Total QualityManagement: Empirical,Conceptual, andPractical Issues

J. Richard HackmanHarvard UniversityRuth WagemanColumbia University

1995 by Cornell University.0001-8392/95/4002-0309/$1.00.

We gratefully acknowledge theassistance of Cathy Sirett in conductinglibrary research for this commentary.

We do not provide here a comprehensivereview of the large and rapidly growingliterature on TOM. For a sampling, seeJablonski (1992), Krishnan et al., (1993),Sashkin and Kiser (1993), and the July1994 special issue of the Academy ofManagement Review (vol, 19. no. 3) on"total quality."

In recent years, total quality management (TOM) hasbecome something of a social movement in the UnitedStates. This commentary returns to the writings of themovement's founders-W. Edwards Deming, JosephJuran, and Kaoru Ishikawa-to assess the coherence,distinctiveness, and likely perseverance of thisprovocative management philosophy. We identify anumber of gaps in what is known about TOM processesand outcomes and explore the congruence between TOMpractices and behavioral science knowledge aboutmotivation, learning, and change in social systems. Thecommentary concludes with a prognosis about the futureof TOM-including some speculations about what will beneeded if TOM is to take root and prosper in the years tocome."

It has now been a decade since the core ideas of totalquality management (TOM) set forth by W. EdwardsDeming, Joseph Juran. and Kaoru Ishikawa gained significantacceptance in the U.S. management community. In thatdecade, TOM has become something of a social movement.It has spread from its industrial origins to health careorganizations. public bureaucracies. nonprofit organizations.and educational institutions. It has become increasinglyprominent in the popular press. in the portfolios of trainersand consultants. and, more recently. in the scholarlyliterature.' Institutions specifically chartered to promote TOMhave been established. and a discernible TOM ideology hasdeveloped and diffused throughout the managerialcommunity. And. in its maturity, TOM has becomecontroversial-something whose worth and impact peopleargue about.Some writers have asserted that TOM provides a historicallyunique approach to improving organizational effectiveness,one that has a solid conceptual foundation and. at the sametime. offers a strategy for improving performance that takesaccount of how people and organizations actually operate(Wruck and Jensen. 1994). A more skeptical view is thatTOM is but one in a long line of programs-in the traditionof T-groups. job enrichment. management by objectives. anda host of others-that have burst upon the managerial scenerich with promise, only to give way in a few years to yetanother new management fashion.In this commentary, we provide a conceptual analysis ofTOM that places these competing claims in perspective. Weask whether there really is such a thing as TOM or whetherit has become mainly a banner under which a potpourri ofessentially unrelated organizational changes are undertaken.We document how TOM activities and outcomes have beenmeasured and evaluated by researchers and note somesignificant gaps in what has been learned. We explore theuneasy relation between behavioral processes that arecentral to TOM practice and mainline organizationalscholarship about those same processes. And we concludewith an overall assessment of the current state of TOMtheory and practice. including some speculations about whatmay be required if this potentially powerful approach is totake root and prosper in the years to come.

309/Administrative Science Quarterly, 40 (1995): 309-342

IS THERE SUCH A THING AS TOM?As is inevitable for any idea that enjoys wide popularity inmanagerial and scholarly circles, total quality managementhas come to mean different things to different people. Thereis now such a diversity of things done under the name "totalquality" that it has become unclear whether TOM still has anidentifiable conceptual core, if it ever did. We begin with aclose examination of what the movement's founders had tosay about what TOM was supposed to be, and then weassess how TOM as currently practiced stacks up againstthe founders' values and prescriptions.Virtually everything that has been written about TOMexplicitly draws on the works of W. Edwards Deming,Joseph Juran, and Kaoru Ishikawa, the primary authorities ofthe TOM movement (for a review, see Crosby, 1989). Ratherthan providing here a precis of their writings, we draw onthem to determine whether there exists among them (1) acoherent philosophical position that specifies the core valuesto be sought in TOM programs and (2) a distinctive set ofinterventions (structures, systems, and/or work practices)that are intended specifically to promote those values.TOM PhilosophyDeming, Ishikawa, and Juran share the view that anorganization's primary purpose is to stay in business, so thatit can promote the stability of the community, generateproducts and services that are useful to customers, andprovide a setting for the satisfaction and growth oforganization members (Juran, 1969: 1-5); Ishikawa, 1985: 1;Deming, 1986: preface). The focus is on the preservationand health of the organization, but there also are explicitlystated values about the organization's context (thecommunity and customers) and about the well-being ofindividual organization members: As Ishikawa (1985: 27)said, "An organization whose members are not happy andcannot be happy does not deserve to exist." The TOMstrategy for achieving its normative outcomes is rooted infour interlocked assumptions-about quality, people,organizations, and the role of senior management.Assumptions. The first assumption is about quality, which isassumed to be less costly to an organization than is poorworkmanship. A fundamental premise of TOM is that thecosts of poor quality (such as inspection, rework, lostcustomers, and so on) are far greater than the costs ofdeveloping processes that produce high-quality products andservices. Although the organizational purposes espoused bythe TOM authorities do not explicitly address traditionaleconomic and accounting criteria of organizationaleffectiveness, their view is that organizations that producequality goods will eventually do better even on traditionalmeasures such as profitability than will organizations thatattempt to keep costs low by compromising quality (Juran,1974: 5.1-5.15; Ishikawa, 1985: 104-105; Deming, 1986:11-12). The strong version of this assumption, implicit inJuran and Ishikawa but explicit and prominent in Deming'swriting, is that producing quality products and services is notmerely less costly but. in fact. is absolutely essential tolong-term organizational survival (Deming, 1993: xi-xii).310/ASQ, June 1995

Total Quality Management

The second assumption is about people. Employees naturallycare about the quality of work they do and will takeinitiatives to improve it-so long as they are provided withthe tools and training that are needed for qualityimprovement. and management pays attention to their ideas.As stated by Juran (1974: 4.54). "The human being exhibitsan instinctive drive for precision, beauty and perfection.When unrestrained by economics, this drive has created theart treasures of the ages." Deming and Ishikawa add that anorganization must remove all organizational systems thatcreate fear-such as punishment for poor performance,appraisal systems that involve the comparative evaluation ofemployees. and merit pay (Ishikawa. 1985: 26; Deming.1986: 101-109).The third assumption is that organizations are systems ofhighly interdependent parts, and the central problems theyface invariably cross traditional functional lines. To producehiqn-quatitv products efficiently, for example, productdesigners must address manufacturing challenges andtrade-offs as part of the design process. Deming and Juranare insistent that cross-functional problems must beaddressed collectively by representatives of all relevantfunctions (Juran, 1969: 80-85; Deming. 1993: 50-93).Ishikawa, by contrast, is much less system-oriented: Hestates that cross-functional teams should not set overalldirections; rather, each line division should set its own goalsusing local objective-setting procedures (Ishikawa, 1985:116-117).The final assumption concerns senior management. Qualityis viewed as ultimately and inescapably the responsibilityof top management. Because senior managers create theorganizational systems that determine how products andservices are designed and produced, the quality-improvementprocess must begin with management's own commitmentto total quality. Employees' work effectiveness is viewed asa direct function of the quality of the systems that managerscreate (Juran, 1974: 21.1-21.4; Ishikawa, 1985: 122-128;Deming, 1986: 248-249).Change principles. TQM authorities specify four principlesthat should guide any organizational interventions intendedto improve quality. The first is to focus on work processes.The quality of products and services depends most of all onthe processes by which they are designed and produced. Itis not sufficient to provide clear direction about hoped-foroutcomes; in addition, management must train and coachemployees to assess, analyze, and improve work processes(Juran, 1974: 2.11-2.17; Ishikawa, 1985: 60; Deming, 1986:52).The second principle is analysis of variability. Uncontrolledvariance in processes or outcomes is the primary cause ofquality problems and must be analyzed and controlled bythose who perform an organization's front-line work. Onlywhen the root causes of variability have been identified areemployees in a position to take appropriate steps to improvework processes. According to Deming (1986: 20). "Thecentral problem of management ... is to understand betterthe meaning of variation, and to extract the information

311/ASQ. June 1995

contained in variation" (see also Juran, 1974: 2.10-2.17;Ishikawa. 1985: chap. 12).The third principle is management by fact. TOM calls for theuse of systematically collected data at every point in aproblem-solving cycle-from determining high-priorityproblems, through analyzing their causes, to selecting andtesting solutions (Juran, 1974: 22.1-28.1; Ishikawa, 1985:104-105; Deming, 1986: chap. 8). Although Deming,Ishikawa, and Juran differ in their preferred analytical tools,each bases his quality-improvement program on collectingdata, using statistics. and testing solutions by experiment.

