Suspended In Self-Spun Webs Of Significance: A Rhetorical Model Of Institutionalization And...

27
SUSPENDED IN SELF-SPUN WEBS OF SIGNIFICANCE: A RHETORICAL MODEL OF INSTITUTIONALIZATION AND INSTITUTIONALLY EMBEDDED AGENCY SANDY EDWARD GREEN JR. YUAN LI University of Southern California NITIN NOHRIA Harvard University This article employs rhetorical theory to reconceptualize institutionalization as change in argument structure. As a state, institutionalization is embodied in the structure of argument used to justify a practice at a given point in time. As a process, institutionalization is modeled as changes in the structure of arguments used to justify a practice over time. We use rhetoric surrounding the institutionalization of total quality management (TQM) practices within the American business community as a case study to illustrate how conceptualizing institutionalization as changes in argu- ment structure can help show how institutions simultaneously constrain and enable social action. Man is an animal suspended in webs of significance he himself has spun. -Clifford Geertz Institutions are more than persistent material prac- tices and structures; they are also accompanied by systems of signs and symbols that rationalize and legitimize those practices (Friedland & Alford, 1991; Strang & Meyer, 1994). Yet how do we know when persistent material practices are “institutionaliz- ing”—that is, acquiring legitimacy? Despite increas- ing attention to conceptualizing and specifying the process of institutionalization (Tolbert & Zucker, 1996), past neoinstitutional research has failed to ad- dress this question adequately because institutional accounts often theoretically or empirically equate the acquisition of legitimacy to the outcome of legitima- tion: persistence, prevalence, or change in material practices (Barley & Tolbert, 1997; Baum & Powell, 1995; Suddaby & Greenwood, 2005; Zucker, 1989). However, equating prevalence or any material change with legitimacy overemphasizes the material aspect of institutions at the expense of their symbolic aspect. Neoinstitutional theory has often been criticized for this overly materialist and realist focus (Abrahamson & Fairchild, 1999; Green, 2004; Phillips, Lawrence, & Hardy, 2004; Strang & Meyer, 1994). Critics have called for a more ideational or cognitive focus that captures how the content or structure of discourse reflects and shapes institutionalization (Strang & Meyer, 1994). We argue that rhetorical theory, which empha- sizes a direct relationship between the language/ cognition and action of actors, may help scholars to develop a symbolic or cognitive conception of in- stitutionalization. Rhetoric is “discourse calculated to influence an audience toward some end” (Gill & Whedbee, 1997: 157). Through rhetoric, actors shape the legitimacy of practices by making persua- sive arguments that justify and rationalize practices (Green, 2004; Suddaby & Greenwood, 2005). We use rhetorical theory to reconceptualize institution- alization as changes in the structure of arguments. As a state, institutionalization is embodied in the structure of arguments used to justify a practice at a given point in time. As a process, institutionaliza- tion is modeled as changes in the structure of argu- ments used to justify a practice over time. Within this framework, to rationalize is to give discursive reasons for practices; to legitimize or institutional- ize is to accept and take these reasons for granted (Green, 2004). Our core insight is that, as a material practice acquires legitimacy and institutionalizes, the complexity of the argument used to support that practice collapses and becomes more simple. Moreover, the collapse and simplification of an argument provide a robust symbolic measure of institutionalization. We thank Chet Miller and three anonymous reviewers for their helpful comments on drafts of this work. We also thank Thomas Goodnight, Paul Adler, Mark Kennedy, Thomas Cummings, and Hayagreeva Rao. All remaining errors and omissions are entirely our own. Academy of Management Journal 2009, Vol. 52, No. 1, 11–36. 11 Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download or email articles for individual use only.

Transcript of Suspended In Self-Spun Webs Of Significance: A Rhetorical Model Of Institutionalization And...

SUSPENDED IN SELF-SPUN WEBS OF SIGNIFICANCE: ARHETORICAL MODEL OF INSTITUTIONALIZATION AND

INSTITUTIONALLY EMBEDDED AGENCY

SANDY EDWARD GREEN JR.YUAN LI

University of Southern California

NITIN NOHRIAHarvard University

This article employs rhetorical theory to reconceptualize institutionalization aschange in argument structure. As a state, institutionalization is embodied in thestructure of argument used to justify a practice at a given point in time. As a process,institutionalization is modeled as changes in the structure of arguments used to justifya practice over time. We use rhetoric surrounding the institutionalization of totalquality management (TQM) practices within the American business community as acase study to illustrate how conceptualizing institutionalization as changes in argu-ment structure can help show how institutions simultaneously constrain and enablesocial action.

Man is an animal suspended in webs of significancehe himself has spun.

-Clifford Geertz

Institutions are more than persistent material prac-tices and structures; they are also accompanied bysystems of signs and symbols that rationalize andlegitimize those practices (Friedland & Alford, 1991;Strang & Meyer, 1994). Yet how do we know whenpersistent material practices are “institutionaliz-ing”—that is, acquiring legitimacy? Despite increas-ing attention to conceptualizing and specifying theprocess of institutionalization (Tolbert & Zucker,1996), past neoinstitutional research has failed to ad-dress this question adequately because institutionalaccounts often theoretically or empirically equate theacquisition of legitimacy to the outcome of legitima-tion: persistence, prevalence, or change in materialpractices (Barley & Tolbert, 1997; Baum & Powell,1995; Suddaby & Greenwood, 2005; Zucker, 1989).However, equating prevalence or any material changewith legitimacy overemphasizes the material aspectof institutions at the expense of their symbolic aspect.Neoinstitutional theory has often been criticized forthis overly materialist and realist focus (Abrahamson& Fairchild, 1999; Green, 2004; Phillips, Lawrence, &

Hardy, 2004; Strang & Meyer, 1994). Critics havecalled for a more ideational or cognitive focus thatcaptures how the content or structure of discoursereflects and shapes institutionalization (Strang &Meyer, 1994).

We argue that rhetorical theory, which empha-sizes a direct relationship between the language/cognition and action of actors, may help scholars todevelop a symbolic or cognitive conception of in-stitutionalization. Rhetoric is “discourse calculatedto influence an audience toward some end” (Gill &Whedbee, 1997: 157). Through rhetoric, actorsshape the legitimacy of practices by making persua-sive arguments that justify and rationalize practices(Green, 2004; Suddaby & Greenwood, 2005). Weuse rhetorical theory to reconceptualize institution-alization as changes in the structure of arguments.As a state, institutionalization is embodied in thestructure of arguments used to justify a practice at agiven point in time. As a process, institutionaliza-tion is modeled as changes in the structure of argu-ments used to justify a practice over time. Withinthis framework, to rationalize is to give discursivereasons for practices; to legitimize or institutional-ize is to accept and take these reasons for granted(Green, 2004). Our core insight is that, as a materialpractice acquires legitimacy and institutionalizes,the complexity of the argument used to supportthat practice collapses and becomes more simple.Moreover, the collapse and simplification of anargument provide a robust symbolic measure ofinstitutionalization.

We thank Chet Miller and three anonymous reviewersfor their helpful comments on drafts of this work. Wealso thank Thomas Goodnight, Paul Adler, MarkKennedy, Thomas Cummings, and Hayagreeva Rao. Allremaining errors and omissions are entirely our own.

� Academy of Management Journal2009, Vol. 52, No. 1, 11–36.

11

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download or email articles for individual use only.

We believe that this symbolic or linguistic con-ceptualization of institutionalization contributes tomanagement and organizational theory in severalways. First, it provides a framework for analyzingshifts and changes in managerial rationality. Weposit that the structure of arguments supporting apractice reflects and also shapes managerial reason-ing and logic. Second, a linguistic conception ofinstitutionalization suggests that a material prac-tice becomes legitimate to the extent it is supportedby a compelling and convincing rationale that ac-counts for its existence and enables its reproduc-tion. Within this framework, rhetoric matters be-cause rhetoric rationalizes the adoption of apractice, enabling it to diffuse and persist. Before apractice can institutionalize and diffuse, that prac-tice must make sense. Rhetoric or discursive rea-soning helps define what a practice means and whythe adoption, broad diffusion, and prevalence ofthat practice are sensible and appropriate. Third, asymbolic conception of institutionalization takesseriously recent suggestions by institutional theo-rists to move beyond models of institutionalizationthat focus on the sociological realism of behavioralconformity or prevalence to explanations that em-phasize shared cognitions and meanings (Friedland& Mohr, 2004; Green, 2004; Strang & Meyer, 1994).Our model extends these ideas emphasizing rheto-ric as an important element of institutionally em-bedded agency. Our emphasis on symbolism andrhetoric is not meant to minimize the complexity ofagency, institutions, or the material. We hope onlyto show the important role of the symbolic in insti-tutional processes.

To illustrate our ideas, we analyze the rhetoricsurrounding the institutionalization of total qualitymanagement (TQM) practices within the Americanbusiness community as a case study. Through tex-tual analysis of the form and content of justifica-tions in a comprehensive database of TQM dis-course, we show how institutionalization,conceptualized as changes in argument structure,developed and changed over time. We concludewith a discussion of the implications of our modelfor theory and research.

INSTITUTIONALIZATION ANDARGUMENT STRUCTURE

Our central argument rests upon three points.First, neoinstitutional accounts in which the insti-tutionalization or the acquisition of legitimacy isconceptualized as changes in the material preva-lence or persistence of a practice are inadequate.Second, the phenomenological or cognitive tradi-tion in neoinstitutional theory suggests an alterna-

tive conceptualization of institutionalization as alinguistic process. Finally, we argue that rhetoricaltheory offers a powerful set of tools and methodsfor explaining and describing a linguistic concep-tion of institutionalization.

Institutionalization as Change inMaterial Prevalence

Institutions are “simultaneously material andsymbolic” (Friedland et al., 1991: 241). Institutionsare the material patterns of human activity throughwhich “individuals and organizations produce andreproduce their material subsistence and organizetime and space” (Friedland et al., 1991: 243). Insti-tutions are “also symbolic systems, ways of order-ing reality, and thereby rendering experience oftime and space meaningful” (Friedland et al., 1991:243). In addition, the symbolic systems of institu-tions have important ideational and linguistic com-ponents critical to providing material practiceswith meaning and legitimacy (Berger & Luckmann,1966; Green, 2004; Strang & Meyer, 1994).

Although the acquisition of legitimacy is at thecore of all institutional accounts (Elster, 1992), pastneoinstitutional research often theoretically andempirically equates the acquisition of legitimacywith changes in the persistence of material prac-tices (Barley et al., 1997; Suddaby & Greenwood,2005). For example, studying the spread of civilservice practices, Tolbert and Zucker (1983) mea-sured the acquisition of legitimacy or institutional-ization as the prevalence or spread of practices inthe presence of a decline in the predictive power ofmaterial variables such as city size, age, and popu-lation. Similarly, in a study examining the spreadof TQM in hospitals, Westphal, Gulati, and Shortell(1997) measured institutionalization as the changein the material component of TQM: the adoption of“standardized” TQM practices vs. “customized”TQM practices (Westphal et al., 1997). In addition,more recent institutional studies describe the sym-bolic activities of institutional entrepreneurs asshaping institutional logics and legitimacy. How-ever, the empirical evidence that legitimacy has infact been achieved is never explicitly described ormodeled (see Covaleski, Dirsmith, & Rittenberg,2003; also see Suddaby & Greenwood, 2005). Thesestudies provide persuasive examples of how varia-tion in language could shape legitimacy, but with-out an explicit measure of legitimacy researchersare unable to ascertain if in fact legitimacy isattained.

Moreover, prevalence or the persistence of apractice or structure is a poor indicator or cause oflegitimacy. Sometimes a practice is prevalent yet

12 FebruaryAcademy of Management Journal

not legitimate, and sometimes a practice is legiti-mate yet not prevalent (Green, 2004). As the liter-ature on management fashion reminds us, whennorms of conformity dominate, increases in preva-lence stimulate legitimacy, yet when norms ofprogress dominate, increases in prevalence reducelegitimacy (Abrahamson, 1996; Kieser, 1997). Forexample, under norms of conformity actors mayinterpret previous adoption as “best practices,” orappropriate solutions to common problems. In thiscase, prevalence increases legitimacy. However,under norms of progress actors may interpret pre-vious adoption as failures to innovate or develop“cutting-edge” practices. In this case, prevalencedecreases legitimacy. As a result of the ambiguouslink between prevalence and legitimacy, studiesthat use prevalence as an indication of legitimacycannot distinguish whether the prevalence of a ma-terial practice is driven by the pursuit of legitimacy(i.e., institutional accounts) or the pursuit of re-sources, efficiency, and so forth (e.g., resource de-pendency or economic accounts) (Donaldson,1995).

