Management Fashion Trends: Understanding the Evolution of Employee-Management Techniques, Summaries of Communication

Download Management Fashion Trends: Understanding the Evolution of Employee-Management Techniques and more Summaries Communication in PDF only on Docsity! In this theory development case study, we focus on the relations across recurrent waves in the amount and kind of language promoting and diffusing, and then demoting and rejecting, management techniques—techniques for transforming the input of organizational labor into organi- zational outputs. We suggest that rather than manifesting themselves as independent, transitory, and un-cumulative fads, the language of repeated waves cumulates into what we call management fashion trends. These trends are protracted and major transformations in what man- agers read, think, express, and enact that result from the accumulation of the language of these consecutive waves. For the language of five waves in employee-man- agement techniques—management by objectives, job enrichment, quality circles, total quality management, and business process reengineering—we measure ratio- nal and normative language suggesting, respectively, that managers can induce labor financially or psychologically. The results reveal a gradual intensification in the ratio of rational to normative language over repeated waves, sug- gesting the existence of a management fashion trend across these techniques. Lexical shifts over time, howev- er, serve to differentiate a fashion from its predecessor, creating a sense of novelty and progress from the earlier to the later fashions. Scholars have recognized for a long time now that so-called fads or fashions affect management techniques (Sumner, 1959), those linguistic prescriptions for how to transform organizational inputs into organizational outputs (Ghaziani and Ventresca, 2005). The “balanced score card” label, for exam- ple, denotes language prescribing behaviors necessary to transform certain financial and non-financial results into multi- dimensional measures of organizational performance. Organi- zations adopt this prescriptive language by behaving, or appearing that they behave, according to its prescriptions (Chevalier, 1991; Zbaracki, 1998; Czarniawska, 2005). Organi- zations disseminate stories about the success (or failure) in the use, or purported use, of management techniques, possi- bly causing their diffusion (or rejection) across other organiza- tions (Zbaracki, 1998). This language also causes organiza- tional stakeholders to react favorably (or unfavorably) to organizations’ adoptions (or rejections) of such management techniques, enhancing (or decreasing) adopting organizations’ and their chief executive officers’ reputations, as well as stakeholders’ contributions to both (Staw and Epstein, 2000). The natural history of fads and fashions in management tech- niques manifests what scholars have called a management technique’s popularity wave (Abrahamson, 1991; Burns and Wholey, 1993; Kieser, 1997; Carson et al., 2000; Scarborough and Swan, 2001). Following a latency phase in the popularity of a management technique, a rapid increase occurs in the amount of language promoting and diffusing this technique. The rapid increase usually gives way to a twofold outcome: first, to an equally rapid decrease in such promotional lan- guage and diffusion and, second, to a rapid increase in lan- guage demoting this technique, causing its widespread aban- donment (Abrahamson and Fairchild, 1999). Ultimately, both ©2008 by Johnson Graduate School, Cornell University. 0001-8392/08/5304-0719/$3.00. Employee-management Techniques: Transient Fads or Trending Fashions? Eric Abrahamson Columbia University Micki Eisenman Baruch College 719/Administrative Science Quarterly, 53 (2008): 719–744 types of languages—those promoting and demoting the tech- niques—diminish. Scholars have tended to use the terms fad and fashion indis- criminately, however, when referring to the popularity waves of management techniques and to ignore that two very dif- ferent theories—the theory of fads and the theory of fash- ions—explain such waves (Cole, 1999; Brindle and Stearns, 2001; Klincewicz, 2006). Theories of fads and theories of fashions describe two very different social processes. Fads are collective behaviors thought to arise from a chance con- junction of forces triggering their diffusion, whereas fashions result from supply and demand in a knowledge market. Each has very different consequences for what we will call “fad- dish” or “fashionable” popularity waves in management techniques, depending on which theory we invoke to explain these waves (Abrahamson, 1991; Strang and Soule, 1998). Currently, a large proportion of the literature on the popularity waves of management techniques relies either explicitly or implicitly on the theory of fads, which suggests that a chance conjunction of social forces causes waves in the popularity of management techniques (Abrahamson, 1991; Gill and Whit- tle, 1993; Abrahamson and Rosenkopf, 1993; Abrahamson and Fairchild, 1999). As a result, the theory of fads suggests that different faddish popularity waves of management tech- niques occur independently of each other. Consequently, scholars have generally studied one faddish wave in the pop- ularity of a single management technique at one time. Fur- thermore, certain scholars have concluded that simplistic, and often illogical as well as social, imitative forces cause the choices impelling faddish popularity waves in management techniques (Gill and Whittle, 1993). As a result, scholars have long considered fads largely insignificant, non-rational swings that come and go, with little or no lasting impact on the lan- guage of management techniques and on organizations themselves (Cole, 1989). Yet a closer look at some of these management techniques shows them to be far from insignifi- cant. Strategic downsizing, for instance, the handmaiden of business process reengineering, gave rise to the often