstanding the contexts in which work-ers respond to changes in traditional indus-trial relations; for example, Appelbaum etal. (2000) identified significant differencesacross industries in workers’ responses tonew work systems. We therefore apply a“meso-level” perspective to the study ofworker attitudes toward the MOA (Cappelliand Sherer 1991; House, Rousseau, andThomas-Hunt 1995). We seek to explainworkers’ responses to change by develop-ing an account of the plant-level contextsurrounding the MOA, focusing on twofactors: the plant’s competitive positionvis-à-vis the broader business environment;and the implementation strategy for theMOA. First, we compare those plants whosecompetitive position was especially precari-ous with those whose future with Chryslerwas more secure. Second, we compareplants whose workers initially rejected theagreement with plants whose workers ap-proved the agreement on their first vote.We also use the attitude data to probe work-ers’ perceptions about the various effectsof the MOA, examining the connections

between these perceptions and workers’overall reactions to this work reform initia-tive.

The Modern Operating Agreementsbetween Chrysler and the UAW

We begin with background informationon the Modern Operating Agreements(MOAs); a full description of the labor-management context surrounding the cre-ation of the MOAs, and of early MOA imple-mentation at different plants, is reportedin Lovell et al. (1991).1 MOA implementa-tion began at six plants at various timesbetween 1987 and 1991. The attitude datareported here were gathered in January1993, when all the plants had at least a yearof MOA experience.

Top officials of Chrysler and the UAWestablished the general terms of the MOAin early 1986, with local contracts contain-ing specific MOA provisions negotiated andratified in late 1986 and 1987. By this time,Chrysler had fully recovered from its brushwith near-bankruptcy in the late 1970s andhad achieved record-breaking economicperformance in 1984–85. This gave bothparties the stability and sense of security to

1We worked as part of a team of researchers fromGeorge Washington University (GWU) and the Mas-sachusetts Institute of Technology (MIT) that wasinvited to study the implementation of MOAs by thetop-ranking officials at Chrysler and the UAW. Thisteam (identified in the acknowledgments) receivedfunding from the Department of Labor for this re-search. Planning meetings for the project began inthe summer of 1987, just after all six MOA plants hadratified the agreement. After initial interviews withsenior management and union officials involved inthe negotiations, researchers began tracking imple-mentation efforts at each of the six plants, primarilythrough periodic plant visits and interviews but alsowith an initial attitude survey that did not investigatethe MOA directly but sought to identify key aspects ofworker satisfaction and the nature of the labor-man-agement climate. Lovell et al. (1991) is a report to theDepartment of Labor that summarizes the findingsand recommendations of the full research team tothat point. The attitude survey that provides the datafor this paper was one of a number of follow-upactivities carried out by the research team in 1993 and1994 at the request of Chrysler and the UAW.


pursue the MOA. Each believed that theseplant-level changes would improve produc-tivity and quality still further and hencepreserve jobs and employee earnings.

Prior to introducing the MOA, Chryslerwas the only one of the Big Three U.S. automakers not to have attempted company-wide work reform in the early 1980s. Gen-eral Motors had initiated Quality of WorkLife programs; at Ford, the reforms wereknown as Employee Involvement. UAWofficials leading the Chrysler departmentwere eager to promote “industrial democ-racy” through the MOA. Chrysler manage-ment, for its part, anticipated that forth-coming plant-level capital investmentscould be used as leverage to promotechanges toward flexible work organization.

By 1987, six plants, representing 25% ofChrysler’s manufacturing operations, hadratified MOA agreements. Table 1 shows

key characteristics of these six plants, in-cluding the event that served as the impe-tus for each plant being selected as a candi-date for the MOA.2 From the start, MOAswere designed to function under joint la-bor-management steering committees atboth the plant and national levels. Chryslerand the UAW appointed equal numbers offacilitators, at both levels, to oversee MOAimplementation, under the oversight of thesteering committees and the (also jointlyrun) National Training Center.

The central feature of the MOA was thereliance on work teams as the fundamentalunit for work organization. Teams weregroups of 15–20 employees, formed based

Table 1. Characteristics of Plants.

Plant (State)

New JeffersonCastle Indianapolis Huntsville Trenton Newark Avenue

Characteristic (IN) (IN) (AL) (MI) (DL) (MI)

Product Steering and Engine Electronic Engine Auto Autosuspension castings components machining assembly assembly

parts and assembly

Impetus for Performance Introducing Starting Receiving Receiving StartingMOA Agree- concerns new process production a new engine a new productionments technology in a new product vehicle of a new

plant product product in aline new plant

Year Plant(s) 1906 1901 1952, 1965, 1952 1951 1907; 1991Built 1977, 1988 (Jefferson North)

Number ofHourlyEmployees1/1/91,includingTemp. Layoff 1,100 990 2,663 2,524 3,859 3,375

Average Age 47.7 43.4 41.4 43.2 43.2 48.6

AverageSeniority (years) 23.6 15.6 10.1 18.8 16.6 24

Date of MOAContract 12/22/86 10/20/86 8/10/86 8/28/86 7/25/87 7/25/86

MOA Kickoff 8/26/87 11/9/87 11/12/87 7/27/88 2/5/88 12/14/87

Teams Winter Summer 8/5/88 3/28/91 12/88– WinterDesignated 87–88 88 1/89 1991–92

2See Lovell et al. (1991) for more details on theevents that precipitated the MOA.


on what the agreement called “logicalgroupings of jobs.” Teams were to takeresponsibility for performance outcomes,and to oversee effort, skill acquisition, andperformance. Each team elected a teamcoordinator who took on some responsi-bilities formerly assigned to supervisors andreceived an extra 10 cents per hour in pay.The MOA contract specified a list of 20duties that teams were required to performon a daily or weekly basis (see the appen-dix); all team members were trained inthese duties. Industrial engineers in eachplant initially set team boundaries, butthen workers chose which team to be on,bidding into team positions in order ofseniority on a special “Sadie Hawkins Day.”

To achieve greater flexibility in the allo-cation of labor, the MOA collapsed a hugenumber of job classifications into a fewbroader categories. For example, at theJefferson assembly plant, 87 semi-skilledjob classes were reduced to three techni-cian positions. The wage structure changedcorrespondingly. No longer did each jobcarry its own pay level. Rather, team mem-bers received Capability Progression Pay(CPP), which allowed them to add to theirbase wage by demonstrating mastery of ad-ditional tasks on their team or adjacentteams.

Other changes included a reduction inmanagerial overhead by increasing the ra-tio of first-line supervisors to workers; theelimination of many status markers (suchas neckties, management cafeterias, andreserved parking); and encouragement ofa participative culture, in which ideas andsuggestions could be raised freely. Thesework reforms were typical of the late 1980s,when the influence of Japanese-type pro-duction systems was strongly felt in the U.S.auto industry (Katz and MacDuffie 1994).But the use of the collective bargainingprocess to establish the MOA meant thatstructural changes in the organization ofwork—the construction of teams, the re-duction of job classes, and the new paysystem—could be carried out systematically,with full backing from the national unionand with the help of the joint steeringcommittees and the matched sets of labor

and management MOA facilitators. Al-though this approach was also being triedduring this period in other industries, suchas steel, it was rare in the auto industry.

The promises about what the MOA couldachieve appealed to many workers, but amajority were clearly skeptical as imple-mentation began, and some greeted thechanges with anger and overt resistance.3

Generally, the high-seniority, experiencedworkers at these plants did not initiate orembrace this work reform initiative; rather,it was thrust upon them.

Reactions to the MOA:Plant-Level Factors

The MOA represented an attempt toimplement sweeping changes in work sys-tems across six plants that differed in theirproducts, production processes, competi-tive business environment, and labor rela-tions history. In this section, we considerthe plant-level factors affecting worker re-sponses to these changes, paying attentionto where these factors might influenceworker attitudes (their affective responses)and activities (their behavioral responses)similarly or differently across plants. Ourdata on worker behaviors are self-reported,as described below; nevertheless, we be-lieve the distinction between attitudes andbehaviors is conceptually important, andwe will show below that there are differentpatterns in the data for these two outcomes.

Business Environment

We consider the business environmentfacing each plant in order to understand

3These impressions are based on interviews at theMOA plants that began shortly after the ratificationvotes for the MOA contract (1987–88) and continuedthrough the period of early implementation (1989–91) and up to the time of the attitude survey in early1993. One of the best examples of successful resis-tance to the initial MOA contract involves skilledtrades workers. As far as we know, planned efforts tocollapse job classes for maintenance workers and tocross-train them for rotation across a variety of main-tenance jobs were never implemented.


how much the sense of competitive threatto the plant’s survival might have affectedworker reactions to the MOA. Chrysler,like many U.S. corporations during thisperiod (Kochan, Katz, and McKersie 1994),was articulating a strategy of outsourcingall non-core activities. This strategy af-fected two of the MOA plants (New Castleand Huntsville) directly because they wereaffiliated with Chrysler’s independent partsdivision, Acustar. In March 1991, a Chryslersenior officer announced that Acustarplants were likely to be sold. The UAW hadnot been informed of this decision beforethe announcement, and responded imme-diately with angry statements to the pressand a cessation of MOA activities at all sixplants. After a few months of negotiations,this issue was resolved, with a decision tosell four non-MOA Acustar plants and apledge by Chrysler not to sell the remain-ing Acustar plants except in dire economicconditions.