A Phenomenological or Cognitive Viewof Institutionalization

Although most institutional accounts equate in-stitutionalization with prevalence, institutional ac-counts grounded in the phenomenological tradi-tion suggest that institutionalization is inherently alinguistic process. Many scholars consider the phe-nomenological tradition the distinguishing contri-bution of institutionalism within sociology (Di-Maggio & Powell, 1991; Scott, 1995). The argumentof this tradition is that “the edifice of legitimationsis built upon language and uses language as itsprincipal instrumentality” (Berger & Luckmann,1966: 64). The phenomenological tradition also em-phasizes the importance of cognitive legitimacy.Cognitive legitimacy is concerned with satisfyingcollective standards of appropriate behavior by ex-plaining or justifying the social order in a way thatmotivates actors to enact actions within a compre-hensible, meaningful world (Berger, Berger, & Kell-ner, 1974; Berger & Luckmann, 1966; Wuthnow,Hunter, Bergesen, & Kurzwell, 1984). Languageshapes cognitive legitimacy through institutionallogics. “Institutional logics are the belief systemsthat furnish guidelines for practical action” (Rao,2003: 705–706). As belief systems, institutional log-ics encode the criteria for legitimacy by shapingindividual and collective understandings of whatmaterial conditions are problematic, as well aswhat material practices represent appropriate solu-tions to these problems (Fligstein, 1990; Friedland

& Alford, 1991; Thornton, 2002; Thornton & Oca-sio, 1999).

Within this framework, a material practice be-comes legitimate to the extent it is supported by acompelling and convincing rationale that accountsfor its existence and enables its reproduction(Berger & Luckmann, 1966; Douglas, 1986; Fried-land & Alford, 1991; Strang & Meyer, 1994). Neoin-stitutional scholars label the development of theserationales “theorization.” Theorization is the “self-conscious development and specification of ab-stract categories and the formulation of patternedrelationship such as chains of cause and effect”(Strang & Meyer, 1994: 104). Theorization is essen-tially the symbolic or discursive process of match-ing adopters to practices and practices to adopters.Specifically, theorization of adopters defines pop-ulations within which the diffusion of the practiceis imaginable, sensible, and legitimate. Similarly,theorization of practices describes the outcomesfrom adopting a practice as effective and legitimate(Strang & Meyer, 1994).

Accordingly, material practices driven by norma-tive, regulative, or mimetic pressures (DiMaggio &Powell, 1983) shape legitimacy to the extent theyare theorized—that is, represented, interpreted,and legitimated through language (Phillips et al.,2004). This linguistic conception of institutional-ization emphasizes the role of language in the pro-duction and acquisition of legitimacy.

In the next section, we use rhetorical theory todevelop a linguistic model of institutionalizationthat links changes in reasons or arguments tochanges in cognitive legitimacy.

Rhetoric

There is a growing interest in the use of rhetoricaltheory in organizational studies (Abrahamson,1997; Eccles, Nohria, & Berkeley, 1992; Emrich,Brower, Feldman, & Garland, 2001; Green, 2004;Nohria & Harrington, 1994; Suddaby & Greenwood,2005; Watson, 1995; Zbaracki, 1998). Although rhe-torical theory is broad and complex, most scholarsdivide it into two large domains: classical rhetoricand new rhetoric. Whereas classical rhetoric em-phasizes what the rhetor communicates (Aristotle,1991; Bizzell & Herzberg, 1990), new rhetoric fo-cuses on the audience for and the social aspects ofthe communication (Burke, 1969a; Enos & Brown,1993; Perelman & Olbrechts-Tyteca, 1969). Thespecific rhetorical perspective we use is the “strongcompatibilist perspective” (see O’Neill, 1997),which combines concerns of both classical andnew rhetoric while also emphasizing the compati-bility of rhetoric and reason. Rhetoric and reason

2009 13Green, Li, and Nohria

are “compatible and inextricably linked, becausereason encompasses both the logic of demonstra-tive argument—premises that are self-evidentlytrue—and the rhetoric of dialectical argument—premises based on endoxa or opinions that are gen-erally accepted” (Green, 2004: 655). The rhetoricalconception of institutionalization we advocatebuilds upon the inseparability of rhetoric and rea-son because cognitive legitimacy—comprehensionand “taken-for-grantedness”—is described as es-sentially a discursive production (Green, 2004).Specifically, to be rational is to provide reasons orarguments that are comprehensible and make per-suasive sense. As these arguments are accepted andtaken-for-granted, the ideas and practices theysupport are rationalized and institutionalized(Green, 2004).

This framework suggests an empirical techniquefor modeling institutionalization. Specifically,neoinstitutional theorists suggest that practices areinstitutionalized to the extent they are embeddedin taken-for-granted assumptions (Zucker, 1977). Arhetorical view of institutionalization builds uponthese ideas, yet it suggests that institutionalizedpractices are embedded in reasons or arguments asopposed to assumptions or beliefs. These reasonsor arguments reflect and shape the institutionallogics or belief systems that guide practical action.

Most importantly, these reasons and arguments, asopposed to immeasurable beliefs, logics, or as-sumptions, can be empirically measured as presentor absent over time as practices diffuse and change.In addition, rhetorical theory provides a model foridentifying and measuring reasons. We discuss thisrhetorical model in the next section.

A Syllogistic Model of Argument

Aristotle’s rhetoric, and in particular his syllogis-tic logic, provides a model for identifying and mea-suring changes in reasons or arguments. Specifi-cally, a syllogism consists of three statedpropositions: a major premise, a minor premise,and a conclusion (Aristotle, 1989, 1991). Individu-als use a major premise and combine it with aminor premise to produce a claim. The differencebetween the major premise and the minor premiseis that the major premise is more entrenched inendoxa. Endoxa are the commonly held opinionsand beliefs of a discursive community. Accord-ingly, the major premise is more commonly heldand socially accepted than the minor premise,which is often drawn from local or individualknowledge and experience (Areni, 2002; Toulmin,1969). Figure 1 illustrates the structure of an argu-ment within the syllogistic model.

FIGURE 1Argument Structure

14 FebruaryAcademy of Management Journal

Grammatically, the minor premise presented inFigure 1 specifies a relation between a subject(Socrates) and a middle term (man), and the majorpremise describes a relation between a predicate(mortals) and the middle term. The conclusion(Socrates is mortal) delineates a valid or an invalidrelation between the subject and the predicatebased on the relations stated in the premises (Ar-eni, 2003; Aristotle, 1989; Corbett & Connors, 1999;McGuire, 1960; Revlin & Leirer, 1978). This syllo-gistic model is elaborated in Toulmin’s jurispru-dence model of argument (Toulmin, 1969; Toul-min, Reike, & Janik, 1984). According to Toulmin’smodel, an argument has three essential elements:warrant, data, and claims. Warrant, data, andclaims mirror major premise, minor premise, andconclusion (Areni, 2003; Hart & Daughton, 2005;Toulmin, 1969).

We use both Aristotle’s and Toulmin’s models ofargument. Although both frameworks agree on thegeneral structure of arguments, Aristotle’s frame-work emphasizes formal logic, or the validity of anargument (Aristotle, 1989), but Toulmin’s modelemphasizes informal logic, or the rhetoric of anargument in everyday use (Toulmin, 1969). We useboth frameworks because our analysis is concernedwith the logic of how the component parts of anargument are structurally related, as well as withwhen the component parts of an argument are suf-ficient or persuasive enough to be accepted andtaken-for-granted.

Argument Structure and the Processof Institutionalization

We argue that initially a conclusion or claimjustifying and supporting a material practice is syl-logistically produced: the claim is explicitly linkedto the major premise and the minor premise. Al-though numerous syllogisms—or potential justifi-cations for material practices—may be made at agiven point in time, some are particularly appeal-ing and successful, and for various reasons thesemay become accepted and taken-for-granted andwill institutionalize into future endoxa (Green,2004). As institutionalization takes place, oneshould observe specific changes in the structure ofthe argument supporting a material practice. Spe-cifically, a decline in stating the major premise ofan argument justifying a material practice indicatesthat the material practice is becoming more legiti-mate, taken-for-granted, and institutionalized.

Arguments with omitted major premises arecalled “enthymemes” (Areni, 2002; Corbett & Con-ners, 1999; Gill & Whedbee, 1997); thus, the begin-ning of an institutionalization process should coin-

cide with the collapse of the syllogism into anenthymeme that suppresses the major premise. Forexample, as the idea “Socrates is mortal” institu-tionalizes, one should observe a decrease in thepublic need to state the major premise, “All menare mortal.” The decrease in the need to state orinvoke the major premise is evidence of its accep-tance by the audience, and it is due to the cognitiveand attention limits of both the speaker and theaudience (Perelman, 1982). The unstated premiseis so obvious to speaker and audience that stating itis redundant and repetitive, wasting precious andlimited cognitive resources (Aristotle, 1991). Al-though an enthymeme may represent a purposefuland persuasive attempt to have the audience sup-ply the major premise, when the enthymeme isused for a period of time after the syllogism hasbeen used for a period of time, the disappearance ofthe major premise more likely represents the extentto which the major premise is accepted, taken-for-granted, and inscribed in the knowledge structureof actors participating in the idea, or institution(Perelman, 1982).

The suppression of the major premise or collapseof argument structure from complete syllogism toenthymeme also indicates an increase in the cogni-tive legitimacy or taken-for-grantedness attached toboth the remaining stated propositions and the be-lief or material practice that the suppressed propo-sition supports. Over time, as institutionalizationincreases even more, the minor premise is sup-pressed, leaving only the claim, and the audienceprovides both the major and minor premise. At thisstage, both the claim and the corresponding beliefor material practice are highly institutionalized. Inaddition, the claim at this point has become a partof endoxa. As such, it may then be used as themajor premise in constructing new institutions.Figure 2 outlines this relationship between argu-ment structure and institutionalization.

Persuasion and cognitive limits are the how andwhy of shifts in argument structure. Holding allelse constant, a claim with a premise is more per-suasive than a claim without a premise. That is, asyllogism is more persuasive than an enthymeme,and an enthymeme is more persuasive than an un-supported claim. Although not the case in all situ-ations, in most situations and in most cases claimswith reasons are more persuasive than claims with-out reasons. In most cases, it is the presence ofjustification that persuades. For instance, juriesneed reasons to convict, voters to vote, scholars tobelieve, managers to act. Complex arguments usemore of people’s limited cognitive and attentionresources than simple arguments. If complex argu-ments are persuasive, the efficient use of limited

2009 15Green, Li, and Nohria

cognitive resources suggests the simplification orcollapse of complex arguments into simple argu-ments. This collapse in argument structure is bothevidence of past persuasion and an indication oftaken-for-grantedness.

This rhetorical model of institutionalization res-onates with several insights from neoinstitutionaltheorists. Specifically, in theorizing the relation-ship of discourse and institutionalization, scholarshave suggested that, at initial stages of institution-alization, discourse actively advocates the moral orpragmatic value of a material practice (e.g., Green,2004). This discursive support produces pragmaticor moral legitimacy and reflects a lack or at least alow level of cognitive legitimacy or taken-for-grant-edness (Greenwood, Suddaby, & Hinings, 2002;Suchman, 1995). At later stages of institutionaliza-tion, discursive justification of a material practiceis low, reflecting broad acceptance of its moral orpragmatic value, and thus cognitive legitimacy ortaken-for-grantedness is high (Aldrich & Fiol, 1994;Greenwood et al., 2002; Jepperson, 1991; Suchman,1995). In short, at initial stages of institutionaliza-tion new material practices are supported with syl-logistic, or expanded, arguments that advocate themoral or pragmatic value of the material practice. Ifthe syllogistic argument is persuasive, the materialpractice acquires moral or pragmatic legitimacy.Over time, a persuasive syllogistic argument col-lapses into a more simple argument or enthymeme,and finally becomes the simplest argument of all, aclaim. This collapse of argument structure, or de-crease in discursive justification supporting thematerial practice, reflects a rise in cognitive legiti-macy: comprehension followed by taken-for-grant-edness (Suchman, 1995). With the rise of cognitivelegitimacy, the material practice becomes infusedwith value (i.e., cognitive legitimacy) beyond itsinitial value (i.e., pragmatic legitimacy or morallegitimacy) (Selznick, 1949).

To summarize, the rhetorical model of institu-tionalization describes how the material and sym-bolic aspects of institutions coevolve. Specifically,during institutionalization material practices are

justified and made legitimate by arguments. Insti-tutionalization takes place when the argumentsused to support a practice collapse from syllogismsinto enthymemes and finally into claims. The col-lapse of argument structure from syllogism to en-thymeme indicates an increase in the taken-for-grantedness and legitimacy attached to the materialpractice and the remaining stated propositions.Over time, as institutionalization increases evenmore, the minor premise is suppressed, leavingonly the claim. At this stage, the claim is highlyinstitutionalized and has become a part of endoxa.As a part of endoxa, it may then be used as themajor premise in constructing new or different ar-guments as well as in supporting new or differentmaterial practices. Alternatively, the argumentsmay not collapse but rather stay contested for along time while the practice continues to diffuse, inwhich case diffusion but not institutionalizationoccurs. Within this framework, the state of institu-tionalization, or level of cognitive legitimacy, of amaterial practice is embodied in the structure of thearguments used to support the practice at a givenpoint in time. As a process, institutionalization ismodeled as changes in the structure of argumentsused to justify a material practice over time.

DATA AND METHODS

We illustrate our rhetorical model of institution-alization by examining the institutionalization ofTQM, which is the development of a set of prac-tices intended to produce and deliver high-qualityproducts and/or services to customers while alsopromoting the stability of employees and the sat-isfaction and growth of organization members(Deming, 1986; Ishikawa, 1985; Juran, 1989). Theadoption of TQM practices often reflects a manage-ment’s decision to follow a customer-focused busi-ness strategy of increasing product or service qual-ity while reducing product or service cost. Thegoals of this strategy are typically to enhance acompany’s profit margins, compete in leaner mar-

FIGURE 2Argument Structure and Institutionalization

16 FebruaryAcademy of Management Journal

kets, increase market share, and better satisfy cus-tomers’ needs.