dis- ruptive and needless firing of thousands of employees (Cameron, 1998). Although the theory of fads implies that transitory swings in the popularity of management techniques have single-fad, unimportant, and non-cumulative impacts on the language of management techniques, theories of fashion—gradual, protracted, and major trends over time—suggest that fash- ions in management techniques, like other fashions, cumu- late over time (cf. Blumer, 1969). Such transformations would occur because of the accumulation of consecutive waves in the popularity of language promoting (or demoting) the diffu- sion of management techniques. Taken together, rather than singly, fashion trends are likely to have multi-fashion, impor- tant, and cumulative impacts on the prescriptive language of management techniques and on the behavioral enactments of these prescriptions. To advance our arguments about man- agement fashion trends, we conduct a theory development case study and use the concept of management fashion trends we develop to study empirically a succession of five 720/ASQ, December 2008 Year — Skirt Diameter in Inches — Polynomial (Skirt Diameter in Inches) son, 1979). Instead, organizational systems sense popular demand and both supply and broadcast cultural forms that satiate this popular demand (Hirsch, 1972). In the case of fashion, fashion-setting organizations constitute the supply side of an institutionalized fashion-setting market, whereas fashion followers constitute the market’s demand side. Fash- ion setters sense the demand of fashion followers, produce fashionable forms, and broadcast them to fashion followers. Blumer (1969) stressed how recurrent fashions cumulate into what he called “fashion trends,” gradual, protracted, and major transformations in the fashionable forms that fashion setters broadcast successfully. Three fashion case studies from outside the field of management help illustrate the incremental processes by which succeeding fashions bring about major changes in social forms. In the first case, on fashion in women’s dresses, most people believe that the lengths of dresses’ hemlines, as well as their widths, fluctu- ate with little apparent rhyme or reason almost year after year. This fluctuation in single fashions masks clear trends across successive fashions. As figure 1 reveals, for example, despite yearly fluctuations in the width of skirts, Richardson 723/ASQ, December 2008 Fads or Fashions? Figure 1. Case Illustration of a fashion trend in fluctuations in skirt width between 1788 and 1933, based on data from Richardson and Kroeber (1940). 120 100 80 60 40 20 0 S ki rt D ia m et er in In ch es 17 88 17 93 17 98 18 03 18 08 18 13 18 18 18 23 18 28 18 33 18 38 18 43 18 48 18 53 18 58 18 63 18 68 18 73 18 78 18 83 18 88 18 93 18 98 19 03 19 08 19 13 19 18 19 23 19 28 19 33
and Kroeber (1940) found a long-term fashion trend between 1787 and 1933, as evidenced by the polynomial trend line, with a peak in 1858 and troughs in 1810 and 1926. In the second case, Robinson (1976) measured the hirsute- ness (length of sideburns, beards, and mustaches) depicted in the pictures of male models in the pages of the Illustrated London News, between 1842 and 1972. As figure 2 shows, despite yearly variability in the fashionability of different degrees of masculine hirsuteness, Robinson (1976) found a long-term fashion trend in the amount of hair on male fashion models’ faces, indicated by the polynomial trend line, with a hirsuteness peak in 1892, followed by a gradual increase in the clean-shavedness of men until the study window closed in 1972. Such short-term variability along long-term fashion trends does not restrict itself to women’s dresses and men’s facial hair. It also appears in architecture and in auto design, for example (Robinson, 1958). In the third case study, Robinson (1975) revealed a fashion trend in the sleekness of cars (ratio of length to height) between 1925 and 1973, an impercepti- bly gradual decline of 21/2 feet in roof height over 50 years. 724/ASQ, December 2008 Figure 2. Case illustration of a fashion trend in fluctuations in the percentage of clean-shaven models between 1842 and 1972, based on data from Robinson (1976). 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1842 1852 1862 1872 1882 1892 1902 1912 1922 1932 1942 1952 1962 1972 Year — Percent Clean Shaven — Polynomial (Percent Clean Shaven) P er ce n t C le an S h av en
These case studies demonstrate that although each instantia- tion of a fashion appears isolated, the sequence of fashions cumulates into a clear directional trend. The advantage of the theory of fashions over the theory of fads lies in providing this cumulative, directional dimension. Viewed in this way, new fashions both grow out of and extend previous fashions. As such, fashions connect and cumulate, rather than consti- tuting individual and unpredictable responses to breakdowns in institutional constraints. Blumer’s (1969) theory explains the evolution of fashion trends and why fashion consumers pursue new fashion extensions along these trends. Blumer’s theory of fashion. According to Blumer (1969), fashion setters attempt to sense fashion consumers’ prefer- ences. Fashion setters hope that the forms they select, and try to launch into fashion, will become highly profitable because they match fashion consumers’ shifting preferences and, as a result, will be widely adopted by these consumers. The offerings of fashion setters who fail to correctly identify consumers’ shifting preferences meet with little consumer demand and may even drive fashion setters out of business (Koplin-Jack and Schiffer, 1948). Thus the rules of the market discipline fashion setters. Fashion setters cannot know consumers’ exact preferences in advance. They can, however, develop an incipient concep- tion of what consumers might want. Blumer’s (1969) theory suggests that consumers’ preferences have two dimensions. First, fashion consumers have a “current preference” for a particular type of form (e.g., a