Despite the reassurances won by the UAWfor their plants, workers at New Castle andHuntsville may have subsequently felt lesssecure about Chrysler’s commitment totheir plants and about the long-term pros-pects for their jobs. What effect might thishave had on their reactions to the MOA?Konzelmann Smith (1996) suggested thatmanagerial initiatives focused onoutsourcing may undermine labor-manage-ment cooperation. Jacoby (1983), simi-larly, suggested that while some competi-tive pressures may be necessary to motivatecooperation, high levels of insecurity mayundermine workers’ acceptance of newwork practices. On the other hand, whilehigh levels of job insecurity may under-mine workers’ expressed enthusiasm fornew work practices, such competitive threatsmay also increase the feeling of urgencyabout making such practices work effec-tively and hence boost workers’ effort lev-els. Cappelli (1999) noted that even whereworkers have low trust in their employingorganizations, they may display high levelsof workplace effort and commitment totheir teammates. Drago (1996), similarly,identified two approaches to participatorymanagement: “transformed” workplaces

offering assurances of continued employ-ment to workers; and “disposable” work-places motivating employee participationthrough fear of layoff or plant closure.

Workers’ compliance with participatoryprograms driven by competitive threat maynot always yield the benefits anticipated bya “mutual gains” approach. In fact, there isno evidence that participatory practices are,on average, coupled with employment se-curity (Osterman 2000), and Eaton (1994)found no link between union leaders’ per-ceptions of security and the survival of par-ticipation programs. Workers’ affectiveresponses to innovative work practices maybe negative if they perceive that competi-tive conditions have made for a “dispos-able” work force; nevertheless, those samecompetitive conditions may provide morepowerful incentives for actual participa-tion.

At all six Chrysler plants, the adoption ofthe MOA was driven in part by pressuresarising from the broader competitive envi-ronment. However, while the rhetoric ofcompetitiveness surrounded the entireMOA effort, workers at the regular (non-Acustar) Chrysler plants had multiple rea-sons to believe that their plants’ immediatefuture was secure, particularly once theyratified the local MOA agreement. TheAcustar plants could feel no such assur-ance. We suggest that in these plants, work-ers may have been less satisfied with theMOA, but more compliant with the rules ofthe new contract. More formally, the hy-potheses are

H1a: Workers in plants threatened with a sell-offwill express more negative attitudes toward theMOA than will workers in non-threatened plants.

H1b: Workers in plants threatened with a sell-offwill display higher levels of effort in MOA-relatedactivities than will workers in non-threatenedplants.

Implementation Process

Walton, Cutcher-Gershenfeld, andMcKersie (1994) drew a distinction between“fostering” and “forcing” strategies forimplementing workplace changes affect-


ing labor-management relations. Theimplementation process for the MOA var-ied across the six plants with respect tothese strategies. At the national level, co-operation between the UAW and Chryslerin establishing the overall contract languageand governance structure for the MOAcould be characterized as a “fostering” strat-egy. Yet some elements of a “forcing” strat-egy were also required, with both Chryslermanagement and national UAW leadersplacing pressure on local union leadersand workers to adopt the MOA. Given theevidence that sustaining work reforms inunionized settings requires the commit-ment of both labor and management and apositive labor relations climate (Eaton 1994;Kochan and Osterman 1994), it is impor-tant to understand the effect of this mix offostering and forcing strategies.

New Castle Machining in Indiana wasthe only plant to volunteer for the MOA,and it did so well before the threatenedAcustar closings. Local plant managersand union officials agreed to respondproactively to this provision in the nationalcontract, and New Castle was the first plantto establish its work teams, just one yearafter approving the MOA. In three otherplants (Jefferson Assembly in Detroit,Huntsville Electronics in Alabama, and In-dianapolis Foundry), the initial responseto corporate-level inquiries about interestin the MOA was tentative and apprehen-sive, but the workers voted to adopt the newagreement. At two plants (Newark Assem-bly in Delaware, and Michigan’s TrentonEngine plant), however, opposition to theMOA was strong and immediate. At both ofthese plants, when the MOA contract wasbrought up for approval, local union mem-bers voted it down. Following these votes,Chrysler management made it clear, instrongly worded statements (and with sup-port from national UAW officials), thatadoption of the MOA was a prerequisite forcorporate investments in new technologyand new products. Subsequently, workersat these plants voted again and narrowlyapproved the MOA.

The initial voting down of the MOA agree-ment at Newark and Trenton could, for

several reasons, have presaged negativeworker reaction to the new work practices.Such a vote could have been an indicatorthat relatively low levels of trust alreadyexisted between labor and management;Cappelli and Sterling (1988) found thatpoor labor relations were a significant pre-dictor of failure to ratify labor contracts inthe auto industry in the early 1980s. Fur-ther, the “forcing” strategy itself may haveled to decreased trust between the parties(Birecree 1993; Walton, Cutcher-Gershenfeld, and McKersie 1994).

The plants that responded more posi-tively to the MOA from the beginning, onthe other hand, may have had a more posi-tive labor relations climate. The initialpositive vote may also have created a higherlevel of commitment to making the MOAwork. Kim (1999), for example, showedthat a majority vote in favor of a gain-sharing program is positively associated withits subsequent performance and survival,and argued that stronger initial commit-ments make programs more robust(Goodman and Dean 1982) and may evenlead to “escalation of commitment” (Stawand Ross 1978) should the program falterinitially. In both situations, we expect theimplementation process for the MOA tohave similar effects on worker attitudes andworker behaviors. Hence our hypothesesfor the implementation process are as fol-lows:

H2a: Workers in plants that experienced a moreextensive “forcing” strategy will express more negativeattitudes toward the MOA than workers who experi-enced a more “fostering” strategy.

H2b: Workers in plants that experienced a moreextensive “forcing” strategy will display lower levels ofeffort in MOA-related activities than workers whoexperienced a more “fostering” strategy.

Summary. We expect workers’ affectiveresponses to the MOA to be significantlymore negative at the Acustar plants, wherethe competitive threat had a negative effecton the labor-management climate. We alsoexpect such responses to be more negativeat the two plants (both non-Acustar) thatvoted down the MOA. These votes led tothe implementation of work reforms only


through a “forcing” strategy that may haveeroded trust between the parties in an al-ready troubled labor-management relation-ship. We also hypothesize that under con-ditions of severe competitive threat, work-ers will be more compliant with the re-quirements of work reform. In contrast,after workers experience a forcing strategy,we hypothesize that they are likely to ex-press negative reactions to work reform intheir actions as well as their attitudes.

Other local differences, beyond thesetwo factors, could also have led to differ-ences in workers’ responses to the MOAthat were unique to each plant. We con-sider this possibility in our data analyses by

including plant dummy variables. In “Sum-mary of Plant-Specific MOA Experiences”(above), we also provide contextual detail,drawn from our fieldwork, about the sixplants (excerpted from Lovell et al. 1991).

Reactions to the MOA:Individual-Level Factors

We anticipate, based on the foregoingconsiderations, that workers’ responses tothe MOA will differ systematically acrossplants. We also consider individual-levelfactors that might have influenced work-ers’ views of the MOA. To the extent thatthese individual factors differed across the

Summary of Plant-Specific MOA Experiences

This summary describes differences in the MOA experience across the six individual plants, both asbackground and to shed further light on selected findings from the survey data. Our account here draws onsite-based research that is described in more detail in Lovell et al. (1991) as well as on the findings from thesurvey.

New Castle Machining. New Castle was the one plant that volunteered for the MOA. The work forceperceived that a shift in Chrysler’s products from rear-wheel to front-wheel drive was going to leave the plantwithout support for its product line. The local union held a series of meetings with members to discussoptions, and they approached Chrysler asking to open their contract for negotiations in the interest ofgaining limited job security. Chrysler offered them an opportunity to become an MOA plant, following thepattern set at four other plants. Only three weeks elapsed between the start of negotiations and the final vote,yet the MOA was supported by 71% of New Castle’s workers. Due to this rapid pace, there was some initialconfusion about the MOA, but previous positive experience with a jointly sponsored quality program helpedpave the way. New Castle became the first plant to do MOA training and to implement teams.

New Castle’s survey responses indicate the highest number of team duties performed among the sixplants, and a very strong belief that the MOA helped the plant’s economic performance. The threatened saleof Acustar plants, however, worked against other positive factors at New Castle. Two other factors may havealso diminished workers’ enthusiasm at New Castle. First, to preserve team stability, the MOA established arule that moving to a different job on the basis of seniority could only happen once per contract (as opposedto the old norm of once a year); the enforcement of this rule actually provoked a wildcat strike. Second,layoffs (due to reduced demand) interfered with MOA training and the progress of teams with problem-solving.

At a key moment of tension, a visit by plant managers and union officials to a successful team plant atGeneral Motors (in Shreveport, La.) helped put the MOA back on track. Still, given the tensions arising fromperceived encroachments on seniority rights and the aura of threat that affected New Castle as an Acustarplant, it seems likely that workers who did not believe the plant’s economic competitiveness had beenimproved would have opposed the MOA particularly strongly.

Indianapolis. “Indy Foundry,” as it was known, was viewed as a highly successful plant in terms of qualityand productivity in the mid–1980s. However, the plant’s technology was outmoded, placing it at competitiverisk. Thus the link between the MOA and corporate investment in new product technology was highly salientto workers and plant managers alike, and the MOA contract was approved by a 2-to–1 margin. Theimplementation of the new technology led to delays, however, and teams were not designated until nearlytwo years after the MOA contract was signed.