The fourth principle is learning and continuous improvement.The long-term health of an enterprise depends on treatingquality improvement as a never-ending quest. Opportunitiesto develop better methods for carrying out work alwaysexist, and a commitment to continuous improvementensures that people will never stop learning about the workthey do (Juran, 1969: 2-3; Ishikawa, 1985: 55--56; Deming,1986: 49-52).TOM Interventions

Despite some differences in emphasis, the three TOMauthorities have a common philosophical orientation andshare a set of core values about people, organizations, andchange processes. They prescribe five interventions torealize those values.

Explicit identification and measurement of customerrequirements. To achieve quality, it is essential to knowwhat customers want and to provide products or servicesthat meet their requirements (Ishikawa, 1985: 43). It isnecessary, therefore, for organization members to assessdirectly customer requirements such as durability, reliability,and speed of service (Juran, 1974: 2.2; Deming, 1986:177-182). Some customers are external to the organization,others are internal, as when the output of some organizationmembers is passed on to others. TOM defines the nextprocess down the line as the "customer" for each process.Within the organization, then, the assessment of customerrequirements serves as a tool to foster cross-functionalcooperation (Ishikawa, 1985: 107-108).With data about customer requirements in hand, qualityimprovement can focus specifically on those aspects ofwork processes that are most consequential for customersatisfaction. Even so, high quality is not assured. Someorganizations actively manipulate customer preferences (forexample, through advertising) to bring them into line withwhat the organization already is able to provide. Andcustomers may define their own requirements in terms ofexisting products and services that may be low in quality(Hayes and Abernathy, 1980). Deming (1993: 7-9) suggeststhat this may be especially characteristic of customers in theUnited States, because they have grown accustomed topoor-quality products and services; U.S. organizations thatrely too heavily on what customers say they want risksetting quality standards far below what employees actuallyare capable of achieving.312/ASQ, June 1995

Total Quality Management

Creation of supplier partnerships. TOM authorities suggestthat organizations should choose vendors on the basis ofquality, rather than solely on price. Moreover, theyrecommend that organizations work directly with rawmaterial suppliers to ensure that their materials are of thehighest quality possible (Juran, 1974: 10.1-10.35; Ishikawa,1985, chap. 9; Deming, 1986: 31-43).Use of cross-functional teams to identify and solvequality problems. Although cross-functional teams can beused in multiple ways in TOM programs, their main purposeis to identify and analyze the "vital few" problems of theorganization (Ishikawa, 1985: 113-119; Deming, 1993:85-89). Juran (1969) refers to such teams as the "steeringarm" of a quality effort. Other teams, also cross-functional.are created to diagnose the causes of problems that havebeen identified by the steering arm and to develop and testpossible solutions to them. Diagnostic teams can be eithertemporary task forces or continuing organizational entities. Inboth cases, department heads are included as teammembers to ensure that stakeholder departments willcooperate when the time comes to implement the team'srecommendations. Juran, far more than Deming, advocatesthe use of quality-improvement teams within functions. Butthe team composition principle is the same: Choose peoplewho can provide access to the data necessary for testingpotential solutions and who are critical to implementing thesolutions developed (Juran, 1969: 78-89).Use of scientific methods to monitor performance and toidentify points of high leverage for performanceimprovement. The three TOM authorities are of one voicein advocating the use of statistical tools to monitor andanalyze work processes (Juran, 1974: chaps. 22-27;Ishikawa, 1985: 109-120; Deming, 1986: chaps. 8-9). Awide variety of statistical tools are available to identify thepoints of highest leverage for quality improvement, toevaluate alternative solutions to identified problems, and todocument the results of process changes. Many of the toolsinvolve applications of probability theory to generate findingsthat then can be summarized pictorially. Literally dozens of"quality tools" have been described in the literature (for areview, see Sashkin and Kiser, 1993). Three of the mostcommonly used tools are control charts, Pareto analysis, andcost-of-quality analysis.

A control chart provides a pictorial representation of theoutputs of an ongoing process. Control charts are used tomonitor the performance of a process and to determinewhether that process is "in control"-whether the varianceproduced by the process is random or attributable to specificcauses. It is assumed that all processes produce variance,but a stable process fluctuates randomly. Therefore, datafrom a stable process will tend to fall within predictablebounds. Scrutiny of a control chart allows the user to (1)determine whether a given process is in need ofimprovement, (2) identify points outside the control range sothat the causes of uncontrolled variance can be sought, and(3) reassess the process after experimental attempts toimprove it are completed (Deming, 1986: 323-346).313/ASQ. June 1995

Pareto analysis is used to identify the major factors thatcontribute to a problem and to distinguish the "vital few"from the "trivial many" causes. Pareto charts are used wheneach separate contributor to a problem can be quantified.For example, a group attempting to identify the vital fewcauses of high inventory costs would list each inventoryitem in order of total dollar value of materials kept in stock.Those materials that turn out to be major contributors toinventory costs are then addressed first (Juran, 1969:43-54).Cost-at-quality analysis is used to highlight the cost savingsthat can be achieved by doing the work right the first time.The analysis involves quantifying all costs associated withmaintaining acceptable quality levels, such as the costs ofpreventing errors, and then comparing these with the costsincurred by failures to achieve acceptable quality, such asthe cost of rework. Cost-of-quality analysis thus helps toidentify those opportunities for improvement that offer thelargest cost savings (Juran, 1974: chap. 5; Ishikawa, 1985:54-55).Use of process-management heuristics to enhanceteam effectiveness. The TOM authorities suggest severaltechniques to help quality teams use their collectiveknowledge effectively in identifying and analyzingopportunities to improve quality. Three of the mostcommonly used devices are flowcharts, brainstorming, andcause-and-effect diagrams.

A flowchart is a pictorial representation of the steps in awork process. Flowcharts, which use standardized symbolsto represent types of activities in a process, help membersidentify activities that are repetitive, that add no value, orthat excessively delay completion of the work (Deming,1993: 58-61).Brainstorming is used by groups to generate lists of ideasabout matters such as the potential causes of a problem,possible solutions, and issues likely to be encountered inimplementing those solutions. Its purpose is to tap thecreativity of group members by explicitly ruling out theevaluation of member contributions to the list and activelyencouraging building on others' ideas. Brainstorming often isfollowed by the Nominal Group Technique or multivoting toreduce and prioritize the list that has been generated(Ishikawa, 1985: 64-65).A cause-and-ettect diagram or "fishbone" was developed byIshikawa to graphically represent the relationship between aproblem and its potential causes. Fishbone diagrams canhelp a group examine thoroughly all possible causes of aquality problem and discern the relationships among them.Group members place the problem at the right-hand side ofthe page (the head of the fish). The "bones" of the fish arelines on which members list the potential causes bycategory; the generic categories are causes related topeople, tools, materials, and methods. Members then collectdata to assess the potency of each of these potential causes(Ishikawa, 1985: 63-65).314/ASQ. June 1995

2For findings from surveys of U.S.organizations about their TOM practices.see Conference Board (1991), DeltaConsulting Group (1993), KPMG PeatMarwick (1991), and Lawler, Mohrman,and Ledford (1992). Neither thesesurveys nor this commentary addressdifferences between TOM practices inthe U.S. and other countnes.

Total Quality Management

According to the founders of TOM, the five interventionssummarized above define the core of total qualitymanagement. Knowledge of customer requirementsprovides a test for considering and evaluating processchanges. Supplier partnerships ensure that materialsentering the organization are of acceptable quality.Cross-functional teams bring the full spectrum of relevantinformation and expertise to bear on decisions aboutsystemwide problems. Scientific methods and statisticalanalyses provide teams with trustworthy data to use in theirdecision making. And process management heuristics canimprove the quality of the decision-making process itself.TOM in PracticeTo assess how TOM actually is practiced in U.S. workorganizations, we address two questions. First. in what waysare contemporary implementations of TOM consistent withthe founders' tenets? Second, in what ways do currentpractices differ from their prescriptions, and do thosedifferences enrich the core ideas of quality management ordiverge from them?2Continuities. Organizations that implement TOM areconsistent with the founders' ideas in developing means forassessing their customers' preferences, altering relationshipswith suppliers, using teams (both cross-functional andwithin-function) to solve problems, investing in training inproblem-solving tools and, to a lesser degree, teachingstatistical analysis to front-line employees. The top-downimplementation strategy used in U.S. organizations iscongruent with the founders' assumption that quality isultimately a management responsibility and that attempts toimprove quality must begin at the top. The five TOMpractices described below, presented in order of theirprevalence, are generally consistent with the ideas andtechniques originally articulated by Deming, Ishikawa, andJuran.A recent survey reports that the single most commonly usedTOM technique is formation of short-term problem-solvingteams with the overall objective of simplifying andstreamlining work practices (Conference Board, 19911.Nearly all manufacturing firms using TOM use such teams,and 90 percent of service firms do so. Problem-solvingteams work on a wide variety of tasks, ranging fromcross-functional involvement in product design to solvingwithin-unit workflow problems.The second most commonly used practice is training,Organizations that implement TOM invest heavily in formaltraining for a large proportion of their employees. Accordingto the Conference Board (1991). 92 percent ofmanufacturing companies and 75 percent of servicecompanies implementing TOM use some form of training aspart of their change effort. Typically, nearly all senior andmiddle managers are trained in quality practices, with amedian of 16 hours of training. About 80 percent of first-linesupervisors and 50 percent of nonmanagement employeesreceive a median of eight hours of training. Olian and Rynes(1991) found the most common training content to be, inorder of frequency, interpersonal skills, quality-improvement