wider tie). Second, they also have a “trending preference” for a series of forms that will increasingly resemble a particular ideal type (e.g., a maximally wide tie). Consumers’ trending preferences thus evolve along what we call a trending axis. A trending preference for nar- rower ties, for instance, defines a trending axis anchored at two oppositional poles. One ideal type of tie manifests itself by the widest possible tie, and the other ideal type embodies itself in the narrowest possible tie. Fashions trend gradually along such trending axes. Succeeding fashions move from wider to narrower dresses, or vice versa, along the sartorial trending axis (Richardson and Kroeber, 1940); male models from more hairy to more depilated, or vice versa, along the hirsuteness trending axis (Robinson, 1976); and cars from boxier to sleeker, or vice versa, along the auto-body trending axis (Robinson, 1975). Blumer’s (1969) theory of fashions explains why fashion trends occur and why succeeding fashions intensify along trending axes. If, in their constant search for fashion follow- ers’ preferences, one of the many fashion setters launches a fashion that deviates from the ongoing fashion trend (launch- es a shorter skirt when the trend for skirts has been longer and longer), and the deviant fashion becomes popular, this fashion reveals to all fashion setters not only fashion con- sumers’ current preference but, more important, the emer- gence of a new trending preference (one toward shorter skirts). At the time of such a fashion trend reversal, the com- munity of fashion setters usually perceives only vaguely the new trending preference, because even fashion consumers experience their trending preferences as indistinct and incipi- 725/ASQ, December 2008 Fads or Fashions? normative subjects such as psychology, dominated linguistic production about this fashion during its upswing. The more rational language of later fashion setters, who had a back- ground in rational subjects such as statistics and quality con- trol, dominated linguistic production about this fashion during its downswing. Abrahamson and Fairchild’s (1999) as well as David and Strang’s (2006) research suggested that the normative and rational language of single fashions could evolve during their life cycle for at least two reasons. First, over the life cycle of a single employee-management fashion, fashion setters might shift their language along the rational-normative trend- ing axis, either from more normative to more rational or vice versa. Alternatively, fashion setters with a more normative (or rational) background might replace fashion setters with a more rational (or normative) orientation. In either case, we can speculate that the community of management fashion setters would use increasingly rational (or normative) lan- guage during each fashion’s period of popularity. We employ this line of reasoning to formulate an orienting question guid- ing our empirical investigation: Question 1b: Does language, within each employee-management fashion, trend along a rational-normative trending axis? Linguistic differentiation. Although the language disseminat- ing employee-management fashions could undergo a move- ment along a trending axis, the language of employee-man- agement fashions cannot appear to be only a simple trending process along such a trending axis. Norms of progress require the impression, for each fashion, that fashion setters have created an entirely new and improved employee-man- agement fashion. In the sartorial case, for instance, con- sumers do not replace this year’s fashionable brown skirt with next year’s fashionable brown skirt simply because the latter is a shorter brown skirt. The skirt must not only appear shorter. It must also differ on some other stylistic dimen- sion—more flowery, puffier, or more colorful, for example. Likewise, managers, in the grip of the business process reengineering fashion would not become enraptured with the language of a new business process reengineering fashion simply because it contained more rational language. For man- agerial demand for the next employee-management fashion to increase rapidly, the language of subsequent fashions must sound radically progressive by being different—some- thing like the language of the Six-Sigma fashion, for instance. We use the term “fashion differentiation” to denote this pro- gressive imperative in the language of employee-manage- ment fashions. Fashion differentiation occurs when new lan- guage differentiates each fashion from its predecessors. We formulate this idea as an orienting question guiding our empirical investigation: Question 2: Does the language of succeeding employee-manage- ment fashions make them appear different? Fashionable language thus should do more than make one fashion appear to be moving along a trending axis—what we call fashion trending. Fashionable language must also make 728/ASQ, December 2008 each fashion appear progressive relative to its predeces- sors—what we call fashion differentiation. Fashionable lan- guage, in short, contains a combination of variation across fashions and continuity along fashion trends (Czarniawska, 2005: 136). METHODS To study the possibility that management fashions trend along a rational-normative trending axis, we focused on fash- ions in rational and normative employee-management tech- niques. We based our initial data search on the sequence of four employee-management techniques revealed by Abra- hamson and Fairchild (1999) over the 1970 to 1999 period: job enrichment, quality circles, total quality management, and business process reengineering. To limit left-censoring issues, we also added the one earlier technique that our data collection method could support, management by objectives. These five techniques constitute different combinations of rational and normative employee-management techniques. Based on prior work (Hackman and Oldham, 1976; Barley and Kunda, 1992; Hammer and Champy, 1993), we classified each employee-management