There were several reasons for using TQM toexamine the relationship of institutionalizationand argument structure. First, most scholars be-lieve that by the early 1990s TQM was institution-alized and a central part of business thinking(Benson, 1993; Cole, 1999; Powell, 1995). Thiswidespread belief assured us that we were examin-ing the institutionalization of a practice. Second,many of the assumptions and core logics of TQMeventually replaced old assumptions and acceptedbusiness logics (Lawler & Mohrman, 1985). Thischange gave us a chance to examine changes in thecognitions or symbolic systems that accompany in-stitutional change. Third, computer-readable, full-text articles about TQM are available, a fact thatallowed us to access TQM rhetoric or argumentstructure online.

A Brief Overview of TQM

TQM, like all institutions, is constituted by boththe symbolic and the material (Zbaracki, 1998). Onthe material side, TQM encompasses physicaltools, methodologies, and practices aimed at qual-ity improvement, including quality control circles,statistical process control, quality function deploy-ment, cross-functional teams, and benchmarking,to name a few (Deming, 1986; Ishikawa, 1985; Ju-ran, 1989). The material practices of TQM spreadrapidly and broadly throughout Fortune 100 firms.For example, as late as 1977, none of those firmshad adopted a formal total quality managementprocess, but by 1992 over 90 percent of the surviv-

ing 56 firms1 had adopted formal TQM practices.Figure 3 presents the diffusion curve (Nohria &Green, 1996). In this short period, TQM practicescoalesced into a major revolution in organizationalpractices that swept across the entire U.S. manage-ment landscape (Winter, 1994).

On the symbolic side, stories and accounts thatgave meaning and legitimacy to TQM practices ac-companied their diffusion. Initial TQM accountspersuasively described these practices as appropri-ate and effective for solving important organization-al problems (Zbaracki, 1998). Specifically, the rhet-oric of TQM accounts created new logics thatredefined the established institutional logic andmeaning of quality, costs, customers, managementrelations, and organizational improvement (Cole,1999; Hackman & Wageman, 1995). Over time,TQM accounts moved from initially making senseof TQM practices to describing the success result-ing from their adoption and use (Giroux & Taylor,2002; Zbaracki, 1998).

As TQM institutionalized, its symbolic and ma-terial components coevolved. For example, theearly adoption of the material component of TQMwas primarily attributed to the severe competitivepressure U.S. manufacturing companies faced fromforeign competition, especially Japanese firms withTQM processes (Easton, 1993). This was ironic con-sidering that many TQM techniques (e.g. statisticalprocess control) were originally developed in theUnited States. The material success of high-quality,

1 Forty-three firms disappeared from the Fortune 100from 1977 to 1992 as result of mergers, successful take-overs, bankruptcies, and reorganizations.

FIGURE 3Percentage of Fortune 100 Firms Adopting TQM

0102030405060708090

100%

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

Percentage

Year

% Adopting

2009 17Green, Li, and Nohria

competitively priced Japanese products in the U.S.market prompted executives in pioneering firmssuch as Motorola, Ford, and Xerox to examine whatJapanese firms were doing that could explain theirsuccess (Nohria, 1993). In response, the media,consultants, and many academics created socialaccounts or arguments that attributed Japanese suc-cess to a superior approach to management. Theseaccounts singled out TQM as the management prac-tice that was at the heart of the Japanese success.Early adopters were described as firms that hadbeen affected by Japanese competition and thusadopted TQM as a way to catch up with their Jap-anese counterparts.

In addition, later diffusion was attributed to theregulatory push by the U.S. Department of Defense,which mandated that the suppliers of governmentcontracts adopt TQM practices. The U.S. Congressparticipated in this regulatory push as well by cre-ating the Malcolm Baldridge National Quality Im-provement Act of 1987 to recognize quality excel-lence and to stimulate quality improvement inAmerican industry (Allaire, 1991; Garvin, 1991).Over time, people came to see TQM as a legitimateway for U.S. firms to regain competitiveness, andfavorable press reports about some early U.S. usersof TQM practices reinforced this perception(Lawler & Mohrman, 1985).

Measuring Institutionalization

Although the diffusion and persistence of prac-tices like TQM are often used as evidence of insti-tutionalization, how do people know TQM prac-tices have actually acquired legitimacy? Forexample, examining Figure 3 reveals that 20 per-cent of Fortune 100 firms had adopted TQM by1982, 50 percent by 1986, and 80 percent by 1989.Yet how does one know that TQM has more legit-imacy at an 80 percent adoption level than at a 20percent adoption level, or even that it has any le-gitimacy at all? To answer this question, we pro-pose using a rhetorical model of institutionaliza-tion to help decouple the diffusion of TQMpractices from their institutionalization.

Data Collection

To develop a detailed, robust understanding ofTQM practices and rhetoric, we conducted a de-tailed literature review of both scholarly and prac-titioner journals. We supplemented this researchwith an extensive set of retrospective interviewswith managers of firms in the Fortune 100. Ourinitial interview sample included all firms on theFortune 100 list compiled from Fortune magazine’s

annual rankings of the largest U.S. industrial com-panies throughout the period 1975–92. Our samplestarted with 100 firms in 1975 and, by 1992, 57 ofthem were still in the Fortune 100 (see footnote 1).One of these refused to participate, leaving us witha final sample of 56 companies. The earliestadopter of TQM in this group adopted it in 1978and by 1992, 53 of the 56 firms had adopted TQM(see Figure 3).

Each of the 56 firms was contacted to set up atelephone interview with the appropriate managerfamiliar with the development of the firm’s qualitymanagement practices. This was generally a vicepresident or director of quality. Usually, we con-tacted four to five managers in each company inorder to find the appropriate managers with knowl-edge of the firm’s TQM practices and TQM adop-tion process. Contacted firms agreed to provide therelevant information through phone interviews,and each interview lasted from 30 to 90 minutes.Ninety-three interviews were conducted with exec-utives at the 56 firms. We also conducted an addi-tional 204 interviews at other Fortune 100 compa-nies (e.g., firms that came into the index after 1975),suppliers and customers of these firms, a smallnumber of firms from the larger Fortune 500, aswell as selected academics and consultants work-ing in the TQM area. Although these additionalinterviews helped inform our overall understand-ing of the TQM movement, the interview dataused in this paper are primarily from the 56 For-tune 100 firms.

Some of the questions we asked that are relevantto this research project were the following: “Didyou adopt TQM, and if so, when and where in yourorganization?” “What is the definition of TQM atyour firm?” “What were the reasons for your firmadopting TQM?” “Why is TQM important to yourfirm?” The transcripts from these interviews pro-vided detailed knowledge about the definition, jus-tification, and discourse surrounding TQM adop-tion. Taken together, our literature review andextensive interviews with TQM managers shapedand influenced how we conceptualized TQM rhet-orics and logics.

To collect TQM discourse, we searched elec-tronic articles related to TQM in selected journals,magazines, and newspapers using Lexis-Nexis. Thesearch string consisted of “tqm or quality manage-ment or q circles or quality circles or tqc or totalquality or quality control.” This search string isessentially a list of synonyms for quality practiceswithin firms and was developed from both ourliterature review and interviews. The publica-tions we searched included the Wall Street Journal,BusinessWeek, the Economist, Industry Week, the

18 FebruaryAcademy of Management Journal

Harvard Business Review, U.S. News & World Re-port, the Financial Times, the New York Times,Fortune, and Forbes. We chose these publicationsbecause interviewees mentioned them as newssources they read regularly. The articles we ob-tained ranged from 1975 to 1995, a time period weselected for several reasons. First, several scholarshave argued that TQM began its spread in Americain the late 1970s and began to deinstitutionalize bythe mid 1990s (Benson, 1993; Cole, 1999; Giroux &Taylor, 2002; Nohria & Green, 1996; Powell, 1995).This date range also resonates with our interviewsample, in which the first firm adopted TQM in1978 and the last did so in 1991. Because our focuswas on the institutionalization of TQM, we chose tostart at 1975 and end at 1995 to capture the entireinstitutionalization of TQM. The population of ourstudy consisted of 6,074 articles, and these articlescontained 49,848 paragraphs. In Figure 4, the leftY-axis represents the number of articles for eachyear in our database, and the right Y-axis representsthe number of paragraphs.

To focus our attention on an argument essentialto the institutionalization of TQM, we used thesoftware program ISYS2 to randomly sample 630paragraphs. Specifically, we sampled 30 para-graphs for each year from 1975 to 1995 that con-tained at least two of the concepts or synonyms ofquality, costs, and waste. We sampled 30 para-graphs per year for two reasons. First, after careful

examination of several initial rounds of samples,we found that 30 paragraphs were enough to cap-ture the patterns in the discourse. Second, we lim-ited the sample size to 30 to balance the need for aclose and detailed examination of text by indepen-dent coders with the need for a sample largeenough to allow us to make statistical statementsabout the discursive patterns observed.

We chose the concepts and synonyms of quality,costs, and waste for several reasons. First, althoughthe TQM literature theorizes several important log-ics and assumptions underlying TQM practices(Giroux et al., 2002; Hackman et al., 1995), oneimportant logic is the notion that quality can beimproved without increasing costs because TQMpractices reduce waste (Cole, 1999, 2000; Deming,1993; Easton & Jarrell, 2000; Hackman & Wageman,1995). The key concepts for identifying this TQMargument are the concepts of waste, quality, andcosts.

Second, a detailed analysis of our notes from theinterviews with Fortune 100 executives and man-agers responsible for TQM programs suggested thatthe quality, cost, and waste argument was the mostimportant and consistent reason for adopting TQMthroughout the diffusion process. Specifically,when asked why TQM was adopted, managers gavemany different reasons, ranging from Japanesecompetition (stated by 33 percent of interviewees)to customer demands (stated by 46 percent) to em-ployee empowerment/participation (stated by 6percent). However, the most popular reason thesemanagers gave for adopting TQM (stated by 69percent of interviewees) was the desire to lower

2 ISYS is a software program that allows a user toidentify paragraphs within a text that contain particularconcepts and synonyms designated by the user.

FIGURE 4Counts of Articles and Paragraphs about TQM, 1975–95

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0

100

200

300

400

500

600

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

Numberof

Paragraphs

Numberof

Articles

Year Articles

Paragraphs

2009 19Green, Li, and Nohria

cost and improve quality in a highly competitiveenvironment.

Third, the quality, waste, and cost argument cap-tures one of the core insights of TQM. Specifically,TQM emphasizes the flexibility of the productionprocess and the concept that managers, throughthe continuous improvement of production pro-cesses, can increase quality without increasingwaste. This idea allows managers to participate inand enact the “quality revolution,” and it is suc-cinctly captured and expressed in the quality,waste, and cost argument. We believe this central-ity makes this argument one of the important justi-fications for TQM adoption, if not the most impor-tant justification.

Fourth, our empirical study discusses the pro-cess of institutionalization for for-profit firms, andthus we assume that a dominant and central logicfor this organizational field is the pursuit of profit—selling more products at a lower cost. The rela-tionships among quality, waste, and costs reso-nate with this central logic because improvements inquality should increase the sales of a product, andthe reduction of waste should lower the cost of thatproduct.

Although the previous four reasons suggest thatthe quality, cost, waste argument is the most im-portant argument used to justify and legitimateTQM practices, the validity of our theoretical claimdoes not depend upon this fact. Specifically, ourtheoretical claim suggests that the need to discur-sively justify the adoption of a material practicedecreases as that practice institutionalizes. Our the-oretical claim does not require that the reasons orarguments used to support that practice be the mostimportant or central, just that they be importantenough to persuade actors that the practice is wor-thy of adoption and use. There may be variousreasons for adopting a practice, and our theorywould suggest that these other argument structuresshould collapse too as that practice becomes takenfor granted and institutionalized. This view is con-sistent with the observation that many of the insti-tutionalized practices that exist in society are sotaken-for-granted that they are ostensibly withoutdiscursive justification (Jepperson, 1991). One canimagine that at the creation of these institutionsvarious arguments supported their existence andpersistence, yet over time these arguments col-lapsed as the practices became taken-for-grantedwithin society. Similarly, many different argu-ments used to justify TQM collapsed and becamesimplified as TQM institutionalized. In sum, for thesake of brevity and clarity we focused on the inter-nal structure of one important argument used to

justify and support the institutionalization of TQMpractices.

Once we decided to focus on the quality, cost,and waste argument, a doctoral student and anundergraduate student in business read and codedall of the collected paragraphs. We provided thecoders with background knowledge about TQMand a list of the synonyms for the three key con-cepts. Examples of the synonyms for quality were“tqm,” “tqc,” “total quality,” and “quality manage-ment.” Examples of synonyms for costs were “ex-pense,” “spending,” “outlay,” and “margin.” Exam-ples of synonyms for waste were “redo,” “defect,”and “rework.”