Indy was unique in part because of its group incentive pay system (bonuses for all if production on a givenshift exceeded quota), which dated back to the 1940s. This incentive plan was retained, rather than beingreplaced by the Capability Progression Pay plan used at the other MOA plants. The new contract added abase level of pay guaranteed regardless of technical problems that hampered fulfillment of productionquotas. This guarantee proved necessary, as problems with the new technology hurt plant performance overthe next year. Despite the guarantee, workers were highly dissatisfied with their pay during this period, since


six plants, they might serve as mediatingvariables between workers’ responses andany plant-level effects we identify (Baronand Kenny 1986). We organize the indi-vidual-level factors we measure through theMOA attitude survey into four categories:demographic characteristics; the perceivedimpact of the MOA on individual economicoutcomes (pay and job classification); theperceived contribution of the MOA to im-proved job design; and the perceived con-tribution of the MOA to collective employeevoice. In this section, we offer hypothesesrelated to each category. For the sake ofsimplicity, we do not distinguish betweenaffective and behavioral worker responses

in these hypotheses; we expect that affec-tive and behavioral responses will move inthe same direction for each of these fac-tors.

Demographics. Employee reactions to or-ganizational change might vary based ondemographic characteristics, though thishas been studied only rarely (Cotton 1993).Age may matter: Blumberg (1980) foundthat younger workers were more likely thanolder workers to engage in job switching inautonomous work groups in a coal mine,while Ketchum and Trist (1992) observedthat older workers were likely to be moreresistant to the introduction of teams. Olderworkers also had more to lose with the

bonuses formerly taken for granted became rare and difficult to attain. Eventually the technical problemssubsided, and by 1991 the guaranteed base pay was eliminated.

Pay and performance remained a sore spot at this plant. The survey results showed that workers at IndyFoundry were much less likely than those at other plants to say that their pay level had increased due to theMOA, and saw much less impact of the MOA on organizational performance. However, those who did seean increase in their pay level were very supportive of the MOA. Older workers were also positive about theMOA at Indy Foundry, in contrast with their counterparts at other plants, probably because the consolidationof job classifications combined with the continuation of the old pay system meant there was less need for themto take on the most physically difficult jobs in order to maximize their income. Instead, these jobs went toyounger workers, who were not able to earn the same pay premium as in the past and were hence less pleased.

Huntsville (Al.) Electronics. Huntsville was the plant with the most consistently negative attitudes towardMOAs and teams, though its reported average of team duties accomplished was higher (14.4) than at anyother plant except for New Castle. The implementation of the MOA at Huntsville was associated with a moveinto a new plant, which made for a very chaotic transition, and the initiative remained contentious after themove.

The roles of team leaders vis-à-vis traditional foremen were central to the Huntsville experience. The firstgeneration of team leaders clearly felt they were supposed to take charge and ensure that teams fulfilled theirTwenty Team Duties as quickly as possible. The relatively high levels of team duties performed suggest thatthe leaders were successful in this regard. Team members, however, rebelled against having the leaders actlike “junior foremen” and insisted on being able to replace team leaders based on a majority vote at any time.As resistance to the MOA initiative grew, heightened by the perilous position of Huntsville as an Acustarplant, the team leaders became a lightning rod for work force frustrations. The turnover of team leadersreached extremely high levels. Eventually the work force began to unload its frustrations on the unionleadership, and officials supportive of the MOA were voted out of office, though the new leadershipcontinued to work under the framework the MOA had established.

Trenton Engine. Trenton voted down the MOA in its initial local union vote by a 2-to-1 margin. Plantmanagers, with the support of the local union president, were pursuing the MOA in order to bring a newproduct line to the plant. They were surprised by the negative vote, which resulted from a small turnout inwhich skilled trades workers—who opposed the collapsing of job classifications that would accompany theMOA—were heavily represented. A subsequent all-plant meeting communicated the importance of the MOAto the plant’s survival. After a petition drive among non-skilled-trades workers, a second vote was held andthe MOA was supported by a 3-to-1 margin. However, the lingering effects of the initial vote were felt in thenext local union election, when an outspoken opponent of the MOA from the skilled trades was elected.Partly due to his influence, and partly due to repeated turnover in the plant manager position, MOAimplementation was continually delayed at Trenton. The MOA kickoff ceremonies occurred nearly two yearsafter the contract signing, when the new engine line was finally introduced, but teams were not establisheduntil nearly three years later. As at the other MOA plants, skilled trades workers were never placed in teams,and thus their negative feelings about the MOA are not reflected in our survey data.

Performance problems followed the full implementation of teams as workers rushed to learn theadditional skills that would maximize their pay under CPP. These problems were still relatively recent,


diminution of seniority. We also know thatworkers with more education are typicallymore likely to undertake training (Altonjiand Spletzer 1991); Chrysler workers withmore education might thus be more posi-tively inclined toward an initiative like theMOA, which included not only requiredtraining, but also opportunities to gain skillsthrough the CPP system.

Responses may also have differed by gen-der. While little prior literature exists toguide us here, we note that women com-prised only a small share of the work forcein all plants save Huntsville. Due to thisminority status, women may have felt lesscomfortable than men in a team-based work-

place. For example, Thompson and Gooler(1996) argued that while diversity can en-hance the performance of work teams, teammembers who do not fit in with the majoritymay have more negative attitudes towardthe team.

Thus, with respect to demographics, wehypothesize that:H3a: Younger workers will react more positively to theMOA than older workers do.

H3b: More educated workers will react more positivelyto the MOA than less educated workers do.

H3c: Men will react more positively to the MOA thanwomen do.

Individual attitudes toward the MOA

though mostly resolved, at the time the survey data were collected. Hence it is not surprising that Trentonworkers were the least likely to see a link between the MOA and improved organizational performance.

Trenton workers had more favorable views of the team aspects of the MOA than did workers at any otherplant. The field research did not suggest clear reasons for this difference. It may have reflected the contrastbetween worker experiences and the highly negative expectations that developed during the long delay, orit may have been due to particular aspects of the previous regime. In any case, the positive view of teams helpsexplain why Trenton workers reported such positive attitudes toward the MOA despite their skepticism aboutits effect on economic performance.

Newark (Del.) Assembly and Jefferson (Mich.) Assembly. These two assembly plants had very different foundingconditions. Newark initially voted against opening its contract in order to negotiate the MOA, a voteconsistent with Newark’s reputation for highly adversarial labor relations. In fact, labor relations at the planthad been improving in the years before the MOA, a process that began, as at New Castle, with a jointlysponsored program on quality improvement. The vote more likely reflected the work force’s negativereaction to Chrysler’s strong language linking the MOA vote to keeping the plant open, and to its insistenceon scheduling the MOA contract vote before, rather than after, upcoming local union elections.

After the initial negative vote, the president of the local union worked with plant management and withvarious factions within the union to gain support for a second vote. The parties mounted an extensiveinformation campaign to correct misperceptions about the MOA. On the second vote, the margin of supportfor opening the contract was high, and this level of support was maintained during the subsequentnegotiations on specific MOA details and the contract ratification. Indeed, the union president was re-elected to a third term immediately after the MOA contract was signed. Subsequently, implementationdifficulties arose due to the MOA being launched concurrently with a new product line, but these weretypically resolved jointly, with the union taking an active role in pressuring management to fix variousproblems. Plant managers reported that this local union pressure often helped them break through delaysin getting needed support from Chrysler at the corporate level.

In contrast, Jefferson approved the MOA by a relatively large margin. Workers knew that this approvalwas required as a condition for Chrysler’s investment in a new plant, Jefferson North, located just a few milesfrom the original Jefferson Avenue plant, the oldest in the country at the time of its closing. Indeed, a keyprovision of the Jefferson local contract for the MOA was an agreement that implementation would not beginuntil production began in the new plant—an event that would not occur until the winter of 1991–92, 5- 1/2years after the contract signing and 4 years after the MOA kickoff ceremonies. The old Jefferson plant closedin February 1990, earlier than originally planned, and the local union, displeased by this, put all pre-MOAtraining activities on hold. Yet this training resumed as planned in the months before Jefferson Northopened, and the confluence of the new plant, the new product (Jeep Grand Cherokee), and the MOAgenerated tremendous excitement and pride.

The survey data suggest that despite these differences, the assembly plants had roughly similar experiences interms of worker reactions. Workers were generally favorable to the MOA, and believed strongly that the MOA wasresponsible for improvements in plant performance. In neither plant, however, were workers keen on performingthe 20 team duties. It may have been more difficult for teams under the assembly line production process to findas much time for the team duties as did teams in less constrained work environments.


might also have been influenced by per-sonal economic outcomes achieved as aresult of the agreement. A prominent fea-ture of the MOA, and an important negoti-ating goal for the UAW, was the guaranteethat no worker would face a cut in pay or amovement to a lower job class. New broadjob classes were set up to reflect the highestearlier job class folded into the new cat-egory, resulting in an increase in task re-sponsibility and job status for the majorityof workers. Similarly, the pay rates attachedto these new job classes provided a payincrease for most workers, albeit a modestincrease overall and hence a quite smallincrease for individuals at the top end ofthe range of the former job classes thatwere being consolidated. Under the Capa-bility Progression Pay plan, workers couldincrease pay over their initial base by suc-cessful mastery of additional skills, yet eventhe base pay level for the new job classes washigher than previous pay. The union’sstated goal was that all workers could po-tentially move to the top of the CPP scaleover time. Our hypothesis can be simplystated, based on a straightforward view ofeconomic self-interest:

H4: Workers whose job class, pay level, or bothincreased after work reforms were implemented willreact more positively to the MOA.

The MOA also had the potential to influ-ence workers’ views positively by improvingjob design. Individuals in jobs with intrin-sically motivating characteristics are typi-cally more satisfied with their work than areother workers (Hackman and Oldham1980). Accordingly, we expect more posi-tive responses to the MOA where job classi-fication consolidation, reassignments, andthe construction of teams under the MOAled workers to perceive their jobs as en-riched—specifically, along the dimensionsof task variety, identity, and significance;autonomy; and feedback from the workitself.

H5: Workers who perceive improvements in their jobcharacteristics will react more positively to the MOA.