315/ASQ. June 1995

processes and problem-solving, team leading and building,running meetings, statistical analysis, supplier qualificationtraining, and benchmarking.The third practice is top-down implementation. In keepingwith the TOM authorities' view that quality is ultimately theresponsibility of top management, most TOM programsbegin with the training of top managers in the qualityphilosophy, followed by the articulation of anorganization-wide quality vision and communication of thatvision throughout the organization (Conference Board, 1991:18). Both education about TOM and implementation of TOMpractices typically take place in cascading fashion, with eachlayer carrying the message to the next lower level of theorganization.The fourth practice is developing relationships with suppliers.At least 50 percent of TOM organizations collaborate withtheir suppliers in some way to increase the quality ofcomponent parts (Lawler, Mohrman, and Ledford, 1992Loften by sending "quality action teams" to consult with theirmajor suppliers. The objective is to help suppliers use TOMto analyze and improve their own work processes (Sashkinand Kiser, 1993).The fifth practice is obtaining data about customers.Although systematic data are not available on the proportionof TOM organizations that directly assess customerpreferences and customer satisfaction, nearly all casestudies of TOM companies include descriptions of themeans such organizations use to obtain customer data.Commonly used devices for obtaining these data includetoll-free complaint lines, marketing research firms, andcustomer focus groups (Olian and Rynes, 1991).Enrichments. Two additional interventions-competitivebenchmarking and employee involvement-have becomestrongly associated with TOM in the United States. Althoughnot explicitly advocated by the TOM founders, theseactivities are generally consistent with their ideas.Benchmarking involves gathering information about "bestpractices" from other organizations. Thus, a company thatwishes to improve its customer service might observeservice practices in firms renowned for their service quality,regardless of their industry. The prevalence of benchmarkingin contemporary TOM programs appears to derive primarilyfrom its inclusion as a Baldridge Ouality Award criterion(Malcolm Baldridge National Ouality Award Consortium,1990).Benchmarking serves multiple functions consistent withTOM philosophy: (1) determining what customers canexpect to get from the competition, as part of assessingcustomer requirements, (2) learning alternative workprocesses, and (3) in some cases, guiding the establishmentof quality-improvement goals. The ambitious quality goals ofmany TOM programs, such as zero "defections," cuttingdefects by 90 percent in two years, or reducing cycle timeby 50 percent may be more likely to be accepted byorganization members once competitive benchmarkingdemonstrates that other organizations achieve them (Olianand Rynes, 1991). Just because a new idea has been316/ASQ. June 1995

Total Quality Management

discovered, however, does not mean it will be used. Ulrich,Von Glinow, and Jick (1993) noted that organizations thatrely heavily on external benchmarking to identify superiorwork processes tend to have the most difficulty gettingthose processes adopted internally.TOM organizations in the U.S. invariably introducemechanisms for employee involvement in quality that extendbeyond those that are integral to the TOM program itself(Lawler, 1994). According to the Conference Board (1991),65 percent of TOM organizations create employeesuggestion systems, and 70 percent have quality meetingsbetween managers and employees and/or focus groups tosolicit ideas about quality. The widespread use of "qualitydays" and other celebrations of quality-related events andachievements further reinforce the aspiration, in manyorganizations, to involve every member in quality-improvement processes. Such celebrations, moreover, areconsistent with Deming's view that social approval andpublic recognition are important sources of humanmotivation (Deming, 1986: 85). Finally, some TOMorganizations create self-managing teams to perform theregular work of the enterprise, thereby further expanding theinvolvement of organization members. KPMG Peat Marwick(1991) found that 15 percent of the TOM organizationsstudied used such teams, and among organizations withmore than five years of TOM experience, almost 50 percentdid so.Divergences. Some aspects of contemporary TOM practicedilute or redirect the core ideas of the movement'sfounders. Noteworthy are the attenuated role of scientificmethods in TOM programs and an increasing reliance onperformance measurement and performance-contingentrewards to motivate and control employees.The use of scientific methods is among the most distinctivefeatures of TOM. In contemporary practice, however, thereis much greater emphasis on group-process techniques andinterpersonal skills than on scientific methods. Nor is thereevidence that organization members actually use even thosestatistics they have been taught. In many organizations, theemphasis on statistics and experimentation is stripped awayvery early in the process of implementing TOM, leaving onlythe rhetoric of "management by fact" (Zbaracki, 1994).A large majority of organizations using TOM modify theirperformance measurement and reward systems so thatachievement of specific quality goals can be assessed andrewarded, even though Deming (1993) explicitly argues thatsuch practices are counterproductive. According to the 1991Conference Board survey, 85 percent of TOM organizationshave developed programs to reward individuals and teamsfor quality achievements. In addition, 75 percent tieperformance appraisals to quality, although principally formanagers (only 46 percent of front-line employees areassessed on their use of quality tools). Such practices mayderive from Juran's eclectic philosophy for managingindividual performance. Unlike Deming and Ishikawa, Juransubscribes to no particular motivation theory. Rather, he callsfor the use of a range of motivational techniques, from job317/ASQ. June 1995

3This is a characteristically psychologicalapproach. which is consistent with ourown backgrounds. For an excellentanalysis from the perspective ofeconomics. see Wruck and Jensen(1994); for one from the perspective ofsociology. see Zbaracki (1994); and forone from the perspective of managementtheory. see Anderson. Rungtusanatham.and Schroeder (1994).

enrichment to quality audits: "All available tools must beused. each directed at the specific problem it is able tosolve. None are panaceas" (Juran. 1969: 18-19).The longer an organization has been involved with TOM. thegreater its reliance on incentives to motivate work towardquality improvement goals. The 1991 KPMG Peat Marwicksurvey found that 60 percent of organizations with five ormore years of TOM experience explicitly rewarded theachievement of quality goals. The survey also showed.however. that organizations with greater experience withTOM tended to place greater emphasis on group.departmental, or organization-wide. rather than individualrewards. This is consistent with the TOM authorities'emphasis on teamwork and between-unit interdependenceand with their view that it is the system. not individualefforts. that ultimately determines quality. Such rewards.however, almost always are linked to quantitativeperformance measures. which Deming believed to be"limiting" (Deming, 1993: 47).ConclusionWe have approached the question. "Is there such a thing asTOM?" in the same way that we would approach analysis ofany construct: by assembling data relevant to its convergentand discriminant validity. Convergent validity. as adaptedfrom Campbell and Fiske (1959) for present purposes.reflects the degree to which the versions of TOMpromulgated by its founders and observed in organizationalpractice share a common set of assumptions andprescriptions. Discriminant validity refers to the degree towhich TOM philosophy and practice can be reliablydistinguished from other strategies for organizationalimprovement. such as participative management.management by objectives. and so on. Only If TOM passesthese two validity tests does it make sense to dig moredeeply into the conceptual. empirical, and practical issuesthat can inform its overall assessment."Convergent validity. We conclude that TOM passes theconvergent validity test. As we have seen, there issubstantial agreement among the movement's foundersabout the key assumptions and practices of total qualitymanagement. Moreover. contemporary TOM practice isgenerally consistent with the founders' ideas. The record isnot perfect. however. Some of the sharpest and mostdistinctive ideas of the TOM founders have been sandeddown a bit over the last decade. And there is. these days.greater adherence to TOM philosophy at the espoused thanat the operational level, as seen. for example. by thediminished role of scientific methods and statistical tools inmany TOM programs. Still. we find that there is impressiveconvergence-across theorists, across practitioners. andacross time-of the basic ideas of total quality management.Discriminant validity. In assessing the distinctiveness ofTOM. we consider three comparison groups: programs thatare subsumed by a full-fledged TOM program. those that areclearly different from TOM, and those that are. like TOM.broad and multifaceted organizational improvementprograms. As articulated by its founders. TOM clearly