Our coding instructions defined the primary unitof analysis as a proposition. The instructions spec-ified five categories to which a proposition couldbe assigned for coding: (1) a major premise 1 statesa positive relationship between quality and costs,(2) a major premise 2 states a positive relationshipbetween costs and waste, (3) a minor premise statesa negative relationship between quality and waste,(4) a claim states a negative relationship betweenquality and costs, and (5) a nonproposition/irrele-vant proposition is a statement that is not a propo-sition or is irrelevant. Because the data were nom-inal, we calculated intercoder reliability usingCohen’s (1960) kappa. The intercoder reliability forthe coding of the content of the argument was .86.The final sample consisted of 346 paragraphs thatcontained at least one of the four propositions. Theparagraphs that were coded as irrelevant weredropped from the sample. An example of an irrel-evant paragraph is the following, which discussesthe effect of lighter fluid on air quality standards incity of Los Angeles:3

The lone board member who opposed the plan,Mike Antonovich, called it a waste of time and saidit placed an unfair burden on store owners. The ruledoes not set any penalties for using the lighter fluid,just for selling it, he said. Mr. Antonovich also saidbackyard chefs could also get around the plan bybuying lighter fluid at stores outside the air qualitydistrict. (1990)

This paragraph has nothing to do with quality prac-tices in the American business community or therelationship of quality, cost, and waste, and thus isirrelevant to our analysis.

3 The year in parentheses after a quotation from apublication indicates the year that the quote appears inour publication database. Similarly, the year in parenthe-ses after an interview quotation indicates the year thatthe interviewee’s firm adopted TQM.

20 FebruaryAcademy of Management Journal

Our analysis of TQM rhetoric consists of fourparts. First, we describe the institutional entrepre-neurs of TQM: the agents and creators of the TQMdiscourse that helped to rationalize and institution-alize TQM adoption. Second, we offer a qualitativeanalysis of TQM discourse. Our qualitative analysispresents, analyzes, and interprets several represen-tative paragraphs in our TQM discourse database.We show that, as TQM diffused, its symbolic com-ponent changed to reflect the institutionalization ofTQM practices. Third, we present a visual analysis,in which we graphically display the entire sampleof paragraphs using the aggregate data-coding re-sults to describe how structural changes in TQMarguments reflect changes in the institutionaliza-tion of TQM. After outlining our qualitative andvisual analyses of the data, we describe two quan-titative analyses conducted to pin down the signif-icance of the structural changes.

DATA ANALYSIS

TQM Entrepreneurs: The Creators ofTQM Discourse

Who are the institutional entrepreneurs or cre-ators and propagators of TQM discourse? The ini-tial institutional entrepreneurs were probably theTQM “gurus,” who published their most famousbooks on quality in the late 1970s to late 1980s (e.g.,Crosby, 1979; Deming, 1982; Feigenbaum, 1983;Ishikawa, 1985; Juran, 1989). Scholars also mentionconsultants, CEOs, government officials, and aca-demics as playing important and active roles in thespread and institutionalization of TQM (Abraham-son & Fairchild, 1999; Cole & Scott, 2000; Lawler,Mohrman, & Ledford, 1992; Zbaracki, 1998). Simi-larly, our article and interview data also show theseTQM entrepreneurs arguing actively for the use ofTQM practices. For example, the following quotedescribes the TQM guru Dr. Deming extolling thevirtues of quality practices to car executives at aconference:

It was only a year ago that Dr. W. Edwards Deming,a harsh critic of U.S. management, crisply presentedhis statistical theories for reviving U.S. manufac-turing productivity to a nationwide audience via anNBC-TV news special. Then the 80-year-old-plusconsultant strolled from center stage. Recently re-tained by Ford Motor Co. as a consultant, he deliv-ered his blunt remarks to hundreds of auto andauto industry-related officials at the annual Auto-motive News World Congress, sponsored by the De-troit-based industry trade paper. (1981)

Another article provides even greater detail onTQM entrepreneurs and rhetors promoting and ad-vocating the use and acceptance of TQM practices:

America’s drive for quality in almost every kind ofenterprise is generating the fervor of a crusade. Al-though the most visible part of the push has been inthe auto industry, everybody’s joining in. “Every-body” includes state and local governments worriedabout their industries becoming noncompetitive,chambers of commerce, and community collegesand universities. . . . Other firms like Dr. J. M. Ju-ran’s Juran Institute in Wilton, Conn., Dr. ArmandV. Fiegenbaum’s General Systems Co. in Pittsfield,Mass., and Philip Crosby’s Quality College in Win-ter Park, Fla., are doing a land-office business inpreaching quality. The Quality Expo shows andseminars in Chicago and Los Angeles sponsored byQuality magazine are experiencing 20% annual in-creases in attendance and exhibitors. . . . In Cam-bridge, Mass., Dr. Myron Tribus, Nobel prize win-ner and director of the Center for AdvancedEngineering Study at the Massachusetts Institute ofTechnology, has organized an American QualityCamp, the Productivity Institute, encompassing ex-isting quality-control professionals and academics.Dr. Tribus has helped sparkplug a half-dozen re-gional efforts across the country and hopes to de-velop an umbrella organization comparable to the40-year-old Japanese Union of Scientists and Engi-neers, which supports all types of quality effortsthere. (1985)

These excerpts from our TQM article databaseare just a few of the countless descriptions of TQMentrepreneurs actively creating the discourse andarguments that rationalize and make sense of TQMpractices. The above excerpts also capture vividlythe structuring activities of institutional entrepre-neurship. For example, the development of indus-try conferences and exhibitions, the establishmentof educational and research institutes, and the cre-ation of nationwide and regional professional asso-ciations and industrial consortia all contributed tothe creation of an infrastructure and network thatfacilitated the dissemination of discourse andarguments.

In addition, our interviews also provide supportfor the notion of engaged TQM entrepreneurs andrhetors actively promoting TQM practices. For ex-ample, one interviewee told the following storyabout how TQM was brought into a company:

The president of the company came up throughmanufacturing. One day he put the Crosby book“Quality is Free” on the VP of Manufacturing’sdesk. The VP of manufacturing had an engineeringbackground. This was very important because man-agers that come up through accounting or engineer-ing are used to managing things and not people. TheVP of manufacturing read the book. He didn’t un-derstand some parts, however he told the seniorlevel people in the company to read “Quality isFree” and do something. In a sense, this was a

2009 21Green, Li, and Nohria

blessing for the company because each section of thecompany custom-tailored their TQM programs to fittheir individual needs. The president told me thisstory on the plane one day. But if you ask anyoneelse, they would say that the VP was the father ofquality at [Company Name]. (1981)

This interviewee’s account describes three differ-ent TQM entrepreneurs. The first was Mr. Crosby,who wrote an extremely popular book extolling thevirtues of TQM. Another TQM entrepreneur, thepresident of a company, took the Crosby book andhanded it to the vice president of manufacturingwith the implied directive to read it. The presidentalso continued to talk about the importance of qual-ity at the firm by sharing stories about its spread toexecutives (including the interviewee). In addition,the VP of manufacturing at the firm in the story alsohas a role as a TQM entrepreneur. The VP read theCrosby book, attempted to understand as much ashe could, and then handed it off to his subordinatesto read and enact within the firm. We provide fur-ther examples from our interviews of TQM entre-preneurs actively advocating for and promoting theadoption of TQM in Appendix A.

In sum, the institutionalization of TQM practiceswas not produced by a disembodied discourse. Ra-tionalizing and making sense of TQM practices re-quires TQM discourse, but the creation and use ofsuch a discourse requires agency. It requires actorsmaking arguments that define situations, interpretproblems, and prescribe solutions to get things ac-complished. In the next section, we analyze thisdiscourse.

Qualitative Analysis

We analyze representative TQM arguments se-lected from the TQM discourse database describedin the methodology section. We hope this analysisprovides a qualitative description of TQM institu-tionalization—that is, how TQM became rational,sensible, and appropriate. We show distinctchanges in business discourse over time as TQMpractices diffused within the business communityfrom 1975 to 1995.

Old endoxa, or major premise 1. A close readingof TQM discourse from the mid and late 1970sreveals a great deal about the institutionalized be-liefs and assumptions of business managers duringthis period. One particular set of assumptions con-cerns the relationship of product quality with prod-uct costs. Consider the following two examples:

We are more a quality magnet than anyone else. Tokeep our reputation, we have always been very par-ticular about quality—the quality of raw materials,

the control of manufacture. Our costs are higherthan our competitors’ because if you want goodquality, you have to pay for it; it is expensive.(1976).

“There’s a notion,” says Feigenbaum of GeneralSystems, “that something about quality means ahigher-quality product must cost more.” (1979)

These two quotes provide insight into the endoxaor taken-for-granted assumption of the relationshipof product quality and product costs within themanagerial community. As other scholars havecommented (Cameron & Barnett, 2000; Cole, 2000;Easton & Jarrell, 2000), most managers believed inthe idea of a trade-off between quality and costduring the early stages of TQM diffusion. Colenoted that one of the basic assumptions of Ameri-can managers in the early 1980s was, “It costs moreto provide a quality product or service and we willnot recover the added cost.” This assumption wasso institutionalized that “there was literally a cog-nitive gap that did not allow U.S. managers to con-ceive of the possibility of a negative relationship ofquality to costs” (Cole, 2000: 74). Similarly, Cam-eron and Barnett wrote, “The correlation betweencost and quality is positive (i.e., the higher thequality, the higher the cost); hence, it is an expen-sive way to achieve high quality” (2000: 275). Werestate this institutionalized or “endoxic” relation-ship in syllogistic form:

Old claim, or major premise 1. Quality andcosts are positively related.

Major premise 1 is in complete opposition to thefinal claims of TQM. Specifically, TQM entrepre-neurs proposed the adoption of TQM practices be-cause these practices improved quality while low-ering a firm’s costs (Deming, 1993; Hackman &Wageman, 1995). For American managers who be-lieved quality and costs were positively related, theclaim that TQM could increase quality and lowercost was like claiming that black is white (Cole,2000: 72). However, as Cole (2000) observed, “Bythe mid- and late 1980s, we start to see many Amer-ican managers begin to accept publicly the coun-terintuitive claims that high quality and low cost gotogether” (2000: 72).

By the early 1990s, TQM and the underlyingprinciple of a negative relationship between costand quality were thoroughly ensconced and insti-tutionalized throughout the American businesscommunity. This raises the question of how theconception of costs and quality as positively re-lated deinstitutionalized and the conception ofcosts and quality as negatively related institution-alized. We propose that changes in the institution-

22 FebruaryAcademy of Management Journal

alization of concepts (e.g., TQM) and their corre-sponding behaviors are reflected in theargumentative structures supporting these con-cepts and behaviors. Moreover, examination ofchanges in the rhetorical structure of TQM argu-ments demonstrates how institutions simulta-neously constrain actors and enable them to createnew institutions.

For example, at any given time the endoxa of aninstitutional environment consists of many poten-tially conflicting premises. Specifically, in the1970s managers accepted major premise 1 (i.e.,quality and costs are positively related). Many ofthese same managers also believed major premise2—that waste and cost are positively related. Thefollowing passage is an example of majorpremise 2:

According to Feigenbaum, “There are significantnumbers of otherwise well-organized plants where[correcting mistakes] represents anywhere from15% to 40% of the plant’s productive capacity. It’swhat we call ‘hidden plant’—the plant that’s thereto repair the defects, retest the product, redo thescrap. For that kind of plant,” he says, “there is nolower-cost way to improve productivity and reducemanufacturing costs” than to institute a strong er-ror-prevention system that will eliminate the needfor some of the hidden plant. (1979)

This quotation suggests that managers believed re-ducing waste and rework in the production processwas an effective way to decrease production costs.This belief was so taken-for-granted that the words“cost” and “waste” were virtually used inter-changeably (Easton & Jarrell, 2000: 111). We restatethis endoxic relationship in syllogistic form:

Major premise 2. Waste and costs are positivelyrelated.

Rhetorically savvy entrepreneurs at the beginningof the TQM movement used major premise 2 tobuild an argument that attacked and began deinsti-tutionalizing major premise 1. Specifically, TQMentrepreneurs constructed a minor premise thatconnected persuasively with major premise 2. Forinstance, the following is a typical statement ofTQM entrepreneurs in the early 1980s:

Dr. Deming said the struggle to improve the qualityof American goods hinges upon the reduction ofwaste, the key weapon being a training system inwhich managers and production workers alike areinstructed in statistical methodology. This, he said,was the foundation of his highly successful workwith Japanese industry in the 1950s. “Japan lis-tened,” Dr. Deming remarked, “and it worked.”(1981)

As demonstrated in this quote, managers believedthat the reduction of waste was an important factorin the improvement of quality. Management gurus,especially Deming, Juran, and Crosby, exerted greatinfluence in the construction of this minor premise(Easton & Jarrell, 2000). Although they differed asto whether quality was a tactical, strategic, or phil-osophical issue (Cole, 2000), these TQM gurusbrought managers’ attention to the seriously ad-verse effect of waste and its relationship to quality(Cole, 2000; Easton & Jarrell, 2000).