Workers’ reactions could also be a func-tion of the impact the MOA had on their

collective voice and future prospects. Con-siderable research suggests that employeesfavor group expression of their concerns atthe workplace (Freeman and Medoff 1984;Kaufman and Kleiner 1993; Freeman andRogers 1999). Such expression may beimportant to workers as a way to addressissues arising from their daily work experi-ence. It may also give them a sense ofefficacy with respect to productivity, qual-ity, and other performance outcomes thataffect the fate of their employer—and hencetheir job security. Here we hypothesizethat:

H6a: Workers who perceive greater collective influ-ence on issues arising from their daily work experiencewill react more positively to the MOA.

H6b: Workers who perceive greater collective influ-ence on the economic performance of their plant willreact more positively to the MOA.

Summary. We will primarily investigatethe role of these individual-level factors aspotential mediators of the plant-level fac-tors identified above. Where worker demo-graphics or perceptions of the MOA’s im-pact do mediate the plant-level factors, theyshould provide additional insight into theconditions favoring the implementation ofwork reforms.

Sample and Survey Methodology

Data for our study come from a tele-phone survey of 2,000 randomly selectedemployees, conducted between January 17and February 5, 1993.4 Larger plants (those

4The timing of the survey was advantageous for tworeasons. First, it was administered five to six yearsafter the initial signing of MOA contracts and, even atthe plant that was slowest to implement the MOA,over a year after the introduction of teams. Thus wewere not picking up the attitudes from a startupperiod, when a Hawthorne effect or initial resistanceto change might have affected workers’ views. At thesame time, the start of the MOA had not been so longago that workers were unable to remember the pre-MOA situation in their plant. Second, January 1993fell about midway between the dramatic recovery ofChrysler from its financial crisis of 1990–91 and itsrecord-breaking sales and profits in 1994. There wasneither the dark cloud of another brush with bank-


with more than 2,000 employees) were sur-veyed more heavily to yield a sample with arepresentative distribution of the six plants.400 employees from the Jefferson (Michi-gan) and Newark (Delaware) assemblyplants, Trenton (Michigan) engine plant,and Huntsville (Alabama) electronics plantwere surveyed, while 200 employees fromthe New Castle (Indiana) machining plantand Indianapolis (Indiana) foundry weresurveyed.

Of the 2,000 respondents, 1,420 wereunionized production workers, 307 wereunionized skilled trades workers, 78 wereunionized technical and clerical employ-ees, 123 were managers, and 72 were fromother categories. The analysis reported inthis paper is restricted to the 1,420 union-ized production workers, the group mostaffected by the MOA. Considering onlythose production workers who confirmedthat they were members of teams reducedthe sample to 1,289. Eliminating surveyswith missing data reduced the sample sizeto 782; there were no identifiable differ-ences between respondents in the finalsample and those with missing data.

The telephone survey consisted of elevenpages of questions designed to find outwhat employees thought of the overall MOAand its many constituent parts. A profes-sional survey research firm carried out theinterviews. Names were picked at randomfrom a list of employees at each plant pro-vided by Chrysler, and the employees werephoned at their homes. A letter from theresearchers, accompanied by cover lettersfrom Chrysler and the UAW, was distrib-uted in each plant a week before calls weremade, making it clear to workers that boththe company and the union approved ofand supported the survey’s effort to “learnabout the views of employees, both positive

and negative, toward the MOA.” The com-plete confidentiality of all individual re-sponses and the research team’s indepen-dence from both the company and theunion were also emphasized in the letter.Interviewers repeated the key informationfrom this letter at the beginning of eachinterview.

Most questions offered respondents twopositive (for example, “agree” and “stronglyagree”) and two negative (for example,“disagree” and “strongly disagree”) poten-tial responses. Both the question and theresponses were read to the respondent bythe interviewer. Interviewers did notprompt respondents with responses such as“neither agree nor disagree” or “partiallyagree, partially disagree.” However, inter-viewers took note of such responses whenrespondents volunteered them, and theseanswers were later coded as “3” on a five-point scale.5 At the end of the interview,respondents were asked open-ended ques-tions about their reactions to the MOA.The telephone interview took 15–20 min-utes to complete.

The design of the survey imposes somelimits on the conclusions we can draw fromthe data. The survey was administered tosingle respondents, without a longitudinalelement. Thus we recognize the danger ofcircularity in the claim that perceptions of

ruptcy nor the euphoria of a huge year-end bonus tocolor worker perceptions of the MOA. Open-endedcomments recorded at the end of each phone surveyreveal that workers were still concerned about em-ployment security but were cautiously optimistic thatChrysler was “out of the woods” in terms of its eco-nomic performance.

5This approach is common practice for telephonesurveys. Not offering a “neither agree nor disagree”response shortens the time needed for the inter-viewer to read all the responses and also reduces thenumber of different responses the interviewee needsto keep in mind. In addition, this approach mini-mizes the likelihood that a respondent will consis-tently choose the middle or “neutral” response be-cause initial discomfort at expressing an opinion setsthe pattern for future responses. However, it doesnot prevent someone who truly “neither agrees nordisagrees” from saying so. When this or any similarcomment was made, it was noted by the interviewerand coded. Thus, while this approach inherentlygenerates greater variance (that is, there are fewer“3’s” when the “neither agree not disagree” approachis not explicitly offered), there is no reason to believethat it prevents respondents from stating their opin-ions clearly or that it distorts or biases the results inany particular direction.


job design, voice, and effects on perfor-mance will influence attitudes toward theMOA. Workers who like the MOA, forwhatever reason, might be likely to ascribepositive effects to it, regardless of whatother data are known to them, and thesurvey does not allow us to rule out thispossibility. Second, we also risk bias fromusing the same method to ascertain bothattitudes toward the MOA and the factorsthat may affect those attitudes.

These concerns are somewhat mitigatedby our plant-level interviews, which sug-gested that workers were capable of distin-guishing between different aspects of theMOA. For example, a worker might favorteams while being skeptical about the MOAoverall, or vice-versa. Furthermore, ourknowledge about the organizational con-text, critical to our interpretations of thedata, was generated by our fieldwork and isthus independent of the survey data. Thesurvey itself also offered multiple, highlyspecific questions about the outcomes thatindividuals might associate with the MOA,encouraging workers to differentiate theirresponses according to their particular ex-periences. In short, while these data pos-sess the limitations common to those col-lected for program evaluation purposes, webelieve they can still generate considerableinsight into the reasons that different work-ers, in different workplaces, respond differ-ently to work reorganization initiatives.

Variables

We use two dependent variables, corre-sponding to affective and behavioral indi-cators of worker responses to the MOA: athree-item composite variable reflectingindividual workers’ attitudes toward theMOA; and the number of the “20 TeamDuties” that workers reported their teamswere carrying out. As explained above, weare examining two independent variablesat the plant level (competitive threat anduse of a fostering versus forcing implemen-tation process) and four categories of inde-pendent variable at the individual level(demographics; individual economic out-comes; job characteristics; and collective

voice and welfare).Dependent variables. To measure attitudes

toward the MOA, we asked respondentsthree different, but related, questions—one general and two explicitly compara-tive. (Distributions of responses are dis-played in Figure 1.)

“Overall, are you satisfied with the way things weregoing in your plant under the MOA?” (1 = “verydissatisfied”; 5 = “very satisfied”)

“I prefer working under the MOA to working underthe old approach.” (1 = “totally disagree”; 5 = “totallyagree”)

“Overall, I prefer the team system to working under theold system.” (1 = “totally disagree”; 5 = “totallyagree”)

Although these three questions wereworded to capture nuances of worker reac-tions to the MOA, correlations among themwere very high. To increase reliability, wecombined the three measures additively(all were scaled from 1 to 5) into a singlescale (Cronbach’s α = 0.81) representingan individual worker’s mean response tothis set of questions.6

To measure behavioral responses to theMOA, respondents were asked how many ofthe 20 team duties were performed by theirteams on a regular basis, from 0 to 20. Thesurvey did not record membership by team,so this measure represents individual per-ceptions of team activities. The reportednumber of duties should provide a roughproxy for individual worker behavior, but itis possible that some workers who per-formed only a small range of the 20 dutiesmay have been on teams that performed alarge share of the team duties (or vice-versa). Systematic biases resulting frommisperceptions, however, seem unlikely: alist clearly specified these duties; MOA fa-cilitators were instructed to make sure thatteams reviewed the list at their weekly meet-

6In analyses available on request, we examine eachof the three questions separately. Small differencesin the patterns of results for each of the three ques-tions suggest that workers did make distinctions andcomparisons among component parts of the MOA.


ing; and highly involved workers on low-involvement teams seem no more likelythan less involved workers on high-involve-ment teams.

Plant-level independent variables. Weoperationalize the competitiveness of theplant’s business environment by coding theAcustar plants as 1 and the non-Acustarplants as 0. We then operationalize whetherthe implementation process at a given plantwas “fostering” or “forcing” by coding thetwo plants (Newark and Trenton) that voteddown the MOA on the initial ballot as 1 andthe four plants that voted in favor of theMOA on the initial ballot as 0.

Individual-level independent variables. Forthe demographic variables, age was takenfrom company records and matched to thelist of individuals. Education was not avail-able from company records and hence isself-reported on an 8-point scale, with 1indicating non-completion of elementaryschool and 8 indicating attainment of apost-graduate or professional degree. Gen-der was coded by the interviewer, with 1indicating a male.

For individual economic outcomes, twoquestions were asked: “Has your job classimproved as a result of the MOA?” and “Hasyour job class improved as a result of theMOA and Capability Progression Pay?”Responses of “Yes” were coded as 1.