318/ASQ. June 1995

Total Quality Management

subsumes a number of smaller and more focused initiatives,such as quality circles, cross-functional project teams, andzero-defects programs. It also is distinct from interventionssuch as job enrichment, performance-contingent rewards,and goal setting, some of which are explicitly disavowed byone or more of the three TOM founders-for example, theemphasis of some job-design models on employeeautonomy about methods, pay-for-performance programsthat tie financial rewards to bottom-line outcomes, andgoal-setting programs that specify explicit performanceobjectives within a department or function. We will havemore to say about these discrepancies later; for now, wesimply note that TOM is clearly different, both conceptuallyand operationally, from at least this sample of other changeprograms.Finally, TOM as described by its founders is readilydistinguishable from other broad, multifaceted initiatives,such as participative management and quality of worklifeprograms. As a concept, participative management woulditself fail the convergent and discriminant validity tests. Somany different interventions are mounted under this labelthat they appear to have only the most general kind ofmanagement philosophy in common. Moreover. participativemanagement has no generally accepted set of "must do"management practices; the way participation is implementedin one organization often bears little resemblance to its usein another. Participative management, then, is the same kindof thing as TOM but is not as clean or distinctiveconceptually.The quality of worklife (OWL) movement, which peaked inprominence in the U.S. in the early 1980s. has manysimilarities to TOM (Lawler, 1986: 119-143). Both initiativesare rooted in identifiable philosophical positions and bothhave associated with them a specific set of interventions.The values of the OWL movement center on fosteringlabor-management cooperation to improve simultaneouslyproductivity and the quality of employees' worklives. Keyorganizational devices in OWL include labor-managementcommittees, surveys of employee satisfaction, and so on.OWL, like TOM, is a broad and multifaceted change effortthat also passes the convergent validity test. But it clearlydiffers from TOM both in philosophy and in practice.Our conclusion is that TOM does pass the discriminantvalidity test with reference to the writings of the TOMfounders. But it is close to failing that test when onefocuses on contemporary organizational practice. Manydevices that are specifically eschewed by the founders arenow commonly implemented in the name of TOM. Andmany practitioners now talk about "involvement" and"empowerment" as if they were synonymous with TOMand implement various employee involvement orempowerment interventions as part of a TOM package(Lawler, Mohrman, and Ledford. 1992). Thus. one can in1995 still point with some confidence to the constellation ofideas and interventions that form the core of TOM, and onecan, with less confidence, show how that constellationdiffers from others. At least for now, there is indeed a"there" there for TOM.319/ASQ, June 1995

MEASURING AND ASSESSING TOM ACTIVITIESAND OUTCOMES

A full-fledged evaluation of a TOM program would includethree distinct types of assessment. First is empiricaldemonstration that TOM has, in fact, been implemented, toconfirm that it is TOM that is being assessed rather than, forexample, some subset of the integrated TOM package,some related intervention, or some wholly different programthat has been carried out using TOM rhetoric.

Second is determination of whether TOM alters how peoplework together to meet customer requirements. This analysisassesses process criteria of effectiveness-the degree towhich the improvements in organizational functioning thatare expected actually are observed. Finally comesassessment of outcome criteria-the degree to whichimprovements in bottom-line organizational effectiveness arefound. It is important to examine both process and outcomecriteria because, as scholars who study decision makingknow all too well. a capricious environment sometimes canintervene between process and outcome in a way that turnsbehaviors that could not have been better into results thatcould hardly have been worse. As will be seen, empiricalevaluation of TOM programs presents a significant challengeto researchers because what must be done to accomplishthe three different assessments involves very differentmethods and analytic strategies.

Is It Really TOM?

To address substantive questions about the effects of TOMon organizations and their members, one must first establishthat TOM actually has been installed. This is theorganizational change equivalent of conducting amanipulation check in the research laboratory to ensure thatthe experimental intervention actually was implemented asintended. To accomplish this task in research on TOM, onewould collect behavioral data to document that the five corefeatures of TOM are in place. Specifically, (1) Areorganization members assessing customer requirements andmeasuring performance against those requirementscontinuously? (2) Are suppliers chosen on the basis ofquality, rather than solely on the basis of cost. and areorganization members working with suppliers to improvesuppliers' quality practices? (3) Are members operatinginterdependently, as teams, across traditional organizationalfunctions, rather than independently or in ways that maintainfunctional separateness? (4) Are members using statisticsand scientific reasoning to formulate and test hypothesesabout work processes and strategies for performanceimprovement? and (5) Are members using process-management heuristics to enhance team problem solvingand decision making? Research on TOM rarely addressesthese questions, except for studies that focus on BaldridgeAward-winning companies. Because application for thataward involves careful inspection of actual quality practices,it is safe to assume that award winners actually haveimplemented the full TOM package (see Malcolm BaldridgeNational Ouality Award Consortium, 1990).320/ASQ. June 1995

4This problem is not unique to TOM.Frank and Hackman (1975) empiricallydocumented the same difficulty inevaluations of the impact ofjob-enrichment programs.

Total Quality Management

More common is research that involves no attempt at all toassess the degree to which TOM has been implemented:Of 99 papers about the effects of TOM published inacademic and practitioner journals between 1989 and 1993,only 4 percent assessed the degree to which TOMinterventions actually were in place. And even when"manipulation checks" were performed, they often involvedinferences based on qualitative accounts of the evolution ofa TOM program rather than on direct measurements ofbehavior (e.g., Fisher, 1992). Existing research findings aboutthe effects of TOM on organizational performance, therefore,may be about programs that-although perhaps well-tailoredto a given organization's needs-are not fUll-fledgedimplementations of TOM. 4

What Are Its Effects on Work Processes?The second challenge in research on TOM is to specify andcollect data about those processes that would be expectedto result from TOM and that should, all else being equal,contribute to organizational effectiveness. Considering groupand organizational performance generally, without specificreference to TOM, Hackman (1987) has suggested threeprocess criteria of unit effectiveness: (1) the level oftask-oriented effort exhibited by unit members, (2) theamount of knowledge and skill members apply to their work,and (3) the appropriateness of the task performancestrategies members use in carrying out the work. To theextent that a work unit has a high standing on these processcriteria, the likeiihood increases that its final product. service,or decision also will turn out well. To the extent thatmembers exhibit insufficient effort, bring insufficient talentto bear on the work, or use task-inappropriate performancestrategies, overall unit effectiveness is likely to suffer.Although these three process criteria are quite general andhave been neither explicated nor endorsed by TOM scholarsor practitioners, they might nonetheless be useful inassessing the impact of TOM interventions on workprocesses. We consider below the ways that a qualityteam's standing on each of the three criteria might beenhanced by TOM interventions.Effort. Ouality teams have challenging and significantwork-specifically, collaborating to generate continuousimprovement in meeting explicitly stated customerrequirements-which motivation theory suggests shouldenhance collective effort. Moreover, use of theprocess-management techniques that are integral to TOMprograms should decrease the degree to which effort iswasted through coordination losses and misdirection.Knovviedqe and skill. TOM quality teams are composed ofmembers from different functions, ensuring that there ismore talent available for work on the collective task thanwould be the case for individuals operating on their own orin homogeneous teams whose members come exclusivelyfrom a single function or unit. Moreover, the use ofstatistical analyses and data-representation techniquesshould lessen the degree to which teams make decisionsbased on misapprehensions about the state of the worksystem. Finally, use of group process-management

3211ASQ. June 1995

heuristics should enhance members' learning from oneanother. thereby increasing the total pool of talent availablefor the work. They also should decrease the likelihood thatexisting talent will be wasted. for example. by overlooking orinappropriately weighting the contributions of members whohave special talent or insight.Performance strategy. Task-performing teams sometimeshead off in the wrong direction or go about their work ininefficient or inappropriate ways. merely because membersare not entirely clear about what they are supposed to do orwhom they are supposed to satisfy. Under TOM, these risksare minimized: Customers are specifically identified and theirrequirements are clearly explicated. Customer requirementsprovide an available and appropriate test for team membersto use in inventing and choosing among alternative ways ofproceeding with their work. Moreover. changes in thoserequirements can provide a clear signal that it is time toabandon or revise existing performance routines. The use ofthe process-management and problem-solving heuristics thatare integral to the TOM package should increase thechances that members' deliberations about such matters willgenerate ways of working together that are especiallywell-aligned with what customers want.It appears. then. that these three process criteria-i-effortexpended. knowledge and skill applied. and task-performance strategies used-may be of use in assessingthe impact of TOM on how. and how well, organizationmembers work together. If TOM is working as intended.organizational units should exhibit a high standing on allthree of them.Process criteria have the important advantage of being moreaccessible to reliable measurement than are outcomecriteria. Moreover. assessing the process criteria allowsresearchers to check empirically the validity of TOMpredictions about unit processes. for example. that TOMtechniques help people work together more efficiently andproductively. Yet process criteria-whether the three thatwe have mentioned or others-are almost never addressedin TOM research. Less than 15 percent of the studies ofTOM programs that we examined document actualbehavioral changes that occur after TOM has been adopted.And those that do address work behaviors rely on anecdotaldescriptions of particular quality teams and theirproblem-solving processes. This oversight in TOM researchhas left a significant gap in knowledge about both theeffects of TOM interventions and the means by which thoseeffects are generated.What Outcomes Are Obtained?It is tempting to go for broke in research on the impact ofinterventions such as TOM, testing directly the relationshipbetween the entire intervention package and globalorganizational outcomes. The logic is straightforward: SinceTOM is supposed to enhance organizational effectiveness,then it should be a simple matter to determine whetherorganizations that use it improve on generally acceptedperformance measures. In fact. it is maddeningly difficult todo such research well. for several reasons.