For example, Crosby defined quality in relationto defects and set the performance standard forquality as “zero defects” (Crosby, 1979). In additionto the zero-defects approach, Deming and Juranemphasized other aspects of eliminating waste,such as cycle-time reduction, storage, benchmark-ing, and just-in-time delivery (Easton & Jarrell,2000). Put simply, waste came to be seen as themajor barrier to high quality. In response, managersformalized and standardized their managementsystems to reduce waste.

Central to the idea of improving quality throughreducing waste is a new conception of quality anda new understanding of the production process asflexible as opposed to fixed. Specifically, quality isno longer seen as a narrow area of functional spe-cialties, obtained only by inspection, detection,and subsequent repair. Instead, quality becomes amajor competitive strategy and is viewed as inte-grative of every aspect of organizational processes.TQM emphasizes that quality is “built in” during aproduction process and can be improved throughcontinuous improvement of the process. Throughconstant examination of technical and administra-tive processes, problems are solved at earlier stagesand defects are prevented, yielding savings forfirms (Anderson, Rungtusanatham, & Schroeder,1994; Cole, 1999: 25–33; Dean & Bowen, 1994). Werestate this relationship as the minor premise for anargument:

Minor premise. Quality and waste are nega-tively related.

We conceptualize the negative quality and wasterelationship as the minor premise, because (com-pared with the major premise), this relationshipwas less taken-for-granted by the American busi-ness community. The minor premise was less tak-en-for-granted because it emphasized a new anddifferent aspect of quality: process quality as op-posed to product quality. Prior to the TQM revolu-tion, American managers emphasized productquality and took production processes as given. Ifthe production process, or process of achievingquality, is taken as given, increases in product qual-

2009 23Green, Li, and Nohria

ity inevitably incur more waste. However, empha-sizing the malleability of production processes al-lows for increases in product quality withoutincreases in waste. At the beginning of TQM insti-tutionalization, this was a relatively new and radi-cal idea in the American business community.Only when TQM proponents like Deming startedpersuading managers that production processeswere not fixed and that they could be improvedcontinuously could people make sense of the neg-ative relationship between quality and waste.

By creating and linking major premise 2 to theminor premise, TQM rhetors created and supported acontroversial new claim—quality and costs are neg-atively related—that contradicted major premise 1(quality and costs are positively related).

The Claim as the Product of Agents

Major premise 2 states a positive relation ofwaste to costs. The minor premise proposes a neg-ative relation of quality to waste. Together thesetwo premises create a valid inference that claims anegative relationship between quality and costs.Consider the following passage:

The Japanese regard waste, whether of material orhuman resources, as an abomination. Emphasis onproduct quality reduces waste: less scrappage,fewer rejects, fewer returns. And happily, the Jap-anese have found that high quality and low costsare not incompatible goals that can be resolvedonly through trade-offs—generally the view in theU.S.—but are linked and complementary objectives.(1982)

This quotation suggests that increases in qualitycan reduce costs, a view consistent with the ap-proach followed by corporate Japan: “By eliminat-ing waste and rework in business processes, onecould achieve both higher quality and lower costs(Cole, 2000: 75).” We restate this relationship insyllogistic form:

Claim. Quality and costs are negatively related.

It is important to consider that the claim repre-sents an essential argumentative support for totalquality management (Deming, 1986; Hackman &Wageman, 1995; Ishikawa, 1985; Juran, 1989). Thisclaim also is the antithesis of major premise 1,stating that quality and costs are positively related.Changes in the structure of TQM arguments—themove from major premise 1 in the early 1970s to thenew claim of the 1980s and 1990s—reflect changesin the institutional environment in general and theTQM institution in particular. Traditionally, costwas thought to be an unavoidable attribute ofhigher quality. Redefined as waste, however, costs

came to be regarded as the root cause of bad qual-ity, and thus eliminating costs became the path tohigher quality.

The new claim actively redefined conventionalbeliefs about costs. For example, within the contextof supporting a controversial new claim (the notionthat quality and costs are negatively related), earlyadvocates of TQM used the minor premise (qualityand waste are negatively related) to actively rede-fine conventional beliefs about costs. This redefi-nition stated that “the costs of poor quality (such asinspection, rework, lost customers, and so on) arefar greater than the costs of developing processesthat produce high-quality products and services”(Hackman et al., 1995: 310). Moreover, this redefi-nition of the meaning of costs is counter to thecommonly held and taken-for-granted understand-ing of costs. By combining major premise 2 with theminor premise, a rhetor actively directed attentionaway from the commonly held beliefs of the man-agement community (e.g., major premise 1), to anew conception (e.g., the claim) that had the poten-tial to change and reshape the then-current endoxicenvironment. In a sense, the initial combining andstating of major premise 2 with the minor premisewas inherently creative and represented an impor-tant aspect of agency, because it invented new pos-sibilities for thought and action (Emirbayer &Mische, 1998).

Within this institutional framework, the poten-tial for agency lies in the ability to detect and ex-ploit contradictions within endoxa among pre-mises. In the case of TQM, savvy rhetors utilizedmajor premise 2 (waste and costs are positivelyrelated) to exploit the potential contradiction be-tween major premise 1 (i.e., quality and costs arepositively related) and the minor premise (qualityand waste are negatively related). By persuasivelypresenting the minor premise and major premise 2,a counterintuitive yet valid claim (quality and costsare negatively related) was inferred. Please refer toAppendix B for additional examples of majorpremise 1 and 2, the minor premise, and the claim.

Where the potential to see and exploit contradic-tions between major premises enables agency, theinability to see these contradictions underlies thepower of structures to constrain and conform. Totake a premise for granted, shielding it from scru-tiny and elaboration, is precisely how institutionsshape and influence perceptions, reason, and ulti-mately action (Douglas, 1986).

The Change in Argument Structure

Modeling changes in argument structure helps usmeasure the process of institutionalization as well

24 FebruaryAcademy of Management Journal

as identify periods of constraint and agency. Spe-cifically, we hypothesize that in the early phase ofinstitutionalization, a complete syllogism is statedto inform the audience how the claim is inferredlogically from the major premise and the minorpremise. As the level of institutionalization in-creases, the syllogism collapses into an enthymemein which the major premise is no longer stated.Figure 5 shows an example of this process usingTQM arguments.

At the enthymematic stage, the audience takesthe missing major premise for granted, thus statingthe minor premise alone is sufficient to support theclaim. The collapse of syllogisms into enthymemesindicates that the new claim has been institution-alized to a certain degree. Thus, at this midwaystage of institutionalization we would expect to seea decline in the major premise of a given argumentrelative to the minor premise and the claim associ-ated with that argument. Failing to state the majorpremise also indicates that a rhetor is letting theunstated premise guide beliefs and actions. Thisomission reflects the social constraint of endoxa:the reasoning process of institutional actors isshaped by the commonly held premises of therhetor and audience.

At an even later stage of institutionalization, theminor premise of an argument is also suppressed,leaving only the claim. The stating of the claimwithout the major or minor premise represents ahigher state of institutionalization, reflecting broadand generalized acceptance by the audience. Thus,at this later stage of institutionalization we wouldexpect to see a decline in the minor premise rela-tive to the claim. The following sentence is anexample of a claim without the major premise orminor premise:

Exploration of the quality terrain has caused whatwas once a revelation to become common knowl-

edge: The very pursuit of quality products, services,and processes results in lower cost, higher produc-tivity, and greater market share. (1991)

In the early stages of the TQM movement, thisstatement would have been nonsensical to Ameri-can managers, because it contradicted the endoxicor institutional environment in which quality andcosts were positively related. However, by the laterstages of TQM institutionalization, this statementwithout a supporting major or minor premise hadbecome a general belief. In fact, the belief is taken-for-granted to such an extent that it could be usedas a major premise for constructing new syllogisms.

Visual-Graphical Analysis

In this section, we visually examine the entiresample of coded TQM arguments from the TQMdiscourse database. We display the results of thecoding graphically to support and extend our qual-itative analysis of TQM arguments. The result ofour coding supports our conceptualization of insti-tutionalization as changes in argument structure.Figure 6 shows the number of propositions fromour sample for each of the four types of proposi-tions discussed above over the time period 1975 to1995.

Major premise 1, stating the positive relationshipbetween quality and costs, appears in the mid tolate 1970s and disappears in the early 1980s. Majorpremise 2, stating the positive relationship betweencosts and waste, also appears in the late 1970s andthroughout the 1980s, but its appearance begins todecline by the mid 1980s and almost disappears bythe 1990s. The minor premise, or the negative re-lationship between quality and waste, emerges inthe late 1970s, maintains its presence during the1980s, rises and falls throughout the 1980s andearly 1990s, and declines at a slower rate than

FIGURE 5TQM Argument Structure and Institutionalization

2009 25Green, Li, and Nohria

major premise 2. Finally, the claim—the negativerelationship between quality and costs—rises dra-matically into the 1980s, moderates in the 1980s,and rises into 1990s. Most importantly, the declineof major premise 2 and the minor premise relativeto the claim supports our argument of an increasein the taken-for-grantedness of TQM. The declineof major premise 1 together with the rise of theclaim also shows evidence of the malleability ofinstitutional logics. Specifically, the logic of thequality and cost relationship essentially invertedfrom quality as positively related to costs to qualityas negatively related to costs.

To further test our predictions concerning thetrends and relationships among these four proposi-tions, we divided the number of each type of TQMproposition for each year by all of the TQM prop-ositions for that year. This procedure allowed us tocalculate a ratio that captures the proportion ofeach particular type of TQM proposition relative toall four potential types of TQM propositions forthat year. Figure 7 shows this ratio.

In Figure 8, we graphed the ratio of propositiontypes using a moving average calculated over threeyears to smooth out the curves. Both Figures 7 and8 show the trends that we expected.

Placing the trends of these four TQM propositiontypes in graphical form helps us to identify changesin the argument structure of TQM over time. Spe-

cifically, Figures 7 and 8 show that from the mid tothe late 1970s, major premise 2, the minor premise,and the claim have approximately the same level ofoccurrence, which indicates that actors are employ-ing syllogisms to argue for a negative relationshipbetween quality and costs. In fact, during the pe-riod of the mid to late 1970s, major premise 1(quality and costs are negatively related) is men-tioned with regularity in managerial discourse. Be-ginning in the early 1980s, the proportion of majormentions of premise 2 begins to decline, decreasingsignificantly after 1988. Although major premise 2declines significantly during the mid and late1980s, the minor premise and claim remain at theirsame levels or increase during this period. Thisobservation reflects the enthymeme replacing thesyllogism as the primary argumentative support forthe claim. In 1987, the proportion of the minorpremise begins to decline dramatically in relationto the claim, reflecting the collapse of the en-thymeme as a needed support for the claim, leavingthe claim alone as the argument. In sum, Figures 7and 8 provide strong visual-graphical evidence thatthe claim increased relative to both the minorpremise and major premise 2 from 1975 to 1995.The graphs also show that major premise 2 declinesfaster than the minor premise. This difference inrate supports our theory that institutionalization isreflected in changes in argument structure from

FIGURE 6The Number and Type of Each TQM Proposition by Year

26 FebruaryAcademy of Management Journal

syllogism to enthymeme and in changes to theclaim over time. Moreover, Figures 7 and 8 alsoshow the decline or deinstitutionalization of majorpremise 1 (quality and costs are positively related)and the rise or institutionalization of the claim(quality and costs are negatively related).

Statistical Analysis

To further support our qualitative and visual-graphical analyses, we also conducted a quantita-tive analysis with two parts. First, we regressed theproportions of the four proposition types on time toshow how each type changed over time. Second,we conducted an ordinal logistic regression to in-vestigate how the three propositions used to con-struct the TQM argument (major premise 2, minorpremise, and claim) moved together as a group.

Table 1a presents the descriptive statistics fromregression of the proportions of the four proposi-tion types on time using the data from our visual-graphical analysis. Table 1b presents the regressioncoefficients, which confirm the trends shown in thevisual-graphical presentation. Specifically, the pro-portions of major premise 1, major premise 2, andthe minor premise decrease significantly over time,while the proportion of the claim increases signif-icantly over time. In addition, the coefficient for the

proportion of major premise 2 is more negativethan that for the proportion of the minor premise,indicating that the major premise of the TQM argu-ment disappeared faster than the minor premise.This finding provides statistical support for ourtheoretical claim that the major premise gets sup-pressed first, and then the minor premise issuppressed.

Second, we ran ordinal logistic regressions toinvestigate how the TQM propositions move in dis-tinct groups over time. Using our original sample of346 coded paragraphs that contained at least one ofthe four propositions, we organized the paragraphsinto distinct groups on the basis of which proposi-tions were present or absent in each paragraph. Weused these distinct groupings to construct our de-pendent variable. Specifically, when a paragraphhad only the old claim present, we coded it 0; aparagraph with a major premise was coded 1; aparagraph with a minor premise but no majorpremise was coded 2; and a paragraph with neithera major premise nor a minor premise, but only aclaim, was coded 3.