For job design, respondents rated theircurrent job along dimensions correspond-ing to skill variety, task identity, task signifi-cance, autonomy, and feedback from thework itself (from Hackman and Oldham’s[1980] short form of the Job DiagnosticSurvey). These items loaded on a singlefactor, which we used to construct a stan-dardized scale. Because virtually every jobfrom the old classification scheme was re-designed as part of MOA implementation,it seems reasonable to ask whether currentperceptions of job characteristics affectworker attitudes toward the MOA.

To assess employee perceptions of col-lective voice and welfare, respondents wereasked to rate “how much attention manage-ment at your plant pays to what your workteam thinks or says” on seven issues: technol-ogy (the use of new technology on your

job); work methods and procedures (the waythe work is done); quality (the quality of theproduct); people problems (handling “peopleproblems”); job assignment (who in yourteam should do what job); work pace (howfast the work should be done); and workeffort (how much work people should do ina day).


Respondents were also asked whetherorganizational productivity, safety, and qual-ity outcomes had improved or worsenedunder the MOA. In factor analyses, “teaminfluence” loaded onto two distinct factors,which we call “team influence on work pro-cesses” and “team influence on work ef-fort.” Perceptions of the influence of theMOA on organizational performance out-comes loaded on a separate, single factor.We constructed standardized scales for eachof these variables based on the factor scores.

Results

Across the sample, most workers pre-ferred the MOA to the previous work ar-rangements. Figure 1 shows that for eachof the three individual questions about theMOA, well over 60% of the workers ex-pressed positive views. Workers respondedmost positively to the question regardingteam preference, with over three-quarters

claiming to prefer teams to the old system,while over two-thirds also expressed a pref-erence for the MOA (67% totally or par-tially agreed). Sixty-four percent said theywere satisfied or very satisfied with the MOA.

In what follows, we consider workers’responses by referring to the compositevariable “employee attitudes toward theMOA” and the self-reported behavioralvariable “team duties done.” The samplemean for employee attitudes was 3.6 on a5-point scale, while the sample mean forteam duties done was 13.8 (of a possible20). Table 2 summarizes the means ofthe dependent and independent vari-ables, both for the sample as a whole, andsplit out by plant.

Tables 3 and 4 show how mean levels ofworker attitudes and behaviors vary by thetwo systematic plant-level differences weidentified; Table 5, discussed below, pre-sents analyses of variances of these responsesat the plant level. Table 3 shows that,

Table 2. Means of Variables.

FullVariable Sample Jefferson Newark Indianapolis Trenton New Castle Huntsville

Satisfaction with the MOA 3.3 3.4 3.6 3.0 3.2 3.6 3.1Preference for the MOA 3.6 3.8 4.0 3.5 3.7 3.7 3.0Preference for Teams 3.9 4.1 4.2 3.9 4.2 4.0 3.2

Reaction to the MOA(composite attitude variable) 3.6 3.8 3.9 3.5 3.7 3.8 3.1

Number of 20 Team DutiesPerformed 13.8 13.7 13.3 12.9 13.1 15.7 14.4

Gender (1 = male) .81 .92 .93 .98 .79 .99 .39Age 45.5 48.9 43.8 44.3 44.8 47.5 43.3Education (1 to 8 scale) 4.2 4.2 4.3 4.3 4.2 4.0 4.3

Pay Level Increases .71 .49 .78 .46 .82 .76 .87Job Class Improves (1 = yes;2 = no) .35 .33 .40 .24 .41 .35 .32

Job Enrichment(standardized scale) 0.0 –.02 .08 –.28 .04 .15 –.07

Team Influence on Process(std. scale) 0.0 .16 .03 –.10 .04 .00 –.23Team Influence on Effort(std. scale) 0.0 –.08 –.12 –.14 .06 .17 .15

MOA ImprovesOrganizational Performance(std. scale) 0.0 .20 .20 –.36 –.50 .42 –.13

Number of Respondentswith Complete Data 782 181 174 63 133 85 146


surprisingly, workers in the plants workersin the plants that entered into the MOA asa consequence of a more explicit “forcing”strategy expressed more positive opinionsabout the MOA than did workers in theother plants. Table 4 shows that workers inthese plants, however, also reported thattheir teams performed fewer of the 20 teamduties. Table 3 also shows, less surprisingly,that workers in the two plants facing thethreat of sell-off had more negative atti-tudes about the MOA than did workers inother plants. Table 4 shows, however, thatthese workers reported performing moreteam duties than did workers in the otherplants. This pattern of results is consistentwith three of our four plant-level hypoth-eses (H1a, 1b, and 2b), and supports theexpectation that affective and behavioralresponses to the MOA might move in dif-ferent directions under the “competitivethreat” condition.

Table 5 displays the results of one-way,two-way, and nested analyses of variance forthe plant-level differences that are pre-sented in Tables 2, 3, and 4. The one-wayanalyses in Table 5a are equivalent to simpletests of differences in means across theconditions and plants; in each case, thedifferences are statistically significant atconventional levels. In Table 5b, the two-way analyses of variance for each depen-dent variable show significant differencesbetween workers’ responses in the plantsfacing the threat of sell-off and responsesin the plants that did not face this threat.Once this factor is considered, the use of aforcing strategy at a particular plant has no

additional explanatory power with respectto attitudes or behaviors.

In Table 5c, we also examine the addi-tional explanatory power of plant-specificfactors by adding the plant dummies. Wedisplay only analyses for the “competitivethreat” variable, because, as the results inTable 5b show, the “forcing” strategy vari-able has no explanatory power once thecompetitive threat is taken into account.The results of the nested analysis of vari-ance suggest that plant-level idiosyncrasiesdo explain further differences in workers’attitudes. For team duties done, however,the inclusion of dummy variables for indi-vidual plants offers no additional explana-tory power once the threat of sell-off isconsidered.

The substantial differences in means formany of the individual-level factors acrossthe six plants, as shown in Table 2, suggestthat some of these factors could mediatethe plant-level effects identified in the analy-ses of variance. We explore the directeffects of these variables and the possibilityof mediation effects through ordinary leastsquares regression analyses, reported inTable 6 (for attitudes toward the MOA)and Table 7 (for team duties done). Fol-lowing the method suggested by Baronand Kenny (1986), we introduce the indi-vidual-level variables in blocks, then con-sider (a) the plant-level competitivethreat variable, (b) the plant dummy vari-ables, and finally (c) the complete modelwith all variables included. (As above, wedo not report equivalent tests for the“forcing” variable.)

Table 3. Attitudes toward the MOA by Plant-Level Factors.(Figures in Cells Are Means for Composite Variable)

Plant Faced Threat of Sell-Off

No Yes Total

Plant Faced “Forcing” No 3.68 3.34 3.52Implementation Strategy N = 244 N = 231 N = 475

Yes 3.82 No observations 3.82N = 307 N = 307

Total 3.76 3.34 3.64N = 551 N = 231 N = 782


Table 4. Team Duties Done by Plant-Level Factors.(Figures in Cells Are Mean Number of Duties Done)

Plant Faced Threat of Sell-Off

No Yes Total

Plant Faced “Forcing” No 13.48 14.84 14.14Implementation Strategy N = 244 N = 231 N = 475

Yes 13.23 No observations 13.23N = 307 N = 307

Total 13.34 14.84 13.78N = 551 N = 231 N = 782

Attitudes

Columns (1) and (2) of Table 6 reprise,in regression form, the central results ofthe one-way and nested analyses of variancein Table 5 for employee attitudes. Column(1) examines the effect of the threat of sell-off, alone. Column (2) contains a modelwith estimated effects for the threat of sell-off and for each individual plant. We omitdummy variables for one Acustar plant(Huntsville) and one non-Acustar plant(Trenton) in order to identify the model.Since Huntsville and New Castle are cov-ered under the “threat of sell-off” variable,the coefficient for New Castle indicates itsdifference from Huntsville. Coefficientsfor the non-Acustar plants are estimates ofeach plant’s difference from Trenton. Asthese columns show, the effect of the threatof sell-off on MOA attitudes is statisticallysignificant, even when the individual plant-level dummy variables are added; the F-testfor the additional variance explained bythe plant dummies is also statistically sig-nificant.

The strongly positive coefficient for NewCastle in column (2) demonstrates, as sug-gested by the means in Table 2, that theworkers at New Castle had significantly morepositive attitudes to the MOA than those atHuntsville, despite their sharing a commonthreat of sell-off. In contrast, coefficientsfor the plants that did not face the threat ofsell-off (Jefferson, Indianapolis, and New-ark) do not differ statistically from that forthe omitted non-Acustar plant (Trenton).Put more generally, reactions to the MOA

at New Castle and Huntsville, both underthreat of sell-off, do differ significantly,suggesting that we need to probe for addi-tional plant-specific factors to understandthis difference.7

Columns (3), (4), and (5) of Table 6examine the effects of demographic factorson attitudes to the MOA. Columns (4) and(5) indicate that, consistent with Hypoth-esis 3c, men responded more favorably thanwomen to the MOA, but there is no supportfor H3a or H3b: age and education did nothave statistically significant effects onworker attitudes. Column (5) shows, how-ever, that the differences between men andwomen are reduced to statistical insignifi-cance by the introduction of plant dummyvariables. This suggests that the statisticallysignificant coefficient for gender in col-umn (3) is actually identifying a plant ef-fect; specifically, the fact that Huntsville,the plant most negative toward the MOA,also had the highest proportion of women.Generally, the plant-level effects are notmediated by the differing demographiccharacteristics of the plants.

7The effects for the Indianapolis and Newark plantsdiffer statistically (p < 0.05) from one another in themodels that do not include the individual percep-tions as predictors (columns 2 and 5 of Table 6),though neither effect is statistically different fromthe effect for Trenton (the omitted plant). Workersat Trenton had more positive attitudes than workersat any other non-Acustar plant in the fully specifiedmodel (column 8); we consistently omitted Trentonto make it easier to compare the different specifica-tions.