322/ASQ. June 1995

Total Quality Management

First, there are serious measurement problems associatedwith even standard indices of firm performance such asmarket share, profitability, or stock price (Brief, 1984;Pennings, 1984; Kaplan and Norton, 1992); these problemsare compounded for public and nonprofit organizations(Hage, 1984). One strategy for circumventing this difficulty isto obtain productivity measures at the individual or unit leveland then to aggregate them across units. Such measuresgenerally have psychometric problems of their own,however, and the link between individual and organizationalproductivity is far from straightforward (Goodman, Lerch, andMukhopadhyay, 1994).Second, as mentioned earlier, exogenous disturbances cansignificantly obscure the link between work processes andorganizational outcomes. And even when a relationship doesexist between intervention-induced process improvementsand organizational outcomes, it may be so weak relative toother influences that it can be nearly impossible to confirmstatistically unless one has a very large sample oforganizations, which is unlikely in intervention research.Wruck and Jensen's (1994) study of the effects of TOM atSterling Chemical provides a good case in point: Despite aTOM program that by all accounts was highly successful.measures of firm performance still exhibited a monotonicdecline because of industry and market factors.

Third, temporal issues can obscure intervention-outcomerelationships (Whetton and Cameron, 1994). There often is adiscrepancy between short-term and long-term organizationalresults, and it is by no means straightforward to decide howlong after an intervention one should wait before analyzingoutcome measures. The longer one waits, the moreopportunity TOM has to realize its effects onorganization-wide results, but the more those results areopen to confounding by other factors. Even longitudinalresearch that documents changes in outcome measuresover time is of limited use in disentangling the effects of afocused intervention from those of other endogenous andexogenous changes.

Taken together, these three difficulties can make it nearlyimpossible to detect statistically the direct effects of TOMon global measures of organizational outcomes. As seriousas these problems are, they are compounded by another,namely, the process by which attributions are made aboutthe reasons for any observed performance changes. Everyresearcher knows that one cannot simply make anintervention, observe subsequent outcomes, and thenconclude that any changes in the outcome measures werecaused by the intervention. Many reports about TOMeffects, however, do precisely that: TOM is implemented,unit productivity or organizational profitability improves, and itis concluded that TOM caused the improvement (e.g.,Littman, 1991; Gilbert, 1992; Raffio, 1992). In fact. theobserved gain could have been the result of other eventsthat coincided temporally with the intervention, or by thephenomenon, sometimes referred to as the Hawthorneeffect, of people working harder when they are beingstudied. But there is an even more prosaic explanation for

323/ASQ. June 1995

performance improvements that are observed afterintroducing an organizational intervention such as TOM.In a work unit that has been operating in business-as-usualmode for some time, inefficiencies and redundancies arelikely to have gradually made that unit far less tight. lean,and efficient than it could be. Managers may select that unitas the target of an organizational change program-perhapsa quality intervention, perhaps something else. As managersand change agents begin to plan for implementation, theyscrutinize the staffing, workflow, and internal organization ofthe unit in great detail. the first close look the unit hasreceived in a long time. Any accumulated inefficiencies arelikely to be noticed and, coincident with the intervention,corrected. If productivity improvements subsequently occur, itmay seem obvious to those responsible for the changeprogram that the favorable outcomes stemmed directly fromthe intervention. Although that may be true, theimprovements may have resulted solely from thefat-trimming that accompanied, but was not integral to theintervention. Without appropriate control units and data onwork processes as well as outcomes. there is no way tochoose between the two explanations.The research literature on TOM effects includes few studieswhose designs permit definitive statements to be madeabout causes and effects. More than 80 percent of thepublished assessments of TOM describe what happenedwhen the program was installed in one particularorganization. The outcomes most frequently reported are (1)improvements in error rates. for example, "more accurateinvoices" (Teresko. 1991). (2) decreased time needed tocomplete a process, for example, "assembly-line timedecreased by 67 days" (McDonnell. 1992). and (3) dollarsavings from process efficiencies, for example. "reducedlaboratory turnaround time resulting in a $10.000 savings"(Koska, 1990). Such findings are consistent with the aims ofTOM interventions, but the absence of appropriate researchdesigns makes it impossible to attribute them directly toTOM.Straightforward evaluation research that attempts to assessthe effects of TOM on global measures of organizationaleffectiveness is fraught with both methodological difficultiesand interpretative dangers. Although well-instrumentedquasi-experimental studies can help with the design andmeasurement problems inherent in this type of research, notall of the problems can be solved. Still. the question thatoutcome-focused evaluation research studies seek toanswer cannot be finessed: It is important to know whetherTOM generates real organizational improvements. One wayto gain purchase on this question is to return to the writingsof the primary TOM authorities to extract their assertionsabout what is sought and expected from TOM programs anduse those assertions as the primary criteria for assessingTOM outcomes. Deming, Ishikawa, and Juran are clearabout what they expect TOM to achieve: (1) betterperformance in meeting customer requirements, (2)improved organizational performance capability, and (3)greater knowledge and work satisfaction on the part oforganization members. Because these three outcomes are324/ASQ. June 1995

Total Quality Management

less vulnerable to the measurement and interpretiveproblems that plague global indices of firm-wide economicperformance, they offer attractive options for research onTOM outcomes.If TOM programs do increase the degree to which customerrequirements are met and, in the process, improve theperformance capability of the organization and the well-beingof individual members, then global and economic measuresof organizational effectiveness surely should improve overthe long term as well. This prediction is stated explicitly bythe TOM theorists whose work we have been reviewing,and it strikes us as entirely reasonable. Yet, for all thereasons discussed above, it may also be a prediction thatcan never be definitively confirmed in empirical research.ConclusionResearch on the effects of TOM has focused largely onglobal outcomes. The results have been strongly positive,but they are almost all based on case reports. In part, thisproblem exists because TOM has captured more attentionfrom practitioners than from researchers: Manyassessments of TOM are descriptions written by a memberof the focal organization. But if knowledge of the effects ofTOM on organizational effectiveness is to cumulate,researchers must focus less on evaluation studies of the30-percent-gain-in-productivity variety and more on researchthat includes both explicit manipulation checks andmeasurements of process criteria.TOM practitioners are expected to focus their attention onwork processes rather than on outcome measures and touse scientific methods to improve those processescontinuously. These prescriptions rarely are applied to thestudy of TOM itself. It is ironic that the designs andmethodologies used in research on TOM fall far short of thestandards of research design, measurement, and analysisthat would be required of organization members studyingtheir own work processes under TOM.

BEHAVIORAL PROCESSES UNDER TOMTo accomplish its purposes, TOM must alter how peopleactually behave at work. As suggested in the previoussection, people should be working harder (i.e.. with moreeffort), smarter (i.e.. with greater knowledge and skill), andmore responsively (i.e., with task performance strategiesbetter attuned to customer requirements) under TOM thanotherwise would be the case. Three behavioral processesare key to achieving these aspirations: motivation, learning,and change.MotivationDeming and Ishikawa identify three different sources ofhuman motivation at work. First is intrinsic motivation, the"joy of climbing a mountain just because it is there"(Ishikawa, 1985: 27) and, more generally, growing, learning,and developing one's self (Deming, 1986: 72-86). Second istask motivation, the good feeling that comes fromaccomplishing things and seeing them actually work(Ishikawa, 1985: 28; Deming, 1986: 72). Third is social325iASQ, June 1995

motivation, the energy that comes from cooperating withothers on a shared task and the incentive provided byrecognition from others (Ishikawa, 1985: 28; Deming, 1986:107).How congruent are TOM interventions with its founders'propositions about motivation? In general. the fit is good.TOM provides people with opportunities to learn and todevelop themselves through joint problem-solving efforts.Meeting clear and often challenging customer requirementsand working to improve work processes continuouslyprovide task challenges that should both test and stretchmembers' skills. And the insistent emphasis on teamworkand cross-functional relationships provides manyopportunities for social interaction and social reinforcement.The fit between TOM practices and other motivationaltheories prominent in the organizational literature, however,is uneasy. Among the evidence we reviewed for thediscriminant validity of TOM was the fact that TOM explicitlyeschews a number of popular motivational devices, includingwork redesign (e.g., job enrichment), goal setting (e.g.,management by objectives), and performance-contingentrewards (e.q.. pay for performance). Although each of theseinterventions has been shown, in some circumstances, toenhance organization members' work motivation (Locke etal.. 1980), all of them are in some significant wayinconsistent with TOM theory and practice.Work redesign. Work-redesign theory specifies thatmotivation is strengthened when the work itself ismeaningful, when performers have considerable autonomyin determining the means by which it is accomplished, andwhen they receive regular, trustworthy knowledge aboutwork outcomes (Hackman and Oldham, 1976). The first andthird of these design specifications-meaningful work andknowledge of results-should be routinely present in TOMorganizations. The tasks of meeting customer requirementsand continuously improving processes are meaningful. andthe emphasis on analyzing data about performanceprocesses ensures that performers will be almostcontinuously aware of how they are doing.The problem comes with the second design specification,autonomy about work methods. Under TOM, much energyis spent identifying the "best" work practices, those thatbring work processes under the greatest possible control.Cross-functional teams undertake research projects todevelop or identify such practices, managers dobenchmarking visits to other orqanizations to learn aboutalternative ways of performing the work, and front-lineemployees are themselves expected to search continuouslyfor improved and simplified work practices (Juran, 1969:9-45); Ishikawa, 1985: 55-56; Deming, 1986: 49-52).Once such practices are identified and documented, they arediffused throughout the organization and standardized, withthe result that work-unit members may wind up with verylittle discretion about how they perform their tasks. Thepotential for overspecification of work procedures is so greatthat one is reminded of industrial engineering during theheyday of scientific management, when it was the job of326/ASQ. June 1995