Our intercoder reliability estimates were thesame as for the visual analysis. Specifically, thecoding of these four paragraph types did not re-quire additional interpretation of the text, just aregrouping of the previously coded data used in the

FIGURE 7The Changing Structure of TQM Arguments, 1975–95

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

90.00

100.00%

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

Ratio ofPropositions

Year

ClaimMinor premiseMajor premise 2Major premise 1

2009 27Green, Li, and Nohria

visual analysis. Each of these four paragraph typesreflects a different type of TQM argument. Category0 is the old and institutionalized claim, whichstates that quality and costs are positively related.Arguments coded as category 1 state the majorpremise and thus are less taken-for-granted thanarguments coded as category 2. The latter state theminor premise without the major premise and thusare less taken-for-granted than arguments coded ascategory 3. Arguments coded as category 3 containonly the new claim and no major premise or minorpremise. In keeping with our theoretical claims, inwhich sequential suppression of the major premiseand the minor premise are assumed in making anargument, we assume that categories 1, 2, and 3 of

our dependent variable are ordered, with higher-value categories indicating higher degrees of taken-for-grantedness. We also assume that category 0, orthe proposition that quality and costs are positivelyrelated, represents the initial endoxic environmentand thus precedes the development of the TQMargument, or categories 1, 2, and 3. The frequencyand percentage of the categories are as follows: 8(2.31%) for category 0, 44 (12.72%) for category 1,82 (23.7%) for category 2, and 212 (61.27%) forcategory 3. The mean and standard deviation forthe dependent variable are 2.44 and 0.80, respec-tively. We constructed a measurement of time asthe predictor variable, with the year 1975 corre-sponding to 1, 1976 to 2, and so forth. The meanand standard deviation for the time variable are11.69 and 5.95, respectively. The correlation be-tween category type and time is 0.35.

Given the ordinal nature of our dependent vari-able, we employed ordinal logistic regression anal-ysis, which allowed us to compare the odds of anargument being in lower or equal categories versushigher categories. Ordinal logistic models were ap-propriate for testing our theoretical argument be-cause they are concerned with the direction of re-sponse and take into account the rank-ordering ofthe outcomes. We used STATA 9.0 to estimate theordered logit model; model 1 in Table 2 shows

TABLE 1aDescriptive Statistics and Correlation Coefficients

for Proportions of Propositionsa

Variable Mean s.d. Time

1. Proportion major premise 1 0.03 0.05 �0.602. Proportion major premise 2 0.12 0.10 �0.783. Proportion minor premise 0.26 0.10 �0.434. Proportion claim 0.60 0.18 0.875. Time 11.00 6.20 1.00

a n � 21.

FIGURE 8A Three-Year Moving Average of the Changing Structure of TQM Arguments, 1975–95

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

90.00

100.00%

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994Year

Ratio ofPropositions Claim

Minor premiseMajor premise 2Major premise 1

28 FebruaryAcademy of Management Journal

the results.4 A significant p-value indicates that themodel is significant overall. A positive coefficientmeans that a higher value on the time variableincreases the likelihood of being in a higher cate-gory of the argument types. In other words, there isan increased chance that an argument made later intime will fall in a higher category.

The odds ratio we obtained using the coefficientis 1.12, which means that there is a 12 percentincrease in the odds of being in a higher responsecategory per unit increase in time. Because ourmain interest was in the structural change of thenew TQM argument, we also ran the ordered logitmodel with a three-category dependent variablewith category 0 omitted. The results are displayedas model 2 in Table 2. This model confirms ourcontention that time has a significant effect on ar-gument structure. As time goes on, the TQM argu-ment progresses from invoking the major premisefor the argument, then to using only the minorpremise, and finally to using only the claim. Thisprocess indicates a decrease in the need to discur-sively justify TQM and thus an increase in thedegree of its taken-for-grantedness.

The ordinal logistic regression analysis con-strains the odds ratio to be constant for all catego-ries. We performed the Brant test in STATA to testwhether the proportional odds assumption was

valid for the models. The Brant test generates achi-square statistic. If the p-value of the statistic islarge, then the proportional-odds assumption isvalid. The p-values for the two models are 0.06 and0.003, respectively. We estimated the generalizedordered logit regressions that relax the proportionalodds assumption for both models. The results donot alter our basic conclusions.

Summary of Results

Taken together, the qualitative, visual-graphical,and statistical analyses support our reconceptualiza-tion of institutionalization as change in argumentstructure. The qualitative analysis provides concreteexamples of changes in arguments as TQM institu-tionalizes. The visual-graphical analysis shows thetrends of argument structures as TQM institutional-izes. The quantitative analyses also show that impor-tant aspects of these trends and structural changes inTQM arguments are in fact statistically significant.

4 The cut points are like the intercepts in simple linearregressions. Specifically, STATA uses an underlying la-tent variable to differentiate the low categories from highcategories in the dependent variable. Cuts points repre-sent where the cuts are made. In our case, when thepredictor variable “time” is set to 0, an argument that hasa latent value equal to or lower than the value of cut point1 would be classified as category 1, and an argument thathas a value between cut point 1 and cut point 2 would beclassified as category 2, et cetera. Cut points are used tocalculate the probability that an argument in a given yearis in a particular category.

TABLE 1bResults of Simple Linear Regression Analysis of Four Propositionsa

Independent VariableProportion

Major Premise 1Proportion

Major Premise 2Proportion

Minor PremiseProportion

Claim

Time �0.01** (0.00) �0.01*** (0.00) �0.01* (0.00) 0.03*** (0.00)Constant 0.08*** (0.02) 0.26*** (0.03) 0.34*** (0.04) 0.33*** (0.04)

95% confidence interval (�.01, �.002) (�.02, �.01) (�.01, .00) (.02, .03)R2 .36 .60 .19 .75

a n � 21. Standard errors are in parentheses.* p � .05

** p � .01*** p � .001

TABLE 2Results of Ordinal Logistic Regression Analysis

of Argument Typea

IndependentVariable Model 1 Model 2

Time 0.12*** (0.02) 0.10*** (0.02)Cut point 1 �2.59 (0.34) �0.81 (0.26)Cut point 2 �0.49 (0.25) 0.65 (0.25)Cut point 3 0.89 (0.25)

Log-likelihood �323.83 �291.16Likelihood ratio �2 37.90**** 27.16****df 1 1Pseudo-R2 .06 .04n 346 338

a Standard errors are in parentheses.*** p � .001

**** p � .0001

2009 29Green, Li, and Nohria

DISCUSSION AND CONCLUSION

Theoretical Implications

Measuring institutionalization as a state and aprocess. The broad question motivating this studywas how one knows when persistent material prac-tices are institutionalizing—that is, acquiring legit-imacy. We propose using the structure of argu-ments supporting material practices to helpconceptualize institutionalization. Doing so pro-vides an empirical model for describing the acqui-sition of legitimacy as the cognitive process bywhich material practices and entities become em-bedded in taken-for-granted assumptions (Zucker,1977). As a state, the institutionalization of mate-rial practices such as TQM is reflected in the struc-ture of an argument at a given point in time. Forinstance, in the early stages of TQM institutional-ization, the notion of quality being negatively re-lated to costs was little understood and highly con-troversial. At this stage, the discourse surroundingTQM explicitly linked quality to waste and wasteto cost in order to justify the use of TQM practices.This explicit link represents a low state or degree ofinstitutionalization for TQM practices at a givenpoint in time.

Similarly, as a process, institutionalization is re-flected in the changing structure of an argumentused to justify a social practice over time. In thecase of TQM, the structure of the argument support-ing TQM practices was initially syllogistic, orstated and explained in full. Over time, this syllo-gism collapsed into an enthymeme, or more simpleexplanation, and then into a single claim or prop-osition. The decrease over time in the frequency ofpropositions required to justify the claim reflectsthe process of TQM acquiring cognitive legitimacyor taken-for-grantedness. We argue that the struc-ture of an argument at a given point in time, as wellas the change in argument structure over time, re-flects a robust and powerful model of the state andprocess of institutionalization.

In addition, this model decouples diffusion frominstitutionalization and provides an empiricalframework for understanding institutions as morethan just the material spread of social practices(Green, 2004). As opposed to equating the institu-tionalization of TQM with its diffusion, our analy-sis provides an alternative symbolic understandingof TQM institutionalization. Specifically, our re-sults show that the structure of arguments used tosupport and legitimate TQM practices changed andcoevolved as TQM practices diffused. This sym-bolic conceptualization of institutionalization con-tributes to neoinstitutional research by providing arobust linguistic measure of cognitive legitimacy,

as well as a framework for understanding how thesymbolic and material aspects of institutionscoevolve.

A rhetorical model of institutionalization alsoresonates with an overall theme in the TQM insti-tutionalization literature, in which the diffusion ofTQM practices starts as customized technical solu-tions that over time decouple from technical con-siderations and transition into fads (Abrahamson,1997; Westphal, Gulati, & Shortell, 1997; Zbaracki,1998). A rhetorical model of institutionalizationsuggests that as TQM practices become taken-for-granted, actors cease interrogating or attending tothe reasons for TQM adoption. Failing to attend toor interrogate reasons for action is both a benefitand a curse of institutions. It is a benefit becauseactors have cognitive limits, and taking successfulpractices for granted allows actors to interrogateand attend to new and more pressing problems. It isa curse, because taking successful practices forgranted reflects an actor’s failure to consciouslymake sure that a practice is indeed effective. Arhetorical model of institutionalization is a re-minder that it is the belief in the effectiveness of apractice that causes it to diffuse and institutional-ize, as opposed to its actual effectiveness. Taking apractice for granted and failing to interrogate andattend to the reasons for its adoption may help thepractice persist, but it also increases the probabilitythat the belief in the practice’s effectiveness maydecouple from the practice’s actual effectivenessand thus lead to the faddish adoption of spuriouspractices.

Institutional entrepreneurs as rhetors. Thisstudy also points to how institutional entrepre-neurs use rhetoric to shape the cognitive legitimacyunderlying institutional order and institutionalchange. This finding increases understanding ofwhere institutions come from and how they changeand transform. Although material conditions, en-dogenous contradictions, and exogenous shocksare all theorized to shape institutions (Clemens &Cook, 1999), a rhetorical model of institutionaliza-tion suggests that rhetoric and the structure of ar-guments also play roles in institutional persistenceand change. Rhetoric shapes institutions becauserhetoric shapes the interpretation and definition ofwhat material conditions, endogenous contradic-tions, and exogenous shocks mean.

For example, why did American firms adoptTQM? Was it the natural and necessary response tothe material decline of American business in the1970s? Or was it the material success of Japanesepractices that drove the diffusion and institution-alization of TQM? Our answer emphasizes the roleof rhetoric. We argue that there are multiple alter-

30 FebruaryAcademy of Management Journal

native interpretations of the material decline ofAmerican business in the 1970s and the materialsuccess of Japanese practices. We also argue thatthese interpretations may affect the adoption andpersistence of practices. Consider that prior toTQM adoption in America, most American manag-ers believed that any increase in quality wouldincrease waste because production processes werefixed. However, by the late 1970s TQM entrepre-neurs began to persuasively describe productionprocesses as flexible. A flexible production processallows for continuous and simultaneous improve-ment in quality and reduction of costs. Similarly, atthe beginning of the TQM movement, TQM prac-tices were described as specific to Japan’s collec-tivist culture and not able to work in the individu-alistic environment of American business. Yet bythe late 1970s TQM entrepreneurs began to persua-sively portray TQM practices as useful and appro-priate for American firms. In each case, the rhetor-ical interpretation of the situation shaped theperception of TQM’s utility and thus TQMadoption.

Similarly, consider the many interpretations forthe cause of Japan’s competitive advantage. SomeAmerican managers interpreted Japan’s competi-tive success as the result of low labor rates, cheapcapital, unfair government support, and culturalcharacteristics of workers (Cole & Scott, 2000; Ga-bor, 1990: 4; Kusumoto, 1987). Other Americanmanagers pointed to Japan’s new equipment andmodern factories (Gabor, 1990: 4). Still other Amer-ican managers suggested that the soft yen andcheap exports were the secret of Japanese success(Gabor, 1990: 4). If one interpreted the decline ofAmerican business as due to too much regulation,one might propose deregulation. If the success ofJapanese business were attributed to unfair Japa-nese competition, tariffs on Japanese imports mightbe advocated. In contrast, if one interpreted thesuccess of Japanese firms to TQM practices, thechoice might be to adopt TQM practices oneself.The key point is that before TQM practices or anyactions can institutionalize and diffuse, they mustmake sense. Our study of TQM arguments shows howparticular practices or paths of action are cognitivelyembedded in the taken-for-granted assumptions orarguments that make sense of these practices andshape their durability and persistence.

This explanation emphasizes how individualsactively form, reproduce, and alter institutions likeTQM, as opposed to explanations that describe in-stitutional change as an adaptive or passive re-sponse to material environmental pressures, de-mands, or shocks. Moreover, our frameworksuggests that conventional sources of institutional-

ization are perhaps credited with more explanatorypower than they actually deserve. Specifically, nei-ther the state, the professions, interorganizationalrelations, communication links, competitive rival-ries, geographic proximity, nor reputation order-ings can rationalize material practices. A rhetoricalview suggests that social actors in conversationswith other social actors provide reasons and argu-ments that rationalize and thus institutionalize ma-terial practices (Green, 2004).