Table 5. Analyses of Variance by Plant-Level Factors.

Dependent Variables Independent Variables Degrees of Freedom F (Prob > F)

a. One-Way Analysis of VarianceEmployee Attitude Forcing strategy 1, 780 11.60**

(0.001)Threat of sell-off 1, 780 19.14**

(0.000)Plant dummies 5, 776 8.47**

(0.000)

Team Duties Done Forcing strategy 1, 780 5.52*(0.019)

Threat of sell-off 1, 780 13.08**(0.000)

Plant dummies 5, 776 3.58**(0.003)

b. Two-Way Analysis of Variance by Context (Partial Squares Method)Employee Attitude Forcing strategy 1, 779 1.80

(0.18)Threat of sell-off 1, 779 9.24**

(0.002)

Team Duties Done Forcing strategy 1, 779 0.31(0.58)

Threat of sell-off 1, 779 7.82**(0.005)

c. Nested Analysis of Variance: Plants within Threat of Sell-OffEmployee Attitude Threat of sell-off 1, 776 8.12 **

(0.005)Threat of sell-off 4, 776 5.69**(plus plant dummies) (0.000)

Team Duties Done Threat of sell-off 1, 776 16.26**(0.000)

Threat of sell-off 4, 776 1.20(plus plant dummies) (0.311)

**Statistically significant at the .05 level; ***at the .01 level.

Columns (6), (7), and (8) display theresults of adding the individual-level per-ceptions of the MOA to the model. Thesevariables add considerable explanatorypower. Workers whose job class (H4a) andjob characteristics (H5) improved, whobelieved their teams had real influence(H6a and H6b), and who attributed im-proved performance to the effect of theMOA (H7) were significantly more likely toexpress opinions favorable toward the MOA;the adjusted r-squared for the full model is42.9%. These are the results most likely tobe affected by response-response biases,since workers already positive toward theMOA might be more likely to perceive fa-

vorably its impact on various aspects oftheir job and the overall organization.

Column (8) shows that the individual-level perceptual measures did not mediatethe effect of the business environment onworkers’ overall reactions to the MOA; the“threat of sell-off” coefficient remains sta-tistically significant and negative after wecontrol for all of these variables. Thesemeasures do, however, suggest a mediatingeffect that helps explain the statisticallysignificant difference in reactions to theMOA between the two Acustar plants, NewCastle and Huntsville. The New Castlecoefficient dropped sharply with the inclu-sion of these individual perceptual mea-


Table 6. OLS Regression Coefficients: Determinants of Attitudes toward the MOA.(N = 782; Standard Errors in Parentheses)

Variable (1) (2) (3) (4) (5) (6) (7) (8)

Threat of Sell-Off –0.42*** –0.60*** –0.33*** –0.56** –0.38** –0.61**(0.10) (0.14) (0.10) (0.15) (0.08) (0.12)

Jefferson Assembly 0.05 0.07 –0.32**(0.14) (0.14) (0.12)

Indianapolis Foundry 0.22 –0.25 –0.13(0.14) (0.19) (0.15)

Newark Assembly –0.21 0.19 –0.20(0.18) (0.14) (0.11)

New Castle Machining 0.66*** 0.62** 0.13(faced threat of sell-off) (0.16) (0.18) (0.15)

Male 0.46 *** 0.33*** 0.15 0.35** 0.21** 0.19(0.11) (0.12) (0.13) (0.09) (0.09) (0.10)

Age –0.007 –0.008 –0.010 –0.010* –0.011** –0.009(0.006) (0.006) (0.006) (0.005) (0.005) (0.005)

Education 0.000 –0.002 0.009 0.040 0.038 0.041(0.052) (0.052) (0.052) (0.041) (0.041) (0.041)

Pay Level Increases 0.01 0.06 0.02(0.08) (0.08) (0.08)

Job Class Improves 0.30*** 0.28*** 0.27***(0.08) (0.08) (0.08)

Job Chars. Scale 0.24*** 0.24*** 0.24***(0.04) (0.03) (0.03)

Team Influences Process 0.10*** 0.11*** 0.11***(0.04) (0.03) (0.03)

Team Influences Effort 0.40*** 0.38*** 0.39***(0.04) (0.04) (0.04)

MOA Has Improved 0.54*** 0.55*** 0.57***Performance (0.04) (0.03) (0.04)

Constant 3.76*** 3.70*** 3.58*** 3.83*** 4.00*** 3.53*** 3.76*** 3.91***(0.52) (0.10) (0.39) (0.40) (0.41) (0.32) (0.32) (0.33)

Adjusted r-Squared 0.023 0.046 0.019 0.031 0.048 0.399 0.415 0.419

F for Additional Variables 19.14*** 5.69*** 6.03*** 10.60*** 4.44*** 83.02*** 22.49*** 2.32(Prob > F) (0.00) (0.00) (0.001) (0.001) (0.002) (0.00) (0.000) (0.055)

Omitted Plants: Huntsville (faced threat of sell-off), Trenton Engine (did not face threat).**Statistically significant at the .05 level; ***at the .01 level.

sures. We know from Table 2 that workersat New Castle were especially likely to be-lieve that the MOA had a favorable impacton performance outcomes. Once thesefavorable perceptions are controlled for,the statistically significant coefficient forNew Castle in relation to Huntsville (incolumns 2 and 5) disappears, and no plant-specific effect can be identified for eitherof them.

Estimates for the individual-level factors

also lead to alternative interpretations ofthe differences among the non-Acustarplants. For example, we find that TrentonEngine, the omitted non-Acustar plant, wasmore positive toward the MOA than werethe other plants after we control for theindividual perceptual measures. We notethat the Trenton mean for “MOA has im-proved organizational performance” wassignificantly lower. In other words, work-ers at the Trenton plant did not see a big


impact of the MOA on organizational per-formance, possibly because implementa-tion was initiated only one year before thesurvey. When we control for this differenceacross plants, the results show that Trentonworkers responded positively to the MOA.Column (8) also reveals a statistically sig-nificant, negative plant-specific effect atJefferson Assembly Plant. This result sug-gests that attitudes toward the MOA atJefferson Assembly were generally morenegative than would be predicted by themodel based on individual-level variables.Further analyses (not shown, but availableon request) indicated that the explanationfor the negative response at Jefferson wasmore complicated than the explanationoffered for the positive effect at the Tren-ton plant. The negative coefficient atJefferson became evident only after weadded a combination of several perceptualand demographic variables to the model.

The dependent variable for Table 6 is seton a scale ranging from 1 to 5; the effects ofthe coefficients for dummy variables in themodels can therefore be interpreted as thechanges in attitudes toward the MOA onthis scale. For example, the fully specifiedmodel in column (8) suggests that a workerin an Acustar plant would score about 0.61points lower on the 5 point scale than woulda worker in a non-Acustar plant. The scalescreated from factor scores were standard-ized (mean = 0, s.d. = 1) so that the coeffi-cients for these variables can be interpretedas effects associated with a one standarddeviation change in the variable of interest.

Behaviors

Table 7 reports results of regressionmodels with “team duties done” as the de-pendent variable. As with Table 6, columns(1) and (2) reprise the results of the one-way and nested analyses of variance in or-der to make subsequent comparisons clear,showing that the threat of sell-off variablewas statistically significant and positivelyrelated to team duties done, but that theplant-level effects beyond this measure werestatistically insignificant. The Acustarplants, New Castle Machining and Hunts-

ville, had higher levels of team fulfillmentof the 20 team duties.

Again, the individual-level variablesadded considerably to the fit of the model.Of the demographic variables, age contrib-uted significantly to the fit of the modelspecified in column (5), being positivelyassociated with “team duties done.” (Thestatistical significance of age disappears inthe most fully specified models, having beenmediated by individual perceptions.) Thisresult contradicts H3a, which suggested thatyounger workers would respond more posi-tively to the MOA. Age (like the threat ofsell-off) is related negatively to attitudes tothe MOA but positively to team duties done.

Individual perceptions of the effects ofthe MOA, in contrast to the variables forage and competitive threat, yielded esti-mates that ran in the same direction forboth attitudes and reported behaviors. Ef-fects of improved job characteristics (H5),greater team influence (H6a and H6b),and a perception that the MOA has im-proved performance (H7) were all statisti-cally significant and positive. Improvementin job class had a weaker association with“team duties done” than with “attitude to-ward the MOA.” The threat of sell-off re-mains statistically significant and positiveeven when all other variables are added.The individual perceptions are importantbut do not mediate the effect of this threat.

Effect sizes in Table 7 can be calculatedeasily because the dependent variable is asimple count of duties. The threat of sell-off, for example, was associated with 1.25additional team duties done in the fullyspecified model in column (8). Perceivedteam influences on work effort were espe-cially strongly associated with performingmore team duties, with a one standard de-viation increase in this factor score corre-sponding to nearly 1.6 more reported du-ties done (column 8).

The variance explained in “team dutiesdone” is considerably lower than in thecorresponding analyses for attitudes towardthe MOA: an adjusted R-squared of 16.1%versus 41.9%. This is not surprising. Theattitudinal variables assessed individual re-sponses, while for the “team duties done”


Table 7. OLS Regression Coefficients: Determinants of Reported Number of “Twenty TeamDuties” Performed.

