24

Case Studies and the Configurational Analysis of Organizational Phenomena

Peer C. Fiss

Within the organization and management lit- and ample opportunities for theory building. eratures, the case study has traditionally occu- Furthermore, case studies occupy a central pied a somewhat peculiar position. On the role for in the curricula of most business one hand, case studies have a long and distin- schools, where cases are used extensively as guished history in the study of organizations. a pedagogical tool. Case studies are attractive They have featured prominently in academic in the classroom because they simulate real- research on organizations, and many of the world experiences, allowing the students to most highly regarded and influential studies take on the roles of specific decision makers in the organization and management literature in actual organizations (Mauffette-Leenders have employed a case-study approach (cf. et al. 2001). Their closeness to the experience Gephart, 2004). Some of the most well-known of life in organizations and their appreciation examples include Selznick’s (1949) study for the complexities of organizational phe- of the Tennessee Valley Authority, Blau’s nomena thus makes case studies attractive to (1962) and Crozier’s (1964) research on the academics and practitioners alike. dynamics of bureaucracy, Allison’s (1971) On the other hand, there has been a study of governmental action around the 1962 considerable debate over the scientific nature Cuban missile crisis, and Dalton’s (1959) of case studies and the ways in which they and Kanter’s (1977) work on life within the are to be conducted. This debate over the modern corporation, to name but a few. In fact, case-study approach goes back to at least one might well argue that case studies form the 1940. Back then – as is largely true the cornerstone on which modern organization today – the ability of the case study to theory has been built, providing rich insights generate new and thus contribute to into the workings of modern organizations theory development was uncontested, but

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 415 415–431 416 THE SAGE HANDBOOK OF CASE-BASED METHODS

controversy focused on whether the case In the following, I examine the logic of study could be used to derive generalizable the case study as it applies to the study of insights (e.g. Lundberg 1941, Stouffer 1941, organizations. I take the position that the case Foreman 1948). Even today, the case study study is distinguished from other organiza- remains probably the least understood and tional research strategies by a configurational least formalized methodology in the study understanding of organizational phenomena of organizations (Ragin and Becker 1992). within a specific spatial and temporal context. Despite a number of works on the use of This configurational nature of the case study the case-study methodology (e.g. Yin 1981, presents both an advantage and a challenge Eisenhardt 1989), there is still relatively little vis-à-vis other research strategies, as it raises agreement on how to write a memorable and particular methodological demands. Perhaps publishable case study, particularly if this most importantly, many of the quantitative involves the use of qualitative evidence (Van methods commonly used to formally examine Maanen 1998). organizational configurations – such as cluster Unfortunately, the classic case studies tend analysis, interaction effects, and hierarchical to be of little help in clarifying how an linear modeling – are not well suited to exemplary case study is to be conducted grasping the fundamentally configurational and written up. Most offer no separate nature of the case-study approach. As I have section on methodology at all, presenting argued elsewhere (Fiss 2007), I will suggest instead the polished product without the that set-theoretic methods such as qualitative guidelines as to how it was created. A few comparative analysis (QCA) provide a viable do offer more detailed insights into how the alternative much better suited to the configu- data was collected, how many interviews rational nature of the case-study approach. were conducted, and what sources were I begin by discussing the nature of cases and used, with some even offering summary case studies and of organizational configura- tables of descriptive statistics regarding the tions as well as how configurations may occur context (e.g. Kanter 1977). However, the at the intra-organizational, organizational, hermeneutic process of inference – how all and supra-organizational levels, and across these interviews, archival records, and notes such levels. I further focus on the methods were assembled into a coherent whole, what commonly used to examine organizational was counted and what was discounted – configurations – both qualitative and quan- remains usually hidden from the reader. titative – and examine the ability of these This is especially true for case studies methods to account for the configurational relying primarily on qualitative field-work nature of the case study. I conclude by con- methods. Acknowledging the fact that ‘there sidering the implications of employing QCA are probably as many “methods” as there for both the case-study methods and the theory are fieldworkers,’ Kunda calls the methods of organizational configurations and lay out an section of his influential ethnography of life agenda for future research on a configurational in US high-tech corporation ‘A Confessional understanding of organizational phenomena. of Sorts’ (1992, p. 229). Other authors have quite forcefully attacked current case-study practices as essentially ‘free-form research OF CASES AND CASE STUDIES where everything goes’ (Maoz 2002, p. 164). As a result of the questions and perhaps the Aserious treatment of the case-study approach mystique surrounding it, the case study thus has to come to terms with the entity that the still presents a probably more risky research approach takes its name from, that is, the case. (and career) strategy, and the majority of As Ragin (1992a) and others have pointed out, published research on organizations tends despite its widespread use and centrality in follows a variable-oriented approach using scientific discourse, the concept of the case is standard statistical estimation. frequently not well defined and the term ‘case’