At the core of a rhetorical model of institutional-ization is an appreciation of the important role ofrhetoric in the interpretation of real and materialchanges. Our intention is not to ignore material orreal changes. We acknowledge both the materialand symbolic sides of institutions. We simply arguethat the symbolic side of institutions is often over-looked, and we hope to elevate the symbolic toequal footing with the material as bases for re-searchers’ understanding of what institutions areand how they operate. The symbolic is more than amere reflection of the material. Accordingly, argu-ments are not just reflections or representations ofexogenous shocks, contradictions, or institutionalcontext. Language in general and arguments in par-ticular play a constructive and creative role in therecognition, interpretation, and meaning of shocks,contradictions, and institutional context. Materialconditions and symbolic systems both shape insti-tutionalization. An institutional context is as em-bedded in arguments as arguments are embeddedin the institutional context. Sometimes materialconditions shape the development of argumentsand reasons for acting. And at other times, argu-ments shape actions by defining and shaping themeaning of material conditions.

Institutionally embedded agency. Our rhetori-cal model of institutionalization also shows thatthe production and acceptance of arguments are theconcrete mechanisms of institutionally embeddedagency—showing how actors are passive as well asactive participants within institutions. Actors areconstrained by institutions, yet they are simulta-neously engaged as active agents in the productionof new institutions because arguments and the in-stitutional logics they shape represent both con-straints and resources. As constraints, argumentsand the material practices they support are resis-tant to change precisely because actors see them asrational representations of reality, objectively true,and morally imperative (Berger & Luckmann, 1966;Dobbin, 1994; Friedland & Alford, 1991; Scott,2001). These arguments or logics can become soself-evident or taken-for-granted that actors whoadhere to them rarely scrutinize or even articulatethem (Dobbin, 1994). Failing to scrutinize or artic-

2009 31Green, Li, and Nohria

ulate institutional logics increases the probabilitythat the practices they support will replicate withstability and persistence. In our study, we saw anexample of this when major premise 1, or the beliefthat quality and costs were positively related,helped to constrain both the thoughts and actionsof American managers, hindering the initial adop-tion of TQM. Most managers were skeptical whenfirst confronted with the promise of TQM becausethe dominant institutional logic suggested that in-creases in quality raised costs (Cole, 2000).

However, arguments do more than just constrainthe thoughts and actions of individuals. As re-sources, arguments provide a multitude of poten-tially conflicting premises that provide opportuni-ties for actors to use arguments to set oneinstitutional logic or belief against another to createnew institutional logics or beliefs (Burke, 1969b;Friedland & Alford, 1991; Townley, 1997). Ourstudy provided an example of such an event indescribing how TQM entrepreneurs used majorpremise 2 (that is, waste and costs are positivelyrelated) to change the way that managers were con-ceptualizing quality and costs. Specifically, initialconceptions of costs for American managers em-phasized the cost and expense of detecting andreworking final products. On the basis of this un-derstanding of costs, the American business com-munity believed major premise 1, or that higherquality inevitably meant higher costs. However,early advocates of TQM practices espoused majorpremise 2, which emphasizes the cost of wastefulproduction practices that produce defective prod-ucts in the first place. The first conception of costrelies on an assumption of a fixed production pro-cess, whereas the second assumes a flexible pro-duction process. In addition, the second concep-tion of cost creates not only the potential fordecoupling quality from cost but also the potentialfor negatively relating quality to costs if a flexibleproduction process improves quality. This rear-ranging and remobilization of institutional logicslaid the foundation for the decline in major premise1 (quality and costs are positively related) and therise of a new claim (quality and costs are negativelyrelated).

Using Levi-Strauss’s idea of the thinker as brico-leur, able to transform anything to something else(Levi-Strauss, 1966), we argue that the mobilizingof existing logics or premises to create new logics orpremises is a form of both institutional thinking(Douglas, 1986: 67) and rhetorical invention (Biz-zell & Herzberg, 1990; Jasinski, 2001:323–330). Inaddition, the ability to invent or construct newarguments or institutional logics through the com-

bining and exploiting of old institutional logics isan important aspect of institutional change andagency (Creed, Scully, & Austin, 2002; Friedland &Austin, 1991). Our main point is not to minimizethe complexity of agency or institutions or to de-scribe them as solely rhetorical. We simply wish toshow that rhetoric plays an important role in theagency of institutions.

Rhetoric plays an important role in agency in thatit helps delineate the intentional manipulation andacceptance of the symbols shaping interpretations.We show how changes in the structures of argu-ments point to times when individuals are con-scious of or contesting arguments, as opposed toaccepting and taking arguments for granted. Whatis reflected in the changing structure of argumentsis how actors reinterpret the world, contest thingsthat had not previously been contested, and createnew interpretations from old ones. This formula-tion resonates with recent sociological conceptionsthat suggest that the “transpositions of schemas andremobilization of resources that constitute agencyare always acts of communications with others”(Sewell, 1992: 20–21) and that agency arises from“the capacity to reinterpret or mobilize an array ofresources in terms of schemas other than those thatconstituted the array” (Sewell, 1992: 20–21). Fromthis perspective, agency involves the willful andintentional construction and arrangement of pre-mises in order to invent or discover new argumentsin support of practices. By focusing on the sym-bolic aspect of agency, we are not arguing that thisis all there is to agency or that agency is bereft ofmaterial aspects. We only argue that the construc-tion and arrangement of premises to reinterpretmaterial conditions or to develop new definitionsor new reasons for acting is an important aspect ofagency.

Finally, a rhetorical view of institutionalizationis not too over- or undersocialized. It recenters therole of agency by highlighting the role of actors inthe creation of arguments. Yet it does not ignore theconstraining influence of previous arguments orshared understandings and beliefs on the construc-tion of new arguments and the practices they sup-port. By reconceptualizing institutionalization asthe changing structure of linguistic arguments usedto justify material practices, one can see that indi-viduals are actively and creatively engaged in theconstruction of endoxa—the web of belief or insti-tutional environment that ultimately constrainsthem. If persuasive, their arguments will becometaken-for-granted and shape future beliefs and ar-guments. In a sense, people are suspended in websof significance they themselves have spun.

32 FebruaryAcademy of Management Journal

REFERENCES

Abrahamson, E. 1996. Management fashion, academicfashion, and enduring truths. Academy of Manage-ment Review, 21: 254–285.

Abrahamson, E. 1997. The emergence and prevalence ofemployee-management rhetorics: The effect of longwaves, labor unions, and turnover, 1875 to 1992.Academy of Management Journal, 40: 491–533.

Abrahamson, E., & Fairchild, G. 1999. Management fashion:Lifecycles, triggers and collective learning processes.Administrative Science Quarterly, 44: 708–740.

Aldrich, H., & Fiol, C. M. 1994. Fools rush in? The insti-tutional context of industry creation. Academy ofManagement Review, 19: 645–670.

Allaire, P. A. 1991. Quality: Where it has been and whereit is going. Journal for Quality & Participation,March: 64–66.

Anderson, J. C., Rungtusanatham, M., & Schroeder, R. G.1994. A theory of quality management underlyingthe Deming management method. Academy of Man-agement Review, 19: 472–509.

Areni, C. S. 2002. The proposition-probability model ofargument structure and message acceptance. Journalof Consumer Research, 29(2): 168–187.

Areni, C. S. 2003. The effects of structural and grammaticalvariables on persuasion: An elaboration likelihoodmodel perspective. Psychology & Marketing, 20: 349–375.

Aristotle [R. Smith, trans. & ed.]. 1989. Prior analytics.Indianapolis: Hackett.

Aristotle [H. Lawson-Tancred, ed.]. 1991. The art ofrhetoric. New York: Penguin.

Barley, S. R., & Tolbert, P. S. 1997. Institutionalizationand structuration: Studying the links between actionand institution. Organization Studies, 18: 93–117.

Baum, J. A. C., & Powell, W. W. 1995. Cultivating aninstitutional ecology of organizations: Comment onHannan, Carroll, Dundon, and Torres. American So-ciological Review, 60: 529–538.

Benson, T. 1993. Quality: If at first you don’t succeed.Industry Week, 242(13): 48–59.

Berger, P. L., Berger, B., & Kellner, H. 1974. The home-less mind; modernization and consciousness. NewYork: Vintage Books.

Berger, P. L., & Luckmann, T. 1966. The social construc-tion of reality. New York: Doubleday.

Bizzell, P., & Herzberg, B. (Eds.). 1990. The rhetoricaltradition: Readings from classical times to thepresent. Boston: Bedford Books of St. Martin’s Press.

Burke, K. 1969a. A rhetoric of motives. Berkeley: Uni-versity of California Press.

Burke, K. 1969b. A grammar of motives. Berkeley: Uni-versity of California Press.

Cameron, K. S., & Barnett, C. K. 2000. Organizational

quality as a cultural variable: An empirical investi-gation of quality culture, processes, and outcomes.In R. E. Cole & W. R. Scott (Eds.), The quality move-ment and organization theory: 271–294. ThousandOaks, CA: Sage.

Clemens, E., & Cook, J. 1999. Politics and institutional-ism: Explaining durability and change. In J. Hagan &K. Cook (Eds.), Annual review of sociology, vol. 25:441–466. Palo Alto, CA: Annual Reviews.

Cohen, J. 1960. A coefficient of agreement for nominalscales. Educational and Psychological Measure-ment, 20: 37–46.

Cole, R. E. 1999. Managing quality fads: How Americanbusiness learned to play the quality game. NewYork: Oxford University Press.

Cole, R. E. 2000. Market pressures and institutional forc-es: The early years of the quality movement. In R. E.Cole & R. W. Scott (Eds.), The quality movementand organization theory: 67–87. Thousand Oaks,CA: Sage.

Cole, R. E., & Scott, W. R. 2000. The quality movement &organization theory. Thousand Oaks, CA: Sage.

Corbett, E. P. J., & Connors, R. J. 1999. Classical rhetoricfor the modern student (4th ed.). New York: OxfordUniversity Press.

Covaleski, M., Dirsmith, M., & Rittenberg, L. 2003. Juris-dictional disputes over professional work: The insti-tutionalization of the global knowledge expert. Ac-counting, Organizations and Society, 28: 323–355.

Creed, W. E. D., Scully, M. A., & Austin, J. R. 2002.Clothes make the person? The tailoring of legitimat-ing accounts and the social construction of identity.Organization Science, 13: 475–496.

Crosby, P. 1979. Quality is free. New York: McGraw-Hill.

Dean, J. J. W., & Bowen, D. E. 1994. Management theoryand total quality: Improving research and practicethrough theory development. Academy of Manage-ment Review, 19: 392–418.

Deming, W. E. 1982. Quality, productivity and compet-itive position. Cambridge, MA: MIT Press.

Deming, W. E. 1986. Out of the crisis. Cambridge, MA:MIT Press.

Deming, W. E. 1993. The new economics for industry,government, education. Massachusetts Institute ofTechnology Center for Advanced Engineering Study,Cambridge.

DiMaggio, P. J., & Powell, W. W. 1983. The iron cagerevisited: Institutional isomorphism and collectiverationality in organizational fields. American Socio-logical Review, 48: 147–160.

DiMaggio, P. J., & Powell, W. W. 1991. Introduction. InW. W. Powell & P. J. DiMaggio (Eds.), The newinstitutionalism in organizational analysis: 1–38.Chicago: University of Chicago Press.

Dobbin, F. 1994. Forging industrial policy: The United

2009 33Green, Li, and Nohria

States, Britain, and France in the railway age.Cambridge, U.K.: Cambridge University Press.

Donaldson, L. 1995. American anti-management theo-ries of organization: A critique of paradigm prolif-eration. Cambridge, U.K.: Cambridge UniversityPress.

Douglas, M. 1986. How institutions think. Syracuse, NY:Syracuse University Press.

Easton, G. S. 1993. The 1993 state of U.S. total qualitymanagement: A Baldridge examiner’s perspective.California Management Review, 35(3): 32–54.

Easton, G. S., & Jarrell, S. L. 2000. Patterns in the deploy-ment of total quality management: An analysis of 44leading companies. In R. E. Cole & R. W. Scott (Eds.),The quality movement and organization theory:89–130. Thousand Oaks, CA: Sage.

Eccles, R., Nohria, N., & Berkeley, J. D. 1992. Beyond thehype: Rediscovering the essence of management.Boston: Harvard Business School Press.

Elster, J. 1992. The nuts and bolts of social science. NewYork: Cambridge University Press.

Emirbayer, M., & Mische, A. 1998. What is agency?American Journal of Sociology, 103: 962–1023.

Emrich, C. G., Brower, H. H., Feldman, J. M., & Garland,H. 2001. Images in words: Presidential rhetoric, cha-risma, and greatness. Administrative Science Quar-terly, 46: 527–560.

Enos, T., & Brown, S. C. 1993. Defining the new rheto-rics. Newbury Park, CA: Sage.

Feigenbaum, A. V. 1983. Total quality control (3rd ed.).New York: McGraw-Hill.

Fligstein, N. 1990. The transformation of corporate con-trol. Cambridge, MA: Harvard University Press.

Friedland, R., & Alford, R. R. 1991. Bringing society backin: Symbols, practices, and institutional contradic-tions. In W. W. Powell & P. J. DiMaggio (Eds.), Thenew institutionalism in organizational analysis:232–263. Chicago: University of Chicago Press.

Friedland, R., & Mohr, J. 2004. The cultural turn in Amer-ican sociology. In R. Friedland & J. Mohr (Eds.),Matters of culture: Cultural sociology in practice:1–70. Cambridge, U.K.: Cambridge University Press.