(N = 782; Standard Errors in Parentheses)

Variable (1) (2) (3) (4) (5) (6) (7) (8)

Threat of Sell-Off 1.49*** 1.26 ** 1.53*** 1.13 1.57*** 1.25**(0.41) (0.63) (0.44) (0.67) (0.41) (0.63)

Jefferson Assembly 0.59 0.45 –0.44(0.60) (0.61) (0.60)

Indianapolis Foundry 0.24 –0.062 0.23(0.61) (0.81) (0.77)

Newark Assembly –0.19 0.36 –0.38(0.81) (0.61) (0.58)

New Castle Machining 1.30 1.40 0.20(0.72) (0.81) (0.76)

Male –0.68 –0.08 –0.53 –0.28 –0.85 –0.33(0.49) (0.51) (0.58) (0.48) (0.46) (0.53)

Age 0.057** 0.061** 0.053** 0.045 0.043 0.047(0.025) (0.025) (0.026) (0.023) (0.024) (0.024)

Education –0.019 –0.011 0.011 0.145 0.135 0.149(0.227) (0.226) (0.226) (0.210) (0.212) (0.211)

Pay Level Increases –0.14 0.10 –0.16(0.41) (0.41) (0.43)

Job Class Improves 0.23 0.13 0.23(0.39) (0.39) (0.39)

Job Chars. Scale 0.62*** 0.63*** 0.63***(0.18) (0.18) (0.18)

Team Influences Process 0.46*** 0.51*** 0.45**(0.18) (0.18) (0.18)

Team Influences Effort 1.56*** 1.50*** 1.57***(0.18) (0.18) (0.18)

MOA Has Improved 1.07*** 1.12*** 1.12***Performance (0.18) (0.18) (0.19)

Constant 13.34*** 13.10 *** 11.81*** 10.66*** 11.11*** 10.89*** 11.82*** 11.07***(0.22) (0.46) (1.70) (1.72) (1.79) (1.65) (1.65) (1.71)

Adjusted r-Squared 0.02 0.023 0.005 0.020 0.019 0.16 0.149 0.161

F for Additional Variables 13.08*** 1.20 2.42 12.15*** 0.94 14.49*** 22.86*** 0.34(Prob. > F) (0.000) (0.31) (0.065) (0.001) (0.443) (0.000) (0.000) (0.850)

Omitted plants: Huntsville (faced threat of sell-off), Trenton Engine (did not face threat).**Statistically significant at the .05 level; ***at the .01 level.

measure, workers reported the behavior oftheir team as a group. Given some hetero-geneity of characteristics and attitudesamong team members, there should behigher error associated with models of teambehavior based on individual responses. Itis also worth noting that the range of behav-ior captured by this dependent variable isrelatively narrow. The difference betweenthe plant doing the most team duties, on

average, and the plant doing the least is justover two (10%) on the twenty point scale.This small range notwithstanding, eachblock of variables, with the exception of theplant dummy variables, contributes signifi-cantly to the fit of the fully specified model.8

8The analyses reported above assume that themodel underlying individuals’ reactions to the MOAis the same in each plant, with only the level (as


Discussion and Conclusions

We have found that the majority of work-ers involved with the Modern OperatingAgreement (MOA), negotiated betweenChrysler and the UAW at six manufactur-ing plants, had a positive view of the changesassociated with that work reorganizationinitiative. Roughly two-thirds to three-quar-ters of the workers interviewed said theywere satisfied with the MOA and preferredit to the “old approach”; by an even widermargin, they preferred teams to past ap-proaches. This was true despite the rockystart the MOA got at the plants where it wasforced in, by the company and the nationalunion, over local opposition. At Chrysler,American autoworkers, typically known fortheir strong attachment to traditional ap-proaches to labor-management relations,demonstrated strong interest in pursuingmutual gains, and most saw the MOA as aviable means to that end.

It is possible that the favorable attitudesexpressed here were due to a “halo effect”resulting from the remarkable turnaroundin Chrysler’s performance over the periodpreceding the survey, from near bankruptcyto the most profitable of the Big Three.

Concerns about this effect are mitigated,however, by the wide variation in responseswithin and across plants. Further, a simplehalo effect is belied by the fact that ourmodels consistently identified characteris-tics of individuals who were favorable to thenew approach, whether we measured thatoutcome in terms of their overall affectiveattitude toward the MOA or in terms oftheir behavior in completing the “twentyteam duties.”

The evidence here, together with field-work observations from our broader studyof the MOAs (Lovell et al. 1991), suggestssome general lessons and conclusions. First,workers may react favorably to a top-downtransformation of industrial relations fea-turing teams and an emphasis on labor-management cooperation. The Chrysler-UAW experience with MOAs supports theconclusion that union involvement in workreforms can be a stabilizing and sustainingforce. The joint labor-management gover-nance of the MOA and the institutionaliz-ing of the MOA’s workplace reformsthrough collective bargaining helped keepthe change effort going after a difficultstart and through Chrysler’s financial diffi-culties of the late 1980s.

Second, there were systematic plant-leveldifferences in worker reactions to the MOA.Workers’ responses differed significantlyas a function of the competitive pressuretheir plants faced, specifically in terms ofthe risk of being sold. At the Acustar plants,workers reported that their teams engagedin more key MOA behaviors, but also ex-pressed more negative attitudes to the ini-tiative. This is consistent with our hypoth-esis that affective and behavioral responsesto work reform can differ when the busi-ness environment presents a strong com-petitive threat to a plant’s survival. Thiseffect was stronger than plant-specific ef-fects; once we controlled for the competi-tive threat, there were no further differ-ences between the plants in behaviors (de-spite some remaining differences in atti-tudes).

On the other hand, the initial process ofimplementing the MOA work reforms didnot seem to have any durable or identifi-

indicated by the business environment and plant-intercept terms) varying. We examined that assump-tion with further analyses (not shown) in which weestimated six separate regression models for eachdependent variable. Each of the six models includeda full set of interaction terms between the variablesand a dummy for one of the plants. Testing thecontribution of the interaction terms to model fit isequivalent to testing for the differences between themodel for the plant in question and the model for thesample as a whole (Hardy 1993). With respect toattitudes, the interaction terms as a group were statis-tically significant only for Trenton and for Indianapo-lis, and in each case they made only a small contribu-tion to variance explained—less than 2%. Generally,we concluded, a single underlying model predictsworker attitudes toward the MOA well across theentire sample. Similar analyses for team duties pro-vided even stronger justification for the finding thatessentially the same model underlies variance acrossall six plants. For none of the six plants could wereject (at conventional significance levels) the hy-pothesis that the interaction terms contributed noth-ing to the explanation of variance in team dutiesdone.


able effects on workers’ responses. Reac-tions to the MOA by workers in plants inwhich the MOA was forced in after an ini-tial negative vote do not seem to have dif-fered over the long term from reactions ofworkers in the other plants, once we tookinto account the specific threat of sell-offfaced by workers at the Acustar plants. Thissuggests that the actual experience of work-ing under the MOA work reforms was moreimportant in shaping worker responses thanwas the forcing strategy used during imple-mentation to overcome the barrier of work-ers’ negative preconceptions.

Third, individual characteristics, experi-ences, and perceptions affected workers’attitudes toward these changes in the in-dustrial relations system. Some of theserelationships (the effects of job character-istics, for example) were predictable, andwell supported by theory and previous re-search. Some specifications of the modelshowed that demographic differences wereimportant predictors of worker responses,but these effects were either insignificantor only marginally significant in fully speci-fied models that took into account the plantdifferences and workers’ perceptions of theeffects of the MOA. We believe that moreresearch is needed on different demo-graphic groups’ responses to new work ar-rangements.

Both the individual-level and plant-levelresults suggest that workers will remainsupportive of new work arrangements onlyto the degree that the arrangements actu-ally deliver on their promises. The resultsthus highlight the risks involved with thesearrangements for both unions and man-agement. Workers’ opinions regarding theMOA reflected their assessments of its ef-fects on their individual interests, on teamactivities, and on plant performance. Work-ers who believed that their teams had influ-ence over key aspects of work processes andwork effort, as well as workers who believedthat economic performance had improveddue to the MOA, were especially favorable

to the program. This confirms the impor-tance of the broader context surroundingsuch initiatives in shaping worker attitudes.

Our analysis demonstrates the value oftaking a “meso” level perspective on topicstypically treated at the “micro” level of analy-sis. The combination of quantitative datafrom individuals with perceptions of group-level and organization-level outcomes,qualitative data from fieldwork, and dataanalyses based on identifying underlyingcharacteristics of different plant groupingsprovides important clues to what aspects ofthe context are relevant for employee atti-tudes.

Chrysler workers’ reactions to the MOAsuggest that the answer to the question“What makes teams take?” is “worker self-interest.” Our analysis, however, suggeststhat “worker self-interest” embraces morethan economic considerations. Workersmay not initially embrace work reformsthat appear to cost them the seniority-re-lated economic gains they achieved throughtraditional collective bargaining. They will,however, be positively inclined toward or-ganizational change that is legitimized andmade secure through the backing of a na-tional union contract; that enhances theirindividual economic interests, even if onlymodestly; that improves their daily job ex-periences by making work more intrinsi-cally motivating; that provides an outletfor collective as well as individual voiceon important matters affecting their dailyjob experiences; and that secures theirfuture by enhancing the competitivenessof the enterprise that employs them, andachieves that improved performancethrough mutual gain, rather than zero-sum, strategies. Amid the current ques-tioning of whether workers are benefit-ing from their embrace of work reorgani-zation efforts over the past decade, thisbroader set of criteria—and knowledgeabout the context in which workers formtheir beliefs and choose their actions—willbest guide such assessments.