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 416 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 417

is used in a variety of different ways. Cases, theories about what matters, where boundaries for example, can be understood as theoretical ought to be set, and what may be disregarded constructs or as empirical units, and their rela- as either unimportant or of a different kind. tionship to the underlying phenomena may The process of ‘casing’ is part and parcel be conceptualized in a formative or reflective of the normal conduct of social science way (e.g. Ragin 1992a). What complicates the research and often disguised by the ways in situation is that many of the various definitions which we encounter our phenomena. Much of cases have considerable merit on their own, of the social world comes to us in ‘chunky’ making a consensus definition that is both form, and frequently the boundaries of a rigorous and encompassing hard to come by. case will be intuitively plausible and useful Rather than attempting a formal definition due to social convention (e.g. the nation of a case here, I will focus on the underlying state, an organization, a subunit, a team). aspect of cases that is most relevant to the However, it is important to remember that questions at hand, namely that cases and the not everything that comes in a naturally process of ‘casing’ delimit the real world consumable form is also best understood in phenomena of interest within time and space that form. Frequently, social phenomena are (Ragin 1992b). In order to examine something perhaps better unwrapped, taken apart, and as a case – whatever that may involve in more then reassembled in order to form a more detail – one has to be able to identify and analytically useful ‘case.’ thus delimit the case from the multitude of As noted above, I will argue here that phenomena and aspects that will be not be one can distinguish the case study from studied. This process of establishing bound- other organizational research strategies by aries around a phenomenon is what reduces its configurational understanding of organiza- complexity to manageable proportions and tional phenomena within a specific spatial and turns the potentially limitless possibilities into temporal context. As the notion of ‘casing’ concrete ‘cases,’ usually by first defining the implies, a case is a holistic entity. In order theoretical category of the case, narrowing it to understand it, we have to study it in its down to a subset of cases within this category, entirety; a mere focus on parts of the case and then selecting specific empirical instances will lead to partial insight, potentially taken of this subset.1 Indeed, at an abstract level, out of context. The research strategy of the ‘every study is a case study because it is an case study must take account of this configu- analysis of social phenomena specific to time rational character of the case. As Eisenhardt and place’(Ragin 1992a, p. 2). While time and suggests, the case study ‘is a research strategy space offer intuitively appealing boundaries which focuses on understanding the dynamics around a case, other conceptualizations are present within a single setting’ (1989, p. 534). of course conceivable, such as cases of Instead of disaggregating the case into its mechanisms (e.g. Hedström and Swedberg features, operationalizing such features as 1998) or sequences (e.g. Abbott 1992, Heise variables, and then testing for correlations 1989). However, time and space are usually between these variables while controlling for the most commonly used dimensions, and as many other features as possible or relevant, I will simply follow that convention here. the case-study approach aims to preserve The process of delimiting or ‘casing’ is a the integrity of the case and understand necessary undertaking because it goes along it as a particular configuration of features with a fundamentally contextual understand- embedded in a specific context and time. This ing of cases, which holds that a case combines configurational nature makes problematic any certain characteristics or features that appear research strategy that focuses on one feature together within it and give the case its while ‘controlling’ for the effect of other essential character. As Walton notes, ‘cases features. As in the classic Indian fable about come wrapped in theories’ (1989, p. 122). the blind men examining an elephant, with More specifically, cases come wrapped in each one reporting on a different part of the