Gabor, A. 1990. The man who discovered quality. NewYork: Penguin Books.

Garvin, D. A. 1991. How the Baldridge Award reallyworks. Harvard Business Review, 69(6): 80–93.

Geertz, C. 1973. The interpretation of cultures. NewYork: Basic Books.

Gill, A. M., & Whedbee, K. 1997. Rhetoric. In T. A. VanDijk (Ed.), Discourse as structure and process: Dis-course studies—A multidisciplinary introduction,vol. 1: 157–184. Thousand Oaks, CA: Sage.

Giroux, H., & Taylor, J. 2002. The justification of knowl-edge: Tracking the translations of quality. Manage-ment Learning, 33: 497–517.

Green, S. 2004. A rhetorical theory of diffusion. Acad-emy of Management Review, 29: 653–669.

Greenwood, R., Suddaby, R., & Hinings, C. R. 2002. The-orizing change: The role of professional associationsin the transformation of institutionalized fields.Academy of Management Journal, 45: 58–80.

Hackman, J. R., & Wageman, R. 1995. Total quality manage-ment: Empirical, conceptual, and practical issues. Ad-ministrative Science Quarterly, 40: 309–342.

Hart, R. P., & Daughton, S. M. 2005. Modern rhetoricalcriticism (3rd ed.). Boston: Pearson/Allyn & Bacon.

Ishikawa, K. 1985. What is total quality control? TheJapanese way [D. J. Lu, trans.]. Englewood Cliffs, NJ:Prentice-Hall.

Jasinski, J. 2001. Sourcebook on rhetoric: Key conceptsin contemporary rhetorical studies. ThousandOaks, CA: Sage.

Jepperson, R. L. 1991. Institutions, institutional effects, andinstitutionalism. In W. Powell & P. J. DiMaggio (Eds.),The new institutionalism in organizational analysis:146–163. Chicago: University of Chicago Press.

Juran, J. M. 1989. Juran on leadership for quality. NewYork: Free Press.

Kieser, A. 1997. Rhetoric and myth in management fash-ion. Organization: The Interdisciplinary Journal ofOrganization, Theory, and Society, 4(1): 49–74.

Kusumoto, S. 1987. Manager’s journal: Japanese strategymade in the U.S.A. Wall Street Journal (easternedition): 1.

Lawler, E., & Mohrman, S. 1985. Quality circles after thefad. Harvard Business Review, 63(1): 65–71.

Lawler, E., Mohrman, S. E., & Leford, G. J. 1992. Em-ployee involvement and total quality manage-ment. San Francisco: Jossey-Bass.

Levi-Strauss, C. 1966. The savage mind. London: Wei-denfeld and Nicholson.

McGuire, W. G. 1960. A syllogistic analysis of cognitiverelationships. In M. J. Rosenberg, C. I. Hovland, W. J.McGuire, R. P. Abelson, & J. W. Brehm (Eds.), Atti-tude organization and change: 65–111. New Ha-ven, CT: Yale University Press.

Nohria, N., & Green, S. E. 1993. Ford: Petersen’s turn-around. Case no. 493-040, Harvard Business School,Boston.

Nohria, N., & Green, S. E. 1996. Efficiency and legiti-macy: The adoption of TQM by large industrialcorporations. Harvard Business School working pa-per no. 96-055, Boston.

Nohria, N., & Harrington, B. 1994. The rhetoric ofchange. Research note no. 9-494-036, Harvard Busi-ness School, Boston.

O’Neill, J. 1997. Rhetoric, science, and philosophy. Phi-losophy of the Social Sciences, 28(2): 205–225.

Perelman, C. 1982. The realm of rhetoric. Notre Dame,IN: University of Notre Dame Press.

34 FebruaryAcademy of Management Journal

Perelman, C., & Olbrechts-Tyteca, L. 1969. The new rhet-oric: A treatise on argumentation. Notre Dame, IN:University of Notre Dame Press.

Phillips, N., Lawrence, T., & Hardy, C. 2004. Discourseand institutions. Academy of Management Review,29: 635–652.

Powell, T. C. 1995. Total quality management as compet-itive advantage: A review and empirical study. Stra-tegic Management Journal, 16: 15–37.

Rao, H. 2003. Institutional change in Tocque Ville: Nou-velle cuisine as an identity movement in Frenchgastronomy. American Journal of Sociology, 108:795–843.

Revlin, R., & Leirer, V. O. 1978. The effects of personalbiases on syllogistic reasoning: Rational decisionsfrom personalized representations. In R. Revlin &R. E. Mayer (Eds.), Human reasoning: 51–81. Wash-ington, DC: V. H. Winston.

Scott, W. R. 1995. Institutions and organizations. Thou-sand Oaks, CA: Sage.

Scott, W. R. 2001. Institutions and organizations (2nded.). Thousand Oaks, CA: Sage.

Selznick, P. 1949. TVA and the grass roots: A study ofpolitics and organization. Berkeley: University ofCalifornia Press.

Sewell, W. 1992. A theory of structure: Duality, agency,and transformation. American Journal of Sociology,98: 1–29.

Strang, D., & Meyer, J. W. 1994. Institutional conditions fordiffusion. In R. W. Scott & J. W. Meyer (Eds.), Institu-tional environments and organizations: Structuralcomplexity and individualism: 100–111. ThousandOaks, CA: Sage.

Suchman, M. C. 1995. Managing legitimacy: Strategicand institutional approaches. Academy of Manage-ment Review, 20: 571–610.

Suddaby, R., & Greenwood, R. 2005. Rhetorical strategies oflegitimacy. Administrative Science Quarterly, 50:36–67.

Thornton, P. 2002. The rise of the corporation in a craftindustry: Conflict and conformity in institutional log-ics. Academy of Management Journal, 45: 81–101.

Thornton, P. H., & Ocasio, W. 1999. Institutional logicsand the historical contingency of power in organiza-tions: Executive succession in the higher educationpublishing industry, 1958–1990. American Journalof Sociology, 105: 801–844.

Tolbert, P. S., & Zucker, L. 1983. Institutional sources ofchange in the formal structure of organizations: Thediffusion of civil service reform 1880–1935. Admin-istrative Science Quarterly, 30: 22–39.

Tolbert, P. S., & Zucker, L. G. 1996. The institutionaliza-tion of institutional theory. In S. R. Clegg, C. Hardy,& W. Nord (Eds.), Handbook of organization stud-ies: 175–190. Thousand Oaks, CA: Sage.

Toulmin, S., Reike, R. D., & Janik, A. 1984. Introductionto reasoning. New York: Macmillan.

Toulmin, S. E. 1969. The uses of argument. Cambridge,UK: Cambridge University Press.

Townley, B. 1997. The institutional logic of performanceappraisal. Organization Studies, 18: 261–285.

Watson, T. J. 1995. Rhetoric, discourse and argument inorganizational sense making: A reflexive tale. Orga-nization Studies, 16: 805–821.

Westphal, J., D., Gulati, R., & Shortell, S. M. 1997. Cus-tomization or conformity? An institutional and net-work perspective on the content and consequencesof TQM adoption. Administrative Science Quar-terly, 42: 366–394.

Winter, S. G. (Ed.). 1994. Organizing for continuousimprovement: Evolutionary theory meets the qual-ity revolution. Oxford, U.K.: Oxford UniversityPress.

Wuthnow, R., Hunter, J. D., Bergesen, A. J., & Kurzwell,E. 1984. Cultural analysis: The work of Peter L.Berger, Mary Douglas, Michel Foucault, and Jur-gen Habermas. Boston: Routledge & Kegan Paul.

Zbaracki, M. 1998. The rhetoric and reality of total qual-ity management. Administrative Science Quar-terly, 43: 602–636.

Zucker, L. G. 1977. The role of institutionalization incultural persistence. American Sociological Re-view, 42: 726–743.

Zucker, L. G. 1989. Combining institutional theory andpopulation ecology: No legitimacy, no history.American Sociological Review, 54: 542–545.

APPENDIX A

More Examples of Statements of TQM Entrepreneursa

We adopted TQM back in the late 1970s when one of the labmanagers read Quality Is Free and attended the CrosbyCollege. These managers came back believers andconvinced the company to send more managers to theCrosby College. (1979)

The TQM initiative started in the ’82 and ’83 period and wasdriven by Demming’s [sic] talks and a very dynamic andstrong executive vice president that saw this as the wave ofthe future. (1982)

The quality movement was introduced to [company] throughmanufacturing people who went to seminars, thought itwas a great idea, and brought it home. (1983)

The company started its TQM program corporate-wide in1983. Mr. Demming came to [city], I assume on the lecturecircuit, and our executives were impressed with him andasked him to visit and speak to the company. (1983)

a Entrepreneurs include TQM “gurus,” consultants,and business managers actively promoting TQM prac-tices. All quotations are from our interviews. Boldfaceindicates highlights by the author; the year in parenthe-ses at the end of each quote is the year the interviewee’sfirm adopted TQM.

2009 35Green, Li, and Nohria

The president of the company saw the changing businessenvironment and discerned that TQM was the only way tokeep the company competitive. We introduce it into thecompany through a Malcolm Baldridge-type process ofinternal measures. The company sent 70 top managers tolearn from the gurus what was going on. (1987)

The idea (TQM) came into the business as a result of all of thenoise in business circles at the time talking about thevirtues of quality. You could not pick up a Fortune orBusinessWeek at the time and not hear about quality. . . .For a company our size it was essential that we maintain onthe cutting edge of new processes of production. Executivesand top management thought that it would be best if we trya pilot program in one of the divisions and then it spreadthrough the rest of the company. It’s now in every division.(1987)

APPENDIX B

Additional Coding Examples

Old claim, or major premise 1: Quality and costs arepositively related.

Indeed, the company enjoys a mystique among its peersin the candy business that few if any companies in anyindustry possess. One source of that mystique is Mars’sfanaticism about the quality and freshness of its productsand the cleanliness of its plants. “They do everythingright to keep the quality of the product up,” says thedirector of a confectionery wholesalers association,sounding almost rhapsodic about Mars’s virtues. “On theother hand, when the price of sugar and cocoa went outof sight, some other companies cut costs, and their qual-ity declined.” (1978)

Major premise 2: Waste and costs are positively related.

Not enough people understand that, if products do nothave to be reworked or remade, it will be easier to meetschedules and cost goals. (1982)The goal is simple: zero defects, yielding major costsavings. The company and the Australian customer areabout to discover whether it is achievable. (1988)“Every time you make a mistake, it costs you money,” hepoints out. “My guess is that the hidden cost of doingthings two or three times in non-manufacturing areas ispretty enormous. It can be very hard to track down—andthe only way to do it is with a very thorough [computer-based] system like TQM.” (1988)

Minor premise: Quality and waste are negatively related.

In a nutshell, he explains, TQM boils down to tracking“recurring problems” and errors so that the causes canbe identified and eliminated. Take a distribution center.Errors of all kinds can creep in—orders entered improp-erly, inventory information entered improperly in a com-puter system, shipping the wrong thing to customers.(1988)

A total-quality organization eliminates waste. Non-val-ue-adding work is waste. We have to eliminate the waste.(1991)To win approval from AT&T’s top management thesedays, proponents of any new quality initiative must firstdemonstrate that the effort will yield at least a 30% dropin defects. (1994)

Claim: Quality and costs are negatively related.

In her book The Keys to Excellence—The Story of theDeming Philosophy she explains that the fundamentalprecept Dr. Deming taught the Japanese is that “buildingquality into a product brings lower costs and henceimprovement in productivity and competitive position.”(1988)“Reducing costs through quality control will determinewhether we’ll be competitive in the next decade,” saysW. Kent Sterret, FP&L’s quality chief. (1989)Deming was soon imparting to America’s former enemy aquality-control philosophy that would become a formi-dable economic weapon. He combined new statisticaltechniques for pinpointing quality problems with 14 ofhis own managerial principles, most of them at odds withthe sacred canon of U.S. business. A sampling: Don’tshake up employees with annual reviews. Don’t offerincentive pay. Don’t set production quotas. Do makeemployees feel secure, enlisting all in a search for betterways to get the job done. Do coddle your customers andsuppliers. And, if you want to lower costs, executives,don’t lower quality. Raise it. (1991)Total quality management is a thoroughly pragmatic (al-beit extremely demanding) way to make increasing prof-its by not only satisfying but delighting customers withbetter products and services for less cost. (1994)

Sandy Edward Green Jr. ([email protected]) is an assistantprofessor of management at the University of SouthernCalifornia, Los Angeles. He received his Ph.D. from theHarvard University Graduate School of Business. Hiscurrent research interests include neoinstitutional the-ory, rhetorical theory, and equity valuation.

Yuan Li ([email protected]) is a Ph.D. can-didate in the Department of Management and Organiza-tion, University of Southern California. Her current re-search interests include neoinstitutional theory, rhetoric,and China’s economic reform.

Nitin Nohria ([email protected]) is the Richard P. Chap-man Professor of Business Administration and seniorassociate dean and director of faculty development at theHarvard Business School. He received his Ph.D. from theSloan School of Management, Massachusetts Institute ofTechnology. His research centers on leadership and cor-porate transformation.

36 FebruaryAcademy of Management Journal