fashion as predominately ratio- nal or normative. In line with Barley and Kunda (1992), we refer to management by objectives as predominantly rational and to quality circles and total quality management as pre- dominantly normative. Because other research did not exam- ine job enrichment and business process reengineering in the rational and normative context, we engaged with the defini- tive texts describing these techniques. Based on our reading of Hackman and Oldham (1976), we classified job enrichment as predominantly normative. Based on our reading of Ham- mer and Champy (1993), we classified business process reengineering as predominantly rational. Due to data limita- tions, we could examine at best a five-fashion-long trend. Question 1 Measures: Rational-Normative Trending Axis To measure linguistic trending processes along a rational-nor- mative trending axis, we used a computer-automated con- tent-analysis methodology. We used abstracts of articles about the five management techniques from the ABI Inform database to measure the language of the five fashions. The ABI Inform database stores articles covering business-related topics from more than 2,000 professional, scholarly, trade, and general-interest periodicals. We used abstracts, rather than full text, for two reasons. First, scholars who have test- ed the difference in language between full texts of articles and their abstracts (e.g., Abrahamson and Fairchild, 1999) have shown that abstracts constitute a good proxy for the entire text. Second, ABI Inform does not include the full text of each article back to the 1970s. To maximize the amount of text analyzed, we wanted to analyze the full population of articles, rather than a likely time-biased sample, and preferred a method that would allow us to go back to the 1970s. We used descriptive rather than inferential statistics because we used the entire population of articles, not a sample. We used the label of the employee-management technique as the sub- ject heading in an ABI Inform database search. The subject 729/ASQ, December 2008 Fads or Fashions? headings used were “management by objectives,” “job enrichment,” “quality circles,” “total quality management,” and “business process reengineering.” The use of subject headings is the standard methodological procedure in these types of content-analytic studies (Abrahamson and Fairchild, 1999; Spell, 1999; Raub and Ruling, 2001; Ghaziani and Ven- tresca, 2005; Giroux, 2006). Using this method, we obtained and analyzed a total of 12,174 abstracts spanning the period from January 1971 to August 2000.2 Rational-normative content-analysis dictionary. Computer- automated analysis differs depending on whether a single word or a word within the context of a larger textual unit (sentence, paragraph, or document) constitutes the unit of analysis. In our analysis, we followed the approach of using the single word as the unit of analysis (Stone et al., 1966). Thus we coded a word as normative or rational regardless of the context in which it appeared (see Abrahamson and Fairchild, 1999, for a similar approach) for three reasons. First, and most important, the word alone denotes member- ship in a linguistic community. For instance, the sheer use of the word “interface” denotes membership in a linguistic community of computer aficionados, regardless of how “interface” is used. Likewise, the word “efficiency” alone denotes membership in a speech community that attends, at least in part, to the rational construct “efficiency.” Second, we had a very large amount of textual data available, and focusing on the word as the unit of analysis allowed us to analyze all the data. Third, making coding judgments based on the context of words is appropriate only for much smaller corpora of text (Kelle, 1995; Kabanoff and Abrahamson, 1997). Moreover, it invariably introduces lower reliability in the measures due to the misjudgments of human coders. Stone et al.’s (1966) approach has the benefit of being com- pletely reliable, as it depends only on computer word counts. We followed a procedure similar to that used by Wade, Porac, and Pollock (1997) and Porac, Wade, and Pollock (1999) to develop a content-analysis dictionary for coding rational and normative words in the language of employee- management techniques. These authors started out by identi- fying the concepts they wanted to analyze and, through an iterative process, developed a content-analysis dictionary that distinguished which words belonged to which concepts. A computer count of the dictionary words measured the preva- lence of each concept in the text analyzed. In our study, we measured the prevalence of rational and normative language. Below, we explain the two-step process we used for gener- ating our rational-normative content-analysis dictionary. Step 1: Neither rational nor normative. Each of the authors independently coded every word that appeared more than 30 times, a total of 4,301 words. The goal in this first step was to distinguish words that were neither rational nor normative. Excluded were parts of speech like “the” or “and” that serve a purely grammatical function, as well as words that had nothing to do with either rational or normative language, e.g., “consulting” or “turbine.” We used the general definition presented earlier to distinguish rational or normative words from words that were neither rational nor normative. Rational 2 The management by objectives technique was popular before 1971 when our data start. The management by objectives data analyzed in this study therefore represent the end of this technique’s fashionable period. 