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 417 415–431 418 THE SAGE HANDBOOK OF CASE-BASED METHODS

animal, the challenge lies in comprehending implications. The case-study approach thus the case in its entirety rather than merely its tends to be comparative in nature, if only in the parts. way in which observations from a case may A configurational understanding of the case inform knowledge about organizations and also clarifies the nature of the case study. life in them more broadly by elucidating the The case study is fundamentally a research features of a larger class of similar phenomena strategy, ‘to be likened to an experiment, a (Gerring 2004, 2007). In this sense, the history, or a simulation, which may be con- case-study approach does frequently resemble sidered alternative research strategies’ (Yin more variable-oriented approaches in that the 1981, p. 59). As Yin notes, this understanding researcher tends to make typological reduc- also helps to remove the frequent confusion tions. However, as Stouffer (1941) notes, between the case study as a research strategy the case-study researcher differs from the and the types of evidence used in it (e.g. statistician in that he can do what the variable- qualitative vs. quantitative data) or the types oriented researcher frequently cannot do, of data collection methods employed to namely conduct an intensive, detailed analysis gather this data (e.g. ethnography vs. survey that can be adjusted during the course of collection). Because the case-study approach the research project. The case-study approach is a research strategy that aims to maintain is thus more dynamic in nature, and the the configurational, holistic nature of the case researchers will constantly compare theory or cases, it is not limited to any particular and data in order to achieve a fit between both form of evidence or data collection, and it (Eisenhardt 1989, Ragin 1994). can involve single or multiple cases, various Furthermore, because the logic of the case methods of data collection and several types study is not built around average tendencies and levels of analysis (cf. Eisenhardt 1989). in large samples, the selection of the case or In fact, combining evidence from multiple cases takes on a critical role. This selection is sources, such as interviews, archival data, usually informed by theoretical, not statistical and surveys frequently leads to the most reasons (Eisenhardt 1989). For example, successful organizational case studies. An Pettigrew (1990) argues that the selection of essential feature of the case-study approach is empirical sites for organizational case studies therefore its propensity to foster triangulation should focus on: (1) extreme situations, across different data sources (Eisenhardt critical incidents, and social dramas; (2) polar 1989, Yin 1994). Clearly not all situations and types that allow for stronger contrasts; and phenomena will lend themselves to a case- (3) sites with a relatively high experience study approach. Yet, a case-study approach levels of the phenomena under study, that is, will frequently be a very desirable, if also the phenomenon under study should manifest demanding approach because it requires an in- itself clearly and there should be easy to depth understanding of the case rather than access this manifestation. Case selection is a superficial understanding that goes little of course probably the most obvious example beyond the operationalization of variables. of the process of ‘casing,’ and again points to Although the view of the case study the importance of drawing boundaries around I employ here emphasizes the configurational the organizational phenomenon of interest as nature of the case as delimited in time and essential to the case study. space, it is important to note that this view does not imply an inability to make comparisons. Whereas the case study is a useful research THE NATURE OF ORGANIZATIONAL strategy when engaging unique phenomena, CONFIGURATIONS most research on organizations is not con- cerned with such one-of-a-kind entities or I have argued that the case-study approach events, but instead aims to develop an under- is unique as a research strategy in its con- standing of organizations that has broader figurational approach, that is, in its focus on

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 418 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 419