AppendixTwenty Team Duties of the Modern Operating Agreement

1. Participate in daily audits 11.Keep overtime equalization records.2. Process salvage. 12.Support and help train team members.3. Assist in development of work assignments. 13.Maintain a clean work area.4. Correct and/or report minor and major tooling 14.Problem solving (quality, productivity,

and maintenance problems. statistical process control, and so on).5. Provide input into production standards. 15. Schedule vacations.6. Assist in methods planning. 16. Coordinate with other teams.7. Monitor and control performance. 17. Be aware of and adhere to EEO Guidelines.8. Adhere to plant safety rules. 18. Assist in employee counseling.9. Monitor and report attendance. 19. Administer Capability Progression Plan.10.Equalize overtime. 20. Seek technical assistance when required.

REFERENCES

Altonji, Joseph G., and James R. Spletzer. 1991.“Worker Characteristics, Job Characteristics, andthe Receipt of On-the-Job Training.” Industrial andLabor Relations Review, Vol. 45, No. 1 (October), pp.58–79.

Appelbaum, Eileen, Thomas Bailey, Peter Berg, andArne L. Kalleberg. 2000. Manufacturing Advantage:Why High-Performance Work Systems Pay Off. Ithaca,N.Y.: ILR Press (an imprint of Cornell UniversityPress).

Arthur, Jeffrey B. 1992. “The Link between BusinessStrategy and Industrial Relations Systems in Ameri-can Steel Minimills.” Industrial and Labor RelationsReview, Vol. 45, No. 3 (April), pp. 488–506.

Babson, Steve. 1995. Lean Work: Empowerment andExploitation in the Global Auto Industry. Detroit: WayneState University Press.

Baron, Reuben, and David A. Kenny. 1986. “TheModerator-Mediator Distinction in Social Psycho-logical Research: Conceptual, Strategic, and Statis-tical Considerations.” Journal of Personality and So-cial Psychology, Vol. 51, No. 6 (December), pp. 1173–82.

Becker, Brian, and Barry Gerhart. 1996. “The Impactof Human Resource Management on OrganizationalPerformance: Progress and Prospects.” Academy ofManagement Journal, Vol. 39, No. 4 (August), pp.779–801.

Birecree, Adrienne M. 1993. “Corporate Develop-ment, Structural Change, and Strategic Choice:Bargaining at International Paper Company in the1980s.” Industrial Relations, Vol. 32, No. 3 (Fall), pp.343–66.

Blumberg, Melvin. 1980. “Job Switching in Autono-mous Work Groups: An Exploratory Study in aPennsylvania Coal Mine.” Academy of ManagementJournal, Vol. 23, No. 2 (June), pp. 287–306.

Cappelli, Peter. 1999. The New Deal at Work. Boston:Harvard Business School Press.

Cappelli, Peter, and Peter Sherer. 1991. “The Miss-ing Role of Context in OB: The Need for a Meso-Level Approach.” In L. L. Cummings and Barry M.

Staw, eds., Research in Organizational Behavior, Vol.13. Greenwich, Conn.: JAI Press, pp. 55–110.

Cappelli, Peter, and W. P. Sterling. 1988. “UnionBargaining Decisions and Contract Ratifications:The 1982 and 1984 Auto Agreements.” Industrialand Labor Relations Review, Vol. 41, No. 2 (January),pp. 195–209.

Compa, Lance, and Barbara Riesman. 1985. “TheCase for Adversarial Unions.” Harvard Business Re-view, Vol. 63, No. 3, pp. 23–36.

Cooke, William N. 1992. “Product Quality Improve-ment through Employee Participation: The Effectsof Unionization and Joint Union-Management Ad-ministration.” Industrial and Labor Relations Review,Vol. 46, No. 1 (October), pp. 119–34.

Cotton, John L. 1993. Employee Involvement: Methodsfor Improving Performance and Work Attitudes. NewburyPark, Calif.: Sage.

Drago, Robert. 1996. “Workplace Transformationand the Disposable Workplace: Employee Involve-ment in Australia.” Industrial Relations, Vol. 35, No.4 (October), pp. 526–43.

Eaton, Adrienne E. 1994. “Factors Contributing tothe Survival of Employee Participation Programs inUnionized Settings.” Industrial and Labor RelationsReview, Vol. 47, No. 3 (April), pp. 371–89.

Freeman, Richard B., and James L. Medoff. 1984.What Do Unions Do? New York: Basic Books.

Freeman, Richard B., and Joel Rogers. 1999. WhatWorkers Want. Ithaca, N.Y.: ILR Press (an imprint ofCornell University Press).

Godard, John, and John T. Delaney. 2000. “Reflec-tions on the ‘High-Performance’ Paradigm’s Impli-cations for Industrial Relations as a Field.” Indus-trial and Labor Relations Review, Vol. 53, No. 3 (April),pp. 482–502.

Goodman, Paul S., and James W. Dean. 1982. “Cre-ating Long-Term Organizational Change.” In PaulS. Goodman & Associates, eds., Change in Organiza-tions. San Francisco: Jossey-Bass, pp. 226–79.

Hackman, J. Richard, and Greg R. Oldham. 1980.Work Redesign. Reading, Mass.: Addison-Wesley.


Hardy, Melissa A. 1993. Regression with Dummy Vari-ables. Newbury Park, Calif.: Sage.

House, Robert J., Denise Rousseau, and Melissa Tho-mas-Hunt. 1995. “The Meso Paradigm: A Frame-work for the Integration of Micro and Macro Orga-nizational Behavior.” In L. L. Cummings and BarryM. Staw, eds., Research in Organizational Behavior,Vol. 17. Greenwich, Conn.: JAI Press, pp. 71-114.

Huselid, Mark A. 1995. “The Impact of HumanResource Management Practices on Turnover, Pro-ductivity, and Corporate Financial Performance.”Academy of Management Journal, Vol. 38, No. 3 (June),pp. 635–72.

Ichniowski, Casey, Thomas A. Kochan, David Levine,Craig Olson, and George Strauss. 1996. “WhatWorks at Work: Overview and Assessment.” Indus-trial Relations, Vol. 35, No. 3 (July), pp. 299–333.

Ichniowski, Casey, Kathryn Shaw, and GiovannaPrennushi. 1997. “The Effects of Human ResourceManagement Practices on Productivity: A Study ofSteel Finishing Lines.” American Economic Review,Vol. 87, No. 3 (June), pp. 291–313.

Jacoby, Sanford M. 1983. “Union-Management Co-operation in the United States: Lessons from the1920s.” Industrial and Labor Relations Review, Vol. 37,No. 1 (October), pp. 18–33.

Katz, Harry, and John Paul MacDuffie. 1994. “Collec-tive Bargaining in the U.S. Auto Assembly Sector.”In Paula Voos, ed., Contemporary Collective Bargainingin the Private Sector. Madison, Wis.: Industrial Rela-tions Research Association, pp. 181–224.

Kaufman, Bruce E., and Morris M. Kleiner, eds. 1993.Employee Representation: Alternatives and Future Direc-tions. Madison, Wis.: Industrial Relations ResearchAssociation.

Ketchum, Lyman D., and Eric L. Trist. 1992. AllTeams Are Not Created Equal: How Employee Empower-ment Really Works. Newbury Park, Calif.: Sage.

Kim, Dong-One. 1999. “Determinants of the Survivalof Gainsharing Programs.” Industrial and LaborRelations Review, Vol. 53, No. 1 (October), pp. 21–42.

Kochan, Thomas A., Harry M. Katz, and Robert B.McKersie. 1994. The Transformation of AmericanIndustrial Relations, 2nd ed. Ithaca: ILR Press.

Kochan, Thomas A., and Paul Osterman. 1994. TheMutual Gains Enterprise. Boston: Harvard BusinessSchool Press.

Konzelmann Smith, Suzanne. 1996. “CooperativeCorporate-Level Strategies and Divergent LaborRelations Outcomes: An Institutional Analysis.”Journal of Economic Issues, Vol. 30, No. 3 (Septem-ber), pp. 797–827.

Lovell, Malcolm, Susan Goldberg, Larry W. Hunter,Thomas A. Kochan, John Paul MacDuffie, AndrewMartin, and Robert B. McKersie. 1991. “Report onthe Chrysler-UAW Modern Operating Agreement(MOA) Experiment.” Washington, D.C.: Depart-ment of Labor.

MacDuffie, John Paul. 1995. “Human ResourceBundles and Manufacturing Performance : Organi-zational Logic and Flexible Production Systems inthe World Auto Industry.” Industrial and LaborRelations Review, Vol. 48, No. 2 (January), pp. 199–221.

Osterman, Paul. 2000. “Work Reorganization in anEra of Restructuring: Trends in Diffusion and Ef-fects on Employee Welfare.” Industrial and LaborRelations Review, Vol. 53, No. 2 (January), pp. 179–98.

Parker, Michael, and Jane Slaughter. 1988. ChoosingSides: Unions and the Team Concept. Boston: SouthEnd.

Pearce, John A. II, and Elizabeth C. Ravlin. 1987.“The Design and Activation of Self-Regulating WorkGroups.” Human Relations, Vol. 40, No. 11, pp. 751–82.

Staw, Barry M., and Jerry Ross. 1978. “Commitmentto a Policy Decision.” Administrative Science Quar-terly, Vol. 23, No. 1 (March), pp. 40–64.

Thompson, Donna E., and Laura E. Gooler. 1996.“Capitalizing on the Benefits of Diversity throughWorkteams.” In Ellen Ernst Kossek and SharonLobel, eds., Managing Diversity: Human ResourceStrategies for Transforming the Workplace. Cambridge,Mass.: Blackwell, pp. 392–437.

Walton, Richard E., Joel Cutcher-Gershenfeld, andRobert B. McKersie. 1994. Strategic Negotiations: ATheory of Change in Labor-Management Relations. Bos-ton: Harvard Business School Press.

Documents

WHAT MAKES TEAMS TAKE? EMPLOYEE REACTIONS TO WORK …