730/ASQ, December 2008 Landis and Koch (1977), the 0.77 Kappa value suggests that our coding process generated a substantially reliable dictio- nary for distinguishing rational and normative language. The coders then resolved coding disagreements by discussing which decision fitted the definition in the code book more precisely. Validity test for the rational-normative content-analysis dictionary. Rational and normative fashions denote con- structs that should differ. To test whether our rational-norma- tive content-analysis dictionary provided valid scores on nor- mative and rational measures (discriminant validity), we assumed that the ratio of rational-to-normative coded words in the abstracts of normative fashions should differ substan- tially from the same ratio for rational fashions. In other words, we assumed that the language of normative fashions would have a smaller ratio of rational-to-normative words than the language of rational fashions. According to the clas- sifications introduced above, both management by objectives and business process reengineering contain primarily rational language. Job enrichment, quality circles, and total quality management, by contrast, contain primarily normative lan- guage. Our results indicate that the average ratio of the num- ber of rational-to-normative words was 1.36 for the normative techniques and 2.58 for the rational techniques—a 90 per- 733/ASQ, December 2008 Fads or Fashions? Table 2 Content Analysis Dictionary for Level-1 Normative Language X Level 2 Individual level X X X X X Organizational level X X X X X Level 3 Analyzing: Human charac- teristics. Administering: Bottom-up management, excluding rewarding. X Rewarding: Evaluating and rewarding performance intrinsically. Organizational: Organiza- tional level constructs. X X Power: Organizational level constructs. X X Level 4 Human char- acteristics Human psy- chology and cognitions Leading Empowering Evaluating Rewarding Collectivities Organizational culture Human resource manage- ment Conflict Unions X Description Terms focusing on human characteristics. Terms focusing on any and all characteristics of socio-psychological processes. Terms wherein managers help employees. Terms wherein managers give employees deci- sion-making power. Measuring human harm to employees. Intrinsic motivators. Terms related to collectivi- ties. Terms related to organiza- tional culture. Terms related to human resource management. Terms related to labor strife. Terms related to labor unions. Sample of coded words Ability, friendly, origi- nal, competent Commitment, emo- tional, insight Encourage, challenge, coach Autonomy, democrat- ic, self-managing Burnout, turnover, resistance Belonging, enjoy- ment, fulfillment Family, group, inter- personal Atmosphere, ideolo- gy, value Hire, promote, train Bargaining, conflict, negotiations Grievances, strikes, labor management cent difference that suggests that our measure has discrimi- nant validity. Question 1: trending measure. To answer questions 1a and 1b, we had to detect the axial direction along which fashion might trend—more normative, for instance—along a rational- normative trending axis. To measure trending along this axis, the computer counted the number of normative and rational words in the text of the abstracts of each fashion. We used this count to calculate three ratio measures that would reveal trending: first, what we call the “rational ratio,” the ratio of rational words to the total words in the body of text for each fashion; second, what we call the “normative ratio,” the ratio of normative words to the total words in the body of text for each fashion; and third, the measure described above as the “rational-to-normative ratio,” the ratio of rational-to-normative words in the body of text for each fashion. Question 1a: Cross-fashion trending. Question 1a asked whether trending along a rational-normative trending axis occurs across fashions. To examine cross-fashion trending, we aggregated the abstracts for each fashion. Then, for each fashion, we calculated each of the three ratios: rational, nor- mative, and rational-to-normative. We then examined the changes in each ratio across the five fashions. Question 1b: Within-fashion trending. Question 1b asked whether single fashions trend along the rational-normative trending axis. To explore this question, we divided the study window into three roughly equal periods: 1971–1979, 1980–1993, and 1994–2001. Instead of calculating the ratio- nal-to-normative ratio for each fashion across all years, we calculated this ratio for each fashion, separately, within each of these three periods. This procedure allowed us to see the changes across periods in the rational-to-normative ratio for each fashion. Question 2 measures: The dual role of fashionable lan- guage. Question 2 asked how the language of employee- management fashions makes them appear both progressive and different. To measure progress, we used the measures of fashion trending described above. We also needed a sec- ond measure of what we called fashion differentiation, how the language of each fashion differentiates it from its prede- cessors to make it appear novel, rather than just progressive- ly farther along the rational-normative trending axis. Differentiation measure. The term lexicon refers to the set of unique words used at least once in a text. We reasoned that the language of one fashionable management technique would differentiate it from the next if it used a different lexicon, because the introduction of a new lexicon generates the impression of novelty in a management technique. To measure the degree of difference in the lexicon of our five fashionable management techniques, we aggregated the abstracts for each technique and formed each technique’s lexicon by extracting a list of the different words used at least once in each technique’s aggregated abstracts. Then we calculated the overlap between a pair of techniques’ lexicons. Following Abra- hamson and Hambrick (1997), we measured lexical overlap as the ratio of the actual number of words that co-occur in both 734/ASQ, December 2008 techniques’ lexicons to the total number of words that could possibly co-occur, that is, the number of words in the smaller of the two lexicons. High overlap in the language of two fash- ions denotes low differentiation between these fashions.3 Validity test for the differentiation