the relative arrangement of parts or elements construct; the construct does not exist inde- that can only be fully understood in their pendently of its manifestations. Taxonomic entirety. To develop these arguments more approaches to organizational configurations fully and see how they apply to the study of were used more often in the past and organizations specifically, it is also necessary include the works of Hambrick (1984) and to clarify what is meant by a configuration Miller and Friesen (1978, 1980). However, and how configurational reasoning is reflected they have recently become less popular as in current research on organizations. A useful compared to typological approaches, which starting point is offered by Meyer et al., who are more aligned with an emphasis on theory define organizational configurations as ‘any testing. multidimensional constellation of conceptu- Regardless of whether configurations are ally distinct characteristics that commonly derived deductively or inductively, research- occur together’ (1993, p. 1175). Two things ers are usually interested in identifying the are particularly notable about this definition. specific constellation of relations between First, it is empirically oriented in that it the different parts that make up the orga- points to the presence of multiple instances nizational configuration (McPhee and Scott of a constellation, be they across entities Poole 2001). Usually, the internal ‘logic’ or time. Second, the definition emphasizes of such organizational configurations is one the co-occurrence of distinct characteristics, of consistency that can be achieved by a thus using commonality as a reference point, variety of mechanisms. For example, the but leaving open what this commonality is ‘fit’ characterizing configurations may be the based on. result of internal, adaptive learning about The definition of Meyer et al. (1993) can how the various elements of the organization thus be used with both typological and taxo- are best configured to achieve more efficient nomic approaches to understanding the nature outcomes. As such, configurations are most of configurations. Whereas both approaches likely to be observed where experimentation share the that a configuration is marked is encouraged and indeed feasible, where high by some characteristic emerging from the levels of interdependence between different constellation of its elements, they present organizational elements exist, and where this different ways of arriving at this characteristic. interdependence is marked by complementar- Typological approaches of configurations are ity, that is, if the interdependence is of such essentially reflective in that they conceive of a form that engaging in one type of activity configurations as containing an internal logic will increase returns from another one (Miller that exists independent of concrete instances 1990, Milgrom and Roberts 1995). A classic and is merely reflected in the empirical example of organizational complementarity manifestations. Accordingly, the typological is found in manufacturing plants, where approach is deductive and causality flows the flexibility of production equipment is from the construct to the empirical mani- related to the breadth of the product line. As festations; the empirical manifestations only Milgrom and Roberts (1995, p. 193) show, take on meaning because of the construct having more flexible production lines that (MacKenzie 2003). This approach has been can be easily switched over to a different used very successfully in research on organi- product makes it less costly (and thus more zational configurations, and examples include valuable) to produce many small batches of works by Delery and Doty (1996), Doty customized products that can be matched to et al. (1993) and Drazin and Van de Ven customer preferences. Conversely, a diverse (1985). By contrast, taxonomical approaches product portfolio increases the value of a to organizational configurations take a more flexible production line that does not rely on inductive, empirically based approach. Here, economies of scale and can be quickly shifted the construct is formative in that causality over, leading to less downtime while the line flows from the empirical manifestations to the is retooled. In practice, it will therefore be rare

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 419 415–431 420 THE SAGE HANDBOOK OF CASE-BASED METHODS