Total Quality Management

process designers to identify the "one best way" to performthe work and the job of front-line producers to do the workprecisely that way. The motivational costs of this approachare well documented (O'Toole, 1977; Hackman and Oldham,1980).Under TOM, standardized procedures are more likely tocome from peers working on cross-functional quality teamsthan from industrial engineers. But it is not the source of theprocedures that matters, nor is it the philosophy that liesbehind their development and diffusion. What matters is thedesign of the work itself. TOM, along with other currentlypopular reengineering initiatives, runs a risk of spawning anew version of old-time scientific management, completewith the human and organizational dysfunctions thatprompted the job-enrichment corrective over two decadesago (Pallas and Neumann, 1993; Anderson,Rungtusanatham, and Schroeder, 1994; Spencer, 1994).Some commentators dispute this view. Adler (1993), forexample, interpreted findings from the NUMMI plant ofGeneral Motors-Toyota as showing that the motivationaldysfunctions of standardized, routine, and repetitive taskscan be mitigated when workers are treated fairly andrespectfully, provided with proper tools and training, andshare in decision making about performance policies andstandards. And Klein (1991, 1994) described how the loss ofemployee autonomy that accompanies standardization andprocess controls can be compensated for, to some extent,by greater collective autonomy about how tasks aredesigned in the first place.The problem is that only a subset of the members of TOMorganizations (often a small subset) have the opportunity toparticipate in the cross-functional quality teams that redesigntasks and develop improved work methods. Such teamstypically do have ample autonomy in their work, sometimeseven including the authority to implement the solutions theydevelop. Moreover, case reports document the energy,enthusiasm, and commitment exhibited by members ofthese teams. There remains, however, a pervasive andworrisome discrepancy between the motivationally engagingwork of special TOM teams and the standardized workpractices of those who perform the routine productive workof the organization.Goal setting. Goal-setting theory predicts that motivation isgreatest when performers focus their attention on achievingclear, specific, and challenging goals (Locke and Latham,1990). Although research evidence generally supports thesepredictions, TOM authorities are ambivalent about both theappropriateness and the efficacy of setting specific goals inTOM programs. Deming is opposed: "The only number thatis permissible for a manager to dangle in front of his peopleis a plain statement of fact with respect to survival. .. [suchas] 'Unless our sales improve 10 per cent next year, weshall be out of business' " (Deming, 1986: 76). Explicitgoals, he says, narrow performers' vision and implicitly invitethem to slack off once a goal is achieved. Ishikawa, bycontrast. views challenging goals as appropriate so long asthey are about solving particular problems and are327/ASQ, June 1995

established in such a way as to allow cooperation amongfunctions (Ishikawa, 1985: 60-61).It is not surprising that there is disagreement among theTOM authorities about goal setting, because the nature ofthe work done by quality teams raises some complex issuesabout how goals and objectives are properly framed. UnderTOM, the analysis of work processes focuses attention ondeviant cases, the ones indicating that a work process is, tosome extent, out of contro!. This emphasis on failures ratherthan successes also is evident in the language of TOM: Onetalks about the 1 percent of the cases for which existingprocesses did not work rather than the 99 percent of thecases for which they did (Ishikawa, 1985: 61-62; Juran,1988: 130-132). Moreover, TOM methods help teammembers identify that subset of the 1 percent thatrepresents the most pernicious or frequent problems so thatthey can be taken care of first, before moving on to the nextmost consequential problems. Both the work and the talkare more about problems than about successes.Some commentators, such as Wruck and Jensen (1994:271-272). have found value in this problem-focusedapproach. They noted that both managers and front-lineemployees exhibit a remarkable tendency to ignore ordismiss problems and mistakes, which usually results in agradual erosion of organizational performance. Because TOMinsistently focuses attention on the things that must befixed, it can partially reverse the natural human tendency toturn away from signs of trouble. But we also know that loftyand challenging goals can engender levels of workmotivation that rarely are observed when people areoccupied mainly with things that are not going right (Congerand Kanungo, 1988; Locke and Latham, 1990). How mightthese two positions be reconciled? Clearly, there is abalance to be achieved between looking at the dark and thebright sides of work processes. We need to knowconsiderably more than we do at present about how thatbalance can best be managed in the context of TOMphilosophy and practice.Pay for performance. There is abundant research evidenceabout both the motivational benefits and the risks of basingcompensation and other extrinsic rewards on measuredperformance. In contrast to their ambivalent views about thevalue of goal setting, TOM authorities are clear and decisiveabout basing pay on performance: Do not do it. Thearguments they marshall to support their position are, ingeneral. the same ones that have been well documented inthe organizational and economic research literatures. One,organizations do get what they pay for, but sometimes theyget only that (Kerr, 1975). Two, specific outcomes that arerewarded can become so salient that performers risk losingsight of the larger picture, for example, whom theorganization exists to serve or what principles are supposedto guide provision of that service. Three, performance-contingent extrinsic rewards can undermine performers'intrinsic motivation (Deci, 1971). Four, reward systems thatplace people in competition for rewards that are distributedfrom a fixed pool not only divert performers' attention fromcustomers' needs but also undermine relationships among328/ASQ, June 1995

Total Quality Management

members and make it difficult for them to work together onthe collective tasks that are the organization's real work.Five. rewards necessarily are based on some measure. butfew measures approximate the full dimensionality of thecontributions that really are needed from organizationmembers (Ishikawa. 1985: 25-28; Deming, 1986: 102).Many TOM organizations avoid all of these problems in asingle stroke. They avoid performance-contingent extrinsicrewards entirely and rely on intrinsic motivation. Thissolution, however, has significant opportunity costs. Thebest motivational state of affairs is obtained when anorganization does not rely exclusively on either intrinsic orextrinsic rewards but. instead. structures the work in a waythat fosters intrinsic motivation (for example, by providingchallenge. autonomy, and direct feedback from customers)and then supports that positive motivation withperformance-contingent extrinsic rewards. As recent findingsof Amabile (1993) show, the supposed trade-off betweenintrinsic and extrinsic motivation is neither as straightforwardnor as general as sometimes has been claimed. It is, in fact.possible to structure work and reward systems to promoteboth types of motivation simultaneously (Wageman. 1995).In TOM organizations. however. it is much easier toaccomplish that for managerial staff. who have considerableautonomy. than for rank-and-file employees who. in theirregular work. have little (Waldman. 1994).TOM authorities acknowledge that social reinforcement.which is extrinsic. can enhance the motivation oforganization members. and they do suggest thatorganization-wide gain-sharing or profit-sharing programs canappropriately be used to recognize and reward collectiveexcellence (Juran, 1974: 18.19; Ishikawa. 1985: 26-27). Ifthis can be done at the organizational level. then there wouldseem to be no conceptual reason to conclude that it couldnot also be done at the group level. A system in whichrewards were contingent on gains in performance (such asbringing work processes increasingly under control). forexample. would seem to fit rather nicely with the TOMphilosophy of continuous improvement. Theno-extrinsic-rewards principle. then. may be more anideological stance of the TOM authorities than the result ofreasoned conceptual judgment. Clearly. further thought andresearch is warranted on how. in TOM programs, one cancapture simultaneously the benefits of both intrinsic andextrinsic rewards.The motivational basis of TOM practices is generally sound.It is true that a number of positive sources of motivation areoverlooked or forgone and that. by diffusing standardized"best practices" throughout the organization. members'regular work may be less well designed motivationally than itcould be. Still. organizations that follow TOM practicesshould not fall victim to the known dysfunctions of poorlyconceived or poorly implemented motivational programs. Aninteresting possibility is to find ways that TOM organizationscan do a better job of having their motivational cake andeating it too-that is. of retaining the considerablemotivational benefits that are built into orthodox TOM whilealso harvestinq the motivational gains that can derive from