measure. Our fashion differentiation measure is perfectly reliable, as it is computer generated. To assess the convergent validity of our lexical overlap measure, we assumed that there would be a high lexical overlap between the language of all the fashions we coded as primarily normative (job enrichment, quality circles, and total quality management), as well as between the lan- guage of the fashions we coded as primarily rational (man- agement by objectives and business process reengineering), when compared with the average lexical overlap across all five fashions. With respect to convergent validity, the magnitude of the results was small but as predicted. The degree of lexical overlap was greater in the language of rational fashions (0.79) and normative fashions (0.73) than the average overlap between all fashions (0.72), but the degree of lexical overlap for the normative fashions was only slightly larger than the average. This second result may have occurred because, as we show below, normative fashions that succeeded each other—like job enrichment, quality circles, and total quality management—tended to have a relatively low overlap as a result of the need to differentiate the next from the previous fashion. In contrast, management by objectives and business process reengineering, whose rational language was already separated by three decades, had a very high overlap (0.79), lending further credence to our measure’s validity. To measure discriminant validity, we assumed that there would be lower lexical overlap between the lexicon generat- ed by the aggregation of all normative fashions and that gen- erated by the aggregation of all rational fashions than between the lexicon of all fashions taken together. With respect to discriminant validity, the degree of overlap between all rational and all normative fashions (0.69) was indeed lower than the overlap between rational fashions (0.79), normative fashions (0.73), and all fashions (0.72). Taken together, our tests of convergent and discriminant validity provide weak but consistent support for the validity of our lexical overlap measure. Cross-fashion differentiation. We used the measure of lexical overlap to assess cross-fashion differentiation by examining the extent of lexical overlap between two fashions according to their sequential appearance, applying the logic presented above that showed the following sequential progression of management fashions: management by objectives, job enrichment, quality circles, total quality management, and business process reengineering. MANAGEMENT FASHION TRENDS Question 1a: Cross-Fashion Trending along the Rational- normative Trending Axis Question 1a asked along which trending axis management fashion trends progress. After the transition from manage- 3 Consider an example in which one text has a lexicon of 5,000 words, another has a lexicon of 3,000, and assume that the same 1,500 words appear in both texts’ lexicons. The maximum number of words that could occur in both texts’ lexicons is 3,000. This would occur when every word in the text with the smaller, 3,000-word lexicon appears in the text with the larger, 5,000-word lexicon. To calculate lexical overlap, we take the actual number of words that the two lexicons have in com- mon (1,500) and divide it by the maxi- mum number of words both texts’ lexi- cons could have in common (3,000). Lexical overlap equals (1,500/3,000)*100, or 50 percent. 735/ASQ, December 2008 Fads or Fashions? were able to use their data to corroborate our findings on the existence of fashion trends in management techniques. Figure 4 graphs the thematic score, either rational or norma- tive, given by Carson et al.’s (2000) coders to each manage- ment fashion they examined. As in the bottom part of figure 3, figure 4 reveals first that, overall, there is a higher rational thematic content across fashions. This finding is indicated by the polynomial trend line depicting each language—the solid line depicting rational language is almost always higher than the dashed line depicting the normative language. Second, as our results in the bottom part of figure 3 suggest, figure 4 reveals a surge in normative themes, followed by a gradual decline in the normative themes. Importantly, the surge in normative themes corresponds to a decline in ratio- nal themes, while the decline in normative themes corre- sponds to a gradual rise in rational themes. Carson et al.’s (2000) results, in direct parallel to our results, can therefore be interpreted as suggesting that changes in fashionable 738/ASQ, December 2008 Figure 4. Findings for normative and rational themes in management fashion based on data from Carson et al. (2000). T h em at ic S co re 6 5 4 3 2 1 0 M BO PE RT EA P Tg ro up s Q W L Q Cs TQ M IS O Be nc hm ar ki ng Em po w er m en t H or iz on ta l O pe ra tio ns Vi si on Re en gi ne er in g A gi le S tr at eg ie s Co re C om pe te nc ie s Normative Rational Polynomial (Normative) Polynomial (Rational)
 management language involve a gradual rebalancing of nor- mative and rational language, rather than an abrupt abandon- ment of one type of language in favor of the opposite one. In this way, Carson et al.’s study, though the authors did not draw the same interpretation from its data, does replicate the findings in our study, providing additional support for our find- ings. Question 1b: Within-fashion Trending along the Rational- normative Trending Axis Question 1b asked whether fashion trends along the rational- normative trending axis also occur within fashions. Based on our finding of a four-fashion trend toward the rational pole of the rational-normative trending axis, we wanted to examine whether we would find a similar trend toward the rational during each of our three periods: 1971–1979, 1980–1993, and 1994–2001. Figure 5 indicates that for every one of the three periods, the rational-to-normative ratio for each fashion mirrors the aggregate pattern in the top part of figure 3. Moreover, the three periods are virtually identical, save for the absence of business process reengineering, which emerged only during the second period. Different periodiza- tions did not reveal that within-fashion trending occurs along the rational-normative trending axis. Rather, it appears that the trending tends to occur across fashions rather than within them. Thus the incremental trending of fashionable manage- ment language along the rational-normative trending axis seems to be animated by fashion replacement, rather than by 739/ASQ, December 2008 Fads or Fashions? Figure 5: Within-fashion trending of rational-to-normative ratio by period. 