to see either flexible production equipment or may form around a shared identity rather than a diverse product portfolio without the other; efficient forms of organizing. Alternatively, both are complements in that each increases organizational configurations may reflect the value of the other; as organizational social logics of appropriateness that suggest components, they are ‘sticky.’ certain forms of organizing as associated with Complementary configurations frequently specific economic activities, logics that might are made up of multiple components com- furthermore exhibit considerable differences, monly found together. For instance, the mod- for instance across cultural contexts (e.g. Scott ern manufacturing process tends to be marked 1995). by a number of components such as flexible It is important to note, however, that machines, short production runs, highly the assumption of an internal logic to skilled workers, horizontal communications, configurations may frequently be too strong. and targeted markets, which tend to make For example, particular configurations of it quite different from the traditional, mass- circumstances may be the result of histori- production system that relied on specialized cal constellations following now discernable machinery, long production runs, low worker internal logic yet resulting in particular and skills, hierarchical controls, and mass markets identifiable effects. As such, functionally (Milgrom and Roberts 1995). Although the oriented arguments that configurations are examples I have given here come from the based on a logic of consistency may not be field of industrial manufacturing, complemen- warranted.As McPhee and Scott Poole (2001) tarities can, of course, relate to any number note, the idea that a configuration reflects of organizational characteristics, including the an underlying logic may to some extent classic ones of organizational structure and present a problem for configurational theories, environment where small, agile organizations and by extension also for the case-study are usually considered to perform better in approach. Specifically, ‘most configurational turbulent environments while large, lumber- theories are what Althusser (1972) called ing organizations tend to do better in stable ‘expressive totalities’ – they are supposed environments. to be consistent because each part reflects Alternatively, instead of learning and fit, the underlying logic of the whole’ (McPhee the logic may be based on an external, and Scott Poole 2001, p. 515). However, a environmental selection mechanism such as good theory should question the assumption organizational birth and death due to market of consistency, that is, the assumption that . Here, some kind of mechanisms all parts of the configuration are equally is needed to generate variation in organiza- necessary or important. For instance, one tional configurations, and efficiency pressures might alternatively conceive of configura- then operate to narrow down this variety into tions as consisting of a core and periphery, a smaller number of viable forms. where core elements are essential whereas Many of the arguments regarding orga- more peripheral elements are less important nizational configurations thus take on a and perhaps expendable. Furthermore, most functional logic in tying empirically observed research on configurations emphasizes the configurations to some form of fit based on internal consistency of the configuration’s pressures towards efficiency and consistency. underlying logic. Yet there may frequently be However, arguments relating to the emer- inconsistencies within configurations, where gence of configurations need not be restricted ‘a better configuration might balance off to efficiency-based responses to either internal conflicting logics, or list the necessary condi- consistency demands or external pressures on tions for success and make sure that the organizations. Organizational configurations configuration meets them’(McPhee and Scott may also be formed around sociocultural or Poole 2001, p. 515). political factors. For example, Peteraf and What emerges, then, is a picture of con- Shanley (1997) argue that strategic groups figurations as embedded in space and time

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 420 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 421

and involving varying levels of complexity, a situation of limited diversity by means dynamism, and analysis. Simple configu- of a truth table – an analytical tool for rations may involve only few and linear listing all possible combinations of causal interdependencies. In contrast, complex con- conditions. The truth table here lists four figurations may involve multiple interdepen- organizational characteristics (A, B, C, D) dencies that are furthermore characterized and one outcome (Z). As a truth table by interactions such as complementarity or uses binary values, there are sixteen pos- substitution effects leading to synergies and sible combinations here. However, not all trade-offs between the different elements. conceivable configurations of organizational Furthermore, configurations need not be characteristics also show empirical instances. static, but may be dynamically changing, In Table 24.1, combinations 7, 12, 14, and 15 suggesting that organizations follow dynamic show a question mark in the outcome column, constellations that change over their life indicating there may be no empirical instance cycles (e.g. Moores and Yuen 2001). Finally, of this combination, indicating a situation of configurations may be cutting across several limited diversity. levels of analysis. For example, organiza- Limited diversity stems from a number of tional configurations may involve elements reasons. First, as Stouffer (1941) points out for at the organizational, intra-organizational, and configurational approaches more generally, supra-organizational level. even relatively few elements can lead to While the number of conceivable organiza- an astronomically large number of different tional configurations is thus staggering, tax- possible complex dynamic configurations, onomic studies of organizations have shown so there will frequently be very few or a relatively small number of configurations no empirical instances of any particular account for a relatively large share of all configuration. The number of rows in a truth organizations in the samples studied (Miller table is calculated as 2k, with k indicating the 1990). This phenomenon, which is known as number of causal conditions. For Table 24.1, limited diversity (e.g. Ragin 1987), presents an the number of possible combinations is thus important issue for the study of organizational 24, or 16, but if one was to double the number configuration. Table 24.1 demonstrates such of causal conditions to 8, the number of possible configurations would jump to 256. Beyond the issues of manifested versus Table 24.1 Truth table exhibiting limited hypothetical configurations, Miller (1986) diversity points to three reasons why there should be relatively few kinds of organizational Organizational Characteristics Outcome configurations. First, competitive pressures Configuration A B C D Z from the environment are likely to weed 1 Yes Yes Yes Yes Yes out unsustainable models, an argument that 2 Yes Yes Yes No No connects to the external selection perspective. 3 Yes Yes No Yes Yes Second, organizations should be drawn to 4 Yes Yes No No Yes 5 Yes No Yes Yes No certain configurations that are internally 6 Yes No Yes No No harmonious and mutually reinforcing, an 7 Yes No No Yes ? argument that connects to an internal selection 8 Yes No No No No mechanism, usually based on experience of 9 No Yes Yes Yes Yes what works in any given context. And finally, 10 No Yes Yes No Yes 11 No Yes No Yes No Miller (1990) points out that organizational 12 No Yes No No ? change tends to be non-continuous and 13 No No Yes Yes No episodic, suggesting that hybrid forms are less 14 No No Yes No ? likely to be explored and that preference will 15 No No No Yes ? usually be given to ideal types presenting 16 No No No No Yes mutually exclusive positions.