329/ASQ. June 1995

well-designed work, from specific, challenging performanceobjectives, and from the use of extrinsic rewards torecognize and reinforce extraordinary accomplishment.LearningTOM is pro-learning, with a vengeance. The movement'sfounders note, correctly, that people inherently want to learnand develop. They also point out that this inclination isfragile, that it can be undermined by social systems thatcreate fear and defensiveness. And even though theinclination to learn is built in, people also require tools andcoaching if they are to express that inclination in their workbehavior.TOM practices create good learning environments both byminimizing fear in the organizational culture and by providingmembers with a rich and diverse set of learning tools.Moreover, TOM exposes workers to data about their workprocesses more or less continuously and encourages themto use scientific methods to analyze and improve thoseprocesses. Finally, members of TOM organizations are askedto reexamine their work processes repeatedly, and do sowith no holds barred: "Ask not just why we do it that wayand can we do it better, but also ask why we do that at all"(Juran, 1969: 118). In Argyris and Schon's (1978) terms,people in TOM organizations are expected to do double-loopas well as single-loop learning.There are two quite different varieties of human learning.One variety, that on which TOM philosophy is based, is thewired-in human inclination to grow and develop incompetence. This inclination is well known to anyone whohas observed the joy experienced by very young children asthey learn to pull themselves to a standing position in theircribs, take their first unsupported steps, or use language toexpress their wishes and feelings. The other variety is therobust but often-overlooked capacity of human organisms toadapt to the many problems that life inevitably brings. Thehuman organism is capable of learning to make do evenunder conditions of profound disappointment and adversity,such as losing a spouse, a limb, or a means of livelihood.These two opposing inclinations-to stretch and grow, andto adapt and make do-are present in all of us. Schools andwork organizations are among the most important settings inwhich these opposing varieties of learning are engaged andplayed out. Experiences in these organizations have largeand enduring effects on one's personality, intellectual style,and orientation toward future learning (Kohn and Schooler,1978; Schooler, 1984). Sometimes the result is a life ofcontinuous and ever-escalating growth and learning; othertimes it is a life characterized mainly by adaptation andacceptance of one's lot.An excessive emphasis on either type of learning can bedysfunctional. Too strong an orientation toward adaptationcan result in what commonly has been found in surveystudies of job satisfaction at highly controlling organizations.People report on the surveys that they are satisfied, butcloser analysis reveals that such "satisfaction" mainlyexpresses their acceptance of a life bereft of opportunities330/ASQ. June 1995

Total Quality Management

for career and personal growth. Growth-oriented learning hasatrophied for such people, and their organizations areobtaining but a small portion of what they actually have tooffer. Too strong an orientation toward personal growth anddevelopment also is sometimes observed in workorganizations, especially those that are driven by humanistic,democratic, or spiritual ideologies. Members of suchorganizations may be encouraged to pursue their personalaspirations to such an extent that the capability of theorganization to mount efficient collective action iscompromised, eventually threatening its very survival.How well do TOM organizations succeed in providing abalance between these two varieties of learning-atpromoting continuous personal learning, but of a type and ina way that also generates efficient and effective collectiveperformance? To address this question, we examine TOMlearning processes in three arenas: (1) learning from oneanother in cross-functional quality and problem-solvingteams, (2) learning about ways to improve work processesand performance, and (3) learning about what the collectivityshould be doing-the double-loop learning espoused byArgyris and Schon (1978).Learning from one another. The cross-functional qualityteams that are among the hallmarks of TOM organizationsstack the cards in favor of learning by the simple fact thatthey are cross-functional; individual members are exposed tomore, and more diverse, points of view than would be thecase if they worked mostly by themselves or inwithin-function units. Moreover, the group-process heuristicsthat teams are taught increase the chances that this built-intalent will be used well. Members are likely to learn, forexample, how to solicit the contributions of quiet orrecalcitrant members and how to weight the contributions ofmembers in accord with their actual knowledge andexpertise rather than on the basis of task-irrelevant factorssuch as demographic attributes or interpersonal style.Moreover, the group-process techniques teams use increasethe chances that, over time, members will actively teach andlearn from one another, thereby increasing the total pool ofknowledge and skill available for the team's work. In all.TOM receives excellent marks for the ways and the extentto which it fosters interpersonal learning.Learning about work processes. TOM strategies forlearning about work processes rely heavily on numerical data(which are to be viewed as friendly even when they point todisconcerting trends), analyzed and interpreted usingscientific and statistical tools (which are viewed as providingprotection against human distortions and biases). There isreason for concern about both the sufficiency of the toolsand the readiness of the people who use them to acceptwhat the data show. Even in the institution of science,where scientific norms and tools are far more robust thanthe learning devices used in work organizations, scientistssometimes exhibit behaviors that are oriented more towardsucceeding than toward learning. Data sometimes arecollected in ways that virtually ensure that the scientist'shypothesis will be supported. Scientists sometimes continueto advocate favored theories despite the ready availability of

3311ASQ. June 1995

disconfirming data. And some scientists even have beenknown to falsify data to make things come out the way theywant them to. Merely focusing on data, and analyzing themusing scientific and statistical tools, is insufficient to ensurethe validity of the conclusions reached-even in scienceitself.

That more than just science is needed is shown byEdmondson's (1995) findings about quality-improvementefforts at a large teaching hospital. Among the data collectedin hospital patient-care units were instances of drug-relatederrors-occasions when patients received the wrong drug,the wrong dose, or the right drug at the wrong time.Although unfortunate for the patient, such events providedfriendly data for use by unit members in analyzing whatwent wrong and developing strategies to keep it fromhappening again. Edmondson found, however, that thedegree to which those data were used-or errors were evenrecorded-depended on the kind of leadership provided bythe unit's nurse manager. Units whose manager had ahands-on, supportive style reported far more errors than didunits with more distant and stringent managers. Closeanalysis of this disconcerting finding showed that it derivednot from any laxity on the part of the hands-on managersbut, instead, from the fact that staff in those managers'units were far more willing to treat error data as friendly, andtherefore to record them, than were staff in units led bymore evaluative nurse managers.

Even in TOM organizations, data and scientific tools can berendered impotent if systems, or leaders, put winning orsucceeding ahead of learning. For this reason, themovement we have seen toward increasing use ofperformance-contingent financial incentives in TOMorganizations is worrisome. The more potent thoseincentives are (and, to be effective, they should be as potentas managers can make them), the more likely it is thatorganization members will abandon the appropriate use ofdata and scientific methods in order to obtain them. Alsotroubling is the fact that implementations of TOMincreasingly are giving only lip service to the use of scientificand statistical tools. Together, these developments raiseconcerns about the future role of scientific methods-whichunquestionably are among the most distinctive features ofTOM philosophy-in actual TOM practice.

Learning about collective goals. To learn how to improvework processes is single-loop learning: The means by whichorganizational purposes are accomplished are open toanalysis and improvement but the purposes themselves arenot. To inquire about collective purposes, however, involvesdouble-loop learning: The focus is on what is being done aswell as how it is being done. Neither the philosophy nor thepractice of total quality management draws significantly onmembers' ideas or experiences about collective purposes.TOM is a single-loop, top-down undertaking that seeks toprovide those on the front line with the direction, the tools,and the coaching that they require to serve the enterprisewell. It is noteworthy that the quotation given earlier fromJuran (that one should ask not just about doing it better, but332/ASQ. June 1995

Total Quality Management

also "why we do that at all") was directed specifically tomanagers, not to rank-and-file workers.Some commentators suggest that TOM achieves anappropriate balance between managerial control andemployee participation, between single- and double-looplearning. In contrast to devices such as quality circles, it isargued, TOM institutionalizes meaningful employeeparticipation even as it retains top-down managerial controlof the enterprise (Hill, 1991). Still. some organizationmembers are likely to chafe at the uncompromisingtop-down orientation of TOM. In response, managers mayfind themselves tempted to engage in pseudo-participation:Members are invited to join in discussions about decisionsthat already have been made or that will be made bysomeone else. Pseudo-participation is ill-advised, becausepeople almost always are able to tell when they are beingmanipulated. A far preferable stance, in our view, is formanagers to be unapologetic about the fact that TOMneither espouses nor practices the engagement of allorganization members in reflective learning about collectivepurposes.Overall. TOM is about as learning-oriented as it is possiblefor a management program to be. Even so, there are manyreports in the case literature on TOM of learning failures andof antilearning group norms developing. It may be that suchreports come mainly from organizations in which TOM hasbeen implemented incompletely or incompetently. It may bethat the previous organizational experiences of membershave left a residue of antilearning habits and attitudes sostrong that it cannot be penetrated even by all of thepro-learning tools and supports that well-implemented TOMprovides. Perhaps enduring systemic forces-such as anorganization's top-down orientation, emphasis on control, orfocus on success above all else-overwhelmslearning-oriented TOM tools and practices. Or it may be thatgeneral cultural norms about learning (for example, that it isdone in school but not at work) are sometimes too strongfor TOM to counter.All these possibilities, and others, merit consideration asexplanations for those instances in which a learningorientation does not blossom even in TOM organizations.Yet we should not be too hard on TOM practitioners for notfully succeeding in realizing this part of the movement'sphilosophy. As Argyris (1993). Senge (1990). and others havenoted, it is extremely difficult to create and sustain athorough-going learning orientation in purposive enterprises.Clearly, there is both opportunity and need for research onhow to create social systems in which learning andproduction go hand in hand (Sitkin, Sutcliffe, and Schroeder,1994). Organizations in which TOM has been well and fullyimplemented provide a high platform for continuing theprocess of learning about learning in social systems.ChangeUnder TOM, organization members are expected to improvework processes continuously so that their customers areserved as well as possible. This can result in performancestrategies uniquely well tailored to environmental constraints