3.000 2.500 2.000 1.500 1.000 0.500 0.000 1971–1979 1980–1993 1994–2001 R at io n al -t o -N o rm at iv e R at io  Management by Objectives  Quality Circles □ Business Process Reengineering  Job Enrichment  Total Quality Management continuous fashion trending, both within and across fashions. The evidence shows that management fashions trend incre- mentally, each increment occurring by fashion replacement. We did not find evidence of continuous trending, which would occur if trending occurred both within each fashion and across succeeding fashions. Question 2: The Dual Role of Fashionable Language Question 2 pushed us to explore how the language of fash- ionable management techniques could serve two purposes. First, this language would cause a trend in fashionable man- agement techniques along a trending axis. The results in fig- ure 3 suggested, as we noted above, that fashionable lan- guage might have served this trending role, pushing fashionable employee-management techniques further along an axis on which rational language dominates normative lan- guage to a greater degree. Fashionable management lan- guage, we proposed, might also serve a second, differentia- tion purpose. This language would differentiate each fashion from its predecessor by introducing a novel lexicon. A first set of results supports this claim. Figure 6 graphs changes in the average lexical overlap between pairs of fashions, depending on the number of intervening fashions between these two fashions. It depicts the degree of average differen- tiation between fashions that succeed each other (no inter- vening fashions), between fashions separated by one fashion (one intervening fashion), and so on. 740/ASQ, December 2008 Figure 6. Lexical overlap as a function of the number of intervening fashions. 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Le xi ca l O ve rl ap R at io 1 2 3 4 Number of Intervening Fashions Abrahamson, E. 1991 “Managerial fads and fash- ions: The diffusion and rejec- tion of innovations.” Acade- my of Management Review, 16: 586–612. 1996 “Management fashion.” Academy of Management Review, 21: 254–285. 1997 “The emergence and preva- lence of employee-manage- ment rhetorics: The effects of long waves, labor unions, and turnover, 1875 to 1992.” Academy of Management Journal, 40: 491–533. Abrahamson, E., and D. C. Hambrick 1997 “Attentional homogeneity in industries: The effect of dis- cretion.” Journal of Organiza- tional Behavior, 18: 513–532. Abrahamson, E., and G. Fairchild 1999 “Management fashion: Life- cycles, triggers, and collective learning processes.” Adminis- trative Science Quarterly, 44: 708–740. Abrahamson, E., and L. Rosenkopf 1993 “Institutional and competitive bandwagons.” Academy of Management Review, 18: 487–517. 1997 “Social network effects on the extent of innovation diffu- sion: A computer simulation.” Organization Science, 8: 289–309. Barley, S. R., and G. Kunda 1992 “Design and devotion: Surges of rational and normative ide- ologies of control in manager- ial rhetoric.” Administrative Science Quarterly, 37: 363–399. Barley, S. R., G. W. Meyer, and D. C. Gash 1988 “Cultures of culture: Acade- mics, practitioners and the pragmatics of normative con- trol.” Administrative Science Quarterly, 33: 24–60. Barthes, R. 1967 Système de la mode. Paris: Édition du Seuil. Blumer, H. G. 1969 “Fashion: From class differen- tiation to collective selec- tion.” Sociological Quarterly, 10: 275–291. Brindle, M. C., and P. N. Sterns 2001 Facing Up to Management Faddism: A New Look at an Old Force. Westport, CT: Quorum Books. Burns, L. R., and D. R. Wholey 1993 “Adoption abandonment of matrix management pro- grams: Effects of organiza- tional characteristics and interorganizational networks.” Academy of Management Journal, 36: 106–138. Cameron, K. 1998 “Strategic organizational downsizing: An extreme case.” In B. M. Staw and L. L. Cummings (eds.), Research in Organizational Behavior, 20: 185–229. Stamford, CT: JAI Press. Carson, P. P., P. A. Lanier, K. D. Carson, and B. N. Guidry 2000 “Clearing a path through the management fashion jungle: Some preliminary trailblaz- ing.” Academy of Manage- ment Journal, 43: 1143–1158. Chevalier, F. 1991 Cercles de qualité et change- ment organisationnel. Paris: Economica. Cohen, J. 1960 “A coefficient of agreement for nominal scales.” Educa- tional and Psychological Mea- surements, 20: 37–46. Cole, R. E. 1989 Strategies for Learning: Small- Group Activities in American, Japanese, and Swedish Industry. Berkeley, CA: Uni- versity of California Press. 1999 Managing Quality Fads. New York: Oxford University Press. Coleman, J. J. 1990 Foundations of Social Theory. Cambridge, MA: Harvard Uni- versity Press. Czarniawska, B. 2005 “Fashion in organizing.” In B. Czarniawska and G. Sevon (eds.), Global Ideas: How Ideas, Objects, and Practices Travel in the Global Economy: 129–146. Frederiksberg, Den- mark: Liber and Copenhagen Business School Press. Czarniawska, B., and G. Sevon 1996 Translating Organizational Change. Berlin: De Gruyter. David, R. J., and D. Strang 2006 “When fashion is fleeting: Transitory collective beliefs and the dynamics of TQM consulting.” Academy of Management Journal, 49: 215–234. Ghaziani, A., and M. J. Ventresca 2005 “Keywords and cultural change: Frame analysis of business model public talk, 1975–2000.” Sociological Forum, 20: 523–559. Gill, J., and S. Whittle 1993 “Management by panacea: Accounting for transience.” Journal of Management Stud- ies, 30: 281–295. Giroux, H. 2006 “It was such a handy term: Management fashions and pragmatic ambiguity.” Journal of Management Studies, 43: 1227–1260. Graham, J. R. 1981 “Quality circle boom part of a growing trend.” Supervision, October: 38–40. Greve, H. R. 1995 “Jumping ship: The diffusion of strategy abandonment.” Administrative Science Quar- terly, 40: 444–473. 