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 421 415–431 422 THE SAGE HANDBOOK OF CASE-BASED METHODS

ORGANIZATIONS AND teams (e.g. Tsui and O’Reilly 1989) and even CONFIGURATIONS ACROSS individual-level configurations such as the fit LEVELS OF ANALYSIS between the individual and the organization (e.g. Chatman 1989). While configurations lie at the heart of the case-study approach, not all studies that exam- Organization level ine organizational configurations can properly be classified as case studies. Case studies and Aconsiderable number of case studies, as well studies of organizational configurations thus as studies of organizational configurations form partially overlapping sets. In the follow- more generally, have focused on the organi- ing, I examine studies from the union of those zation level, and particularly the relationships sets rather than the intersection, to provide an between strategy, structure, and processes. overview of the various forms of addressing Much of the literature on strategic groups organizational configurations either explicitly operates at this level (e.g. Dess and Davis or implicitly. Configurations can occur at the 1984, Cool and Schendel 1987, Ketchen et al. intra-organizational, organizational, and the 1997, McNamara et al., 2003). While many supra- and inter-organizational levels, as well of the studies at this level have employed as across such levels. Because of the extensive quantitative methods, there are also a number literature, it is necessary to note that the of case studies using mainly qualitative studies I use here are for the purpose of illus- methods, such as work on organizational tration rather than for providing an exhaustive activity systems (e.g. Siggelkow 2001, 2002) overview of work on configurational thinking or Chandler’s (1962) Strategy and Structure, in organization and management studies. which examined strategic and organizational change in the United States. Likewise, Philip Selznick’s (1949) T.V.A. and the Grass Roots is also largely located at this level of analysis, Intra-organization level although not exclusively. The study of intra-organizational phenomena has been at the focus of a number of the classic Supra-organization level case studies. Gouldner’s (1954) Patterns of Industrial Bureaucracy focuses almost A smaller number of studies have exam- exclusively on intra-organizational phenom- ined configurations at the inter-organizational ena as he described three configurations of level. Examples here include Bensaou and ‘mock,’ ‘representative,’ and ‘punishment- Venkatraman’s (1995) study of interorga- centered’ bureaucracy that emerge from rule nizational relationships in the automobile creation and enforcement. Similarly, Kanter’s industry, or Malhotra et al.’s (2005) research (1977) Men and Women of the Corporation on supply-chain partnership configurations. examines how the power structure of a large Child (2002) likewise uses a configurational corporation shapes both the behavior and approach to examine twenty cases of interna- personalities of its employees, while Dalton’s tional joint venture formation, whereas Dubbs earlier (1959) Men Who Manage likewise et al. (2004) examine configurations at the focused on issues of formal and informal organizational network and system level in power relations within corporations. More the care industry. At an even more recently, Kunda’s Engineering (1992) macro-level, research on Business Systems focused on the organizational culture in the and the Varieties of Capitalism approach have engineering department of a large American suggested that economic systems are likewise high-tech corporation. Other research has best understood from a configurational per- taken an even more micro-level approach, spective (e.g. Whitley 1999, Hall and Soskice focusing on configurations at the group level 2001) and analyzed using configurational such as the demographic composition of methods (e.g. Kogut and Ragin 2006).