333/ASQ. June 1995

and resources, especially those emanating from customers.It is one of the great strengths of the TOM approach. Butwhat happens when the environment changes? Ideally,organizational units would be so closely in touch with theenvironment-including process innovations introduced byother organizations as well as changes in customers'needs-that they could adapt their performance strategieswith little lag and, perhaps, even stay a step or two ahead ofcustomers' wishes. This level of responsiveness, however,is unusual even in TOM organizations where continuousimprovement is a core value and organizational policies andpractices are specifically tailored to foster responsiveness toenvironmental changes.The difficulties encountered by TOM units in responding toexogenous changes are of two types-one rooted in howpeople process information, the other in their tendency todevelop emotional commitment to locally invented ways ofoperating.Information processing. Once an individual. a group, or anorganization develops a strategy for handling a certain kindof problem, that strategy is likely to become a standardroutine, what we call. at the individual level. a habit (Marchand Simon, 1958; Gersick and Hackman, 1990; Louis andSutton, 1991). Routines contribute enormously to theefficiency of daily life. Over time, however, they can becomeso integral to social systems that members continue to relyon them even when the situation changes markedly. InLanger's (1989) terms, individuals and groups tend to goabout their work mindlessly, which can result in significantperformance decrements when environmental opportunitiesand constraints change.Moreover, when members do discover that things are notworking as well as they did formerly, they commonlyrespond by executing their existing behavioral routines morevigorously than ever, rather than using the early signs oftrouble as an occasion for reflection on the adequacy ofthose routines. The individual-level dynamics of thisphenomenon are well known. A state of arousal increasesthe likelihood that one's dominant response (the one highestin the hierarchy of possible responses) will be exhibited withminimal cognitive mediation. Similar dynamics are observedat the group and organizational levels of analysis: Underarousing conditions (of which learning that standard routinesare no longer generating hoped-for results is an instance),performers are as likely to exhibit those routines with evengreater vigor as they are to inspect, reflect upon, andreconsider them (Staw. Sandelands, and Dutton, 1981;Ocasio, 1995).Emotional commitment. The second impediment toresponsiveness is rooted more in emotionality than inefficiency. When a work unit has invented its ownperformance strategy-which certainly is more characteristicof TOM organizations than of those with traditionalmanagement hierarchies-members can become quitereluctant to change it. The research literature is filled withcases in which a group or organization had ample datashowing that strategic change was called for, but performers

334/ASQ. June 1995

Total Quality Management

either ignored those data. dismissed them as irrelevant. orengaged in manifestly irrational behaviors to avoid having todeal with them (Janis. 1982; Argyris. 1993). Only when thesituation gets so bad that it threatens the unit's very survival(at which time it may be too late to recover) can one counton a social system taking seriously the need to makesignificant change (Miller and Friesen. 1980; Jensen, 1993).Both processes sketched above contribute to the difficultieswork units have in altering their performance strategies inresponse to environmental change. Moreover, the likelihoodof responsive change may be lessened under preciselythose circumstances in which change is needed most(Miller. 1993). There are exceptions. of course. Some groupsand organizations do manage to stay even with. or evenahead of, changes in their environments. Some evenwelcome signs of trouble and use them as occasions forinnovation rather than as a time to dig in their heels and domore of what has always been done. Still. the tendency tohold onto familiar, locally invented performance strategiesappears to be nearly as pervasive as it is pernicious.TOM organizations should be more likely than most totranscend the limitations of human information processingand emotionality in responding to environmental change. Thelearning orientation of TOM organizations. their commitmentto management by fact. and their close links with customersshould increase the likelihood that members will attend tounsettling data before they become too threatening to dealwith. Moreover, TOM units are especially likely to encounterthe conditions that Gersick and Hackman (1990) haveidentified as those that prompt inspection andreconsideration of habitual routines, such as having to dealwith novelty. receiving an external intervention, or coming toterms with structural alterations of the performing unit. Evenso. the challenge of altering performance strategies-especially those that have been locally invented-is asignificant one for TOM organizations. Precisely because oftheir orientation to learning, data. and customers. suchorganizations should be a fruitful site for research thatidentifies and explores alternative ways of meeting thatchallenge.

PROSPECTS FOR TOMAlthough dinner may seem assured to a snake who noticesa rabbit strolling nearby. there is no guarantee ofnourishment. If the rabbit is extraordinarily large, it may getstuck in the snake's throat; snakes have. on occasion. diedwhen their eyes were too large for their throat. And if therabbit is just a baby, consumption and digestion are easy.but there is little real nourishment. Eating a baby rabbit ishardly worth the trouble it takes to catch it.

Organizational change programs. including TOM, can gowrong for the same two reasons that snakes sometimeshave trouble with their dinners. One. the changes may be soambitious and involve such fundamental alterations of thesocial system that. for all their potential merit. theorganization cannot accommodate to them. Espousedchanges may appear to fail when in fact they never got

335/ASQ. June 1995

SGrant. Sharu, and Krishnan (1994) providean alternative view: that TOM is arevolutionary philosophy that does requirefundamental changes in the verypremises of an organization.

implemented. Two. the changes may be morewindow-dressing than real. as in a program that exhortspeople to alter their behavior but that requires managers todo little other than issue the exhortation. In this case.implementation is easy, but the old organizational structuresand systems remain untouched and continue to generate thesame behavioral dynamics as before.When implemented fully and well, TOM can thread its waybetween these two extremes (Reger et al.. 1994). TOMchanges are real rather than ephemeral. and they aregenerally consistent with research evidence about thefactors that promote performance effectiveness. When TOMdoes tilt toward one or the other extreme, it invariably istoward interventions that are too modest. The reason is thatTOM, by philosophy and design, skirts four features of worksystems that are fundamental to organizational behavior andperformance: (1) how front-line work is structured. (2) howgains are allocated, (3) how opportunities for learning areapportioned, and (4) how authority is distributed. For TOM,the aspiration is to implement changes that are substantialenough to make a real difference without altering the corepremises of the enterprise-in effect. to achievefundamental change without changing the tundarnentals."This aspiration presents practitioners with a series oftroublesome dilemmas, the resolution of which often resultsin changes that are less risky, but also far less impactful,than would have been the case had the fundamentals beenaddressed frontally.

Design of work. Cross-functional quality teams and taskforces are among the most common features of TOMorganizations, and the work of such teams is usually quitewell designed motivationally. By contrast. much lessattention is given to the design of the work of front-lineproducers. Aside from the opportunity some employeeshave to work on quality teams or to perform supportactivities of the type formerly done by organization staff. themotivational structure of front-line jobs is unaltered in many,perhaps most, TOM implementations:Dilemma 1: Motivating front-line organization members towardcontinuous improvement and the highest quality of output-butdoing so without fundamentally changing the motivational structureof their work.Allocation of gains. TOM philosophy is explicit that extrinsicrewards, includinq pay. should not be contingent onmeasured individual or team performance. This stanceleaves some powerful motivational cards on the table. It is,moreover, unstable over the longer term. When workersperceive that they are contributing more to the organizationthan they did previously, their initial response may be prideand pleasure. That may suffice for a while. Eventually,however, members of profit-making firms will realize thatsomebody is making more money as a result of their greatercontributions, and it is not them. At that point, they maybegin to withdraw their commitment to the enterprise, andsigns of a motivational backlash may even be seen:Dilemma 2: Creating the kind of alignment with, and commitmentto, organizational purposes that can come from sharing in collective

336/ASQ. June 1995

Total Quality Management

gains-but doing so without actually altering how organizationalrewards are apportioned.Opportunities for learning. The TOM aspiration ofcontinuous improvement in meeting customer requirementsis supported by a thorough-going learning orientation,including substantial investments in training and thewidespread use of statistical and interpersonal techniquesdesigned to promote individual and team learning. Thisorientation extends even beyond the organization'sboundaries. as is seen in programs to teach quality practicesto suppliers and in benchmarking visits to capture the bestof other organizations' ideas and innovations.Although front-line employees are encouraged to findever-better ways to accomplish their assigned tasks, theyare not invited to reflect on the purposes their work serves.Moreover, once something has been discovered thatimproves work practices-whether from the deliberations ofa quality team. from benchmarking. or from a front-line workunit-then that improvement is likely to be identified as abest practice that everyone i