743/ASQ, December 2008 Fads or Fashions? and render them intrinsically motivating to employees. In opposition, business process reengineering explicitly attempts to introduce routine work processes and extrinsic motivators to fit employees to these newly routinized and optimized processes. Recurrent swings and trends in the popularity of employee management techniques, therefore, generate fundamentally different mechanisms for managing employees, with sometimes massive consequences for employees caught up in the swings’ ebb and flow. REFERENCES Hackman, J. R., and G. R. Oldham 1976 “Motivation through the design of work: Test of a the- ory.” Organizational Behavior and Human Performance, 16: 250–279. Hackman, J. R., and R. Wageman 1995 “Total quality management: Empirical, conceptual, and practical issues.” Administra- tive Science Quarterly, 40: 309–343. Hammer, M., and J. Champy 1993 Reengineering the Corpora- tion: A Manifesto for Busi- ness Revolution. New York: Harper Collins. Haunschild, P. R., and C. M. Beckman 1998 “When do interlocks matter?: Alternate sources of informa- tion and interlock influence.” Administrative Science Quar- terly, 43: 815–844. Hirsch, P. M. 1972 “Processing fads and fash- ions: An organization set analysis of cultural industry systems.” American Journal of Sociology, 77: 639–659. Jackson, B. 2001 Management Gurus and Man- agement Fashions: A Drama- tistic Inquiry. London: Rout- ledge. Kabanoff, B., and E. Abrahamson 1997 “OB meets the information superhighway.” In C. L. Coop- er and S. E. Jackson (eds.), Handbook of Organizational Behavior: 453–474. New York: Wiley. Kelle, U. 1995 Computer-Aided Qualitative Data Analysis: Theory, Meth- ods and Practice. London; Thousand Oaks, CA: Sage. Kieser, A. 1997 “Rhetoric and myth in man- agement fashion.” Organiza- tion, 4: 49–74. Klincewicz, K. 2006 Management Fashions: Turn- ing Best-Selling Ideas into Objects and Institutions. New Brunswick, NJ: Transaction Books. Koplin-Jack, N., and B. Schiffer 1948 “The limits of fashion con- trol.” American Sociological Review, 13: 730–738. Landis, J. R., and G. G. Koch 1977 “The measurement of observer agreement for cate- gorical data.” Biometrics, 33: 159–174. Meyer, J., and B. Rowan 1977 “Institutionalized organiza- tions: Formal structure as myth and ceremony.” Ameri- can Journal of Sociology, 83: 340–363. Peterson, R. A. 1979 “Revitalizing the culture con- cept.” American Review of Sociology, 5: 137–166. Peterson, R. A., and N. Anand 2004 “The production of culture perspective.” Annual Review of Sociology, 30: 311–334. Porac, J. F., J. B. Wade, and T. G. Pollock 1999 “Industry categories and the politics of the comparable firm in CEO compensation.” Administrative Science Quar- terly, 44: 112–144. Raub, S., and C. C. Ruling 2001 “The knowledge manage- ment tussle: Speech commu- nities and rhetorical strategies in the development of knowl- edge management.” Journal of Information Technology, 16: 113–130. Richardson, J., and A. L. Kroeber 1940 “Three centuries of women’s dress fashions: A quantitative analysis.” Anthropological Records, 5: 111–154. Robinson, D. E. 1958 “Fashion theory and product design.” Harvard Business Review, 36 (November–December): 126–138. 1975 “Style changes: Cyclical, inex- orable, and foreseeable.” Har- vard Business Review, 53 (November–December): 121–131. 1976 “Fashions in shaving and trimming of the beard: The men of the Illustrated London News, 1842–1972.” American Journal of Sociology, 81: 1133–1141. Rogers, E. M. 1995 Diffusion of Innovations, 4th ed. New York: Free Press. Ruling, C. C. 2002 Management Fashion Adop- tion: Sensemaking and Identi- ty Construction in Individual Managers’ Adoption Accounts. Weisbaden, Ger- many: Deutscher Universitats- Verlag. Scarbrough, H., and J. Swan 2001 “Explaining the diffusion of knowledge management: The role of fashion.” British Jour- nal of Management, 12: 3–12. Smelser, N. 1963 Theory of Collective Behavior. New York: Free Press. Spell, C. S. 1999 “Where do management fashions come from, and how long do they stay?” Journal of Management History, 5: 334–348. Staw, B. M., and L. D. Epstein 2000 “What bandwagons bring: Effects of popular manage- ment techniques on corporate performance, reputation, and CEO pay.” Administrative Sci- ence Quarterly, 45: 523–556. Stone, P. J., D. C. Dunphy, M. S. Smith, and D. M. Ogilvie (eds.) 1966 The General Inquirer: A Com- puter Approach to Content Analysis. Cambridge, MA: MIT Press. Strang, D., and J. W. Meyer 1993 “Institutional conditions for diffusion.” Theory and Soci- ety, 22: 487–511. Strang, D., and S. A. Soule 1998 “Diffusion in organizations and social movements: From hybrid corn to poison pills.” Annual Review of Sociology, 24: 265–290. Sumner, W. G. 1959 Study of the Sociological Importance of Usages, Man- ners, Customs, Mores and Morals. Dover, DE: Dover Publications. Turner, R. H., and L. M. Killian 1972 Collective Behavior. Engle- wood Cliffs, NJ: Prentice-Hall. Van den Bulte, C., and G. L. Lilien 2001 “Medical innovation revisited: Social contagion versus mar- keting effort.” American Jour- nal of Sociology, 106: 1409–1435. Wade, J. B., J. F. Porac, and T. G. Pollock 1997 “Worth, words, and the justi- fication of executive pay.” Journal of Organizational Behavior, 18: 641–664. Zbaracki, M. J. 1998 “The rhetoric and reality of total quality management.” Administrative Science Quar- terly, 43: 602–638. 744/ASQ, December 2008 Copyright of Administrative Science Quarterly is the property of Administrative Science Quarterly and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.