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 422 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 423

Cross-level configurations organization, observation of life in organi- zations such as ethnography and participant While the studies I have discussed so far observation, the use of focus groups, or the mainly inhabit a single level of analysis, this examination of various archival records classification is not a very rigid one, and regarding an organization and life around it. effects located at different levels frequently Such qualitative methods for the analysis of impinge upon the configurations at the main organizations are grounded in the hermeneutic level of analysis. However, a number of stud- tradition of the humanities (e.g. Gephart ies have moved towards a truly multi-level 2004) and usually involve the researcher’s examination of configurations. For examples, inductive and interpretive treatment of the Crozier’s (1964) classic work on The Bureau- evidence (Van Maanen 1998). The flexi- cratic Phenomenon explicitly connects the bility of these methods makes them very bureaucratic control system to the cultural attractive for the case-study approach, as environment in which French administrative they can be applied where quantification is organizations are embedded. Similarly, the problematic or has to be deferred until a Miles and Snow (1978) typology of firms later point in the analysis. However, this as Defenders, Prospectors, Analyzers, and very flexibility also presents a challenge in Reactors explicitly theorizes configurations the generally more positivistically oriented of structure, strategy, and environment, field of organization and management studies. thus spanning the organization and supra- Although there are a number of guides as organizational levels. Extending the Miles to how qualitative research might proceed and Snow framework in a different direction, in a systematic manner (e.g. Miles and Moores and Yuen (2001) examine configura- Huberman 1984,Yin 1984, Van Maanen 1988, tions of strategy, structure, leadership style, King et al. 1994, Emerson et al. 1995), as and decision making style, thus combining well as advice on how to get this research characteristics at the organization and indi- published in top flight research journals (e.g. vidual levels. Furthermore, Greckhamer et al. Gephardt 2004, Suddaby 2006), the typical (2008) use QCA to examine how industry, case study still relies to a considerable extent corporate, and business-unit effects combine, on the persuasiveness of its narrative. While leading to truly multi-level analyses. good qualitative research is systematic and disciplined, there is are few agreed-on rules for drawing conclusions and verifying their METHODOLOGICAL CHALLENGES OF robustness (Miles and Huberman 1984). Due ORGANIZATIONAL CASE STUDIES to this lack of standardization the analysis still largely resembles an rather than a science, Because the case-study approach is marked and as a result ‘one cannot ordinarily follow by a configurational understanding of organi- how a researcher got from 3600 pages of field zational phenomena within a specific spatial notes to the final conclusions, sprinkled with and temporal context, it faces a particular vivid quotes though they may be’(Miles and challenge, namely analyzing and under- Huberman 1984, p. 16). standing complex interdependencies between However, even the most hermeneutic of various factors and causal conditions that in approaches has to start with observing differ- combination characterize the case in question. ences and similarities and likely with either To address this challenge, a variety of diff- counting their commonness or assigning some erent methodological approaches have been weight to their importance. At the most gen- employed, each of which offers certain advan- eral level, the actual analysis of the collected tages while also carrying certain liabilities. evidence will thus usually involve a search The most prevalent method for analyzing for similarities and dissimilarities within cases still relies on qualitative research meth- the data that eventually leads to empirical ods such as interviews with members of an generalizations, frequently along the lines of

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 423 415–431 424 THE SAGE HANDBOOK OF CASE-BASED METHODS

inquiry presented by John Stuart Mill in his subject to all sorts of cognitive biases, such as A System of Logic (1843), and particularly the tendency to search for information in a way as incorporated in the methods of agreement that confirms our preconceptions (e.g. Wason and difference (e.g. Stouffer 1941, Ragin 1960) or our propensity to neglect the base 1987). Consider, for example, Eisenhardt’s rate of events (e.g. Kahneman and Tversky (1989) influential work on how organizational 1973). The recognition of how quickly the researchers can employ case studies to organizational researcher is overwhelmed by build better theories. Eisenhardt describes the amounts of data and how difficult it two forms of analysis: that of within-case is to identify patterns and draw inferences data and the search for cross-case patterns. has led to an increased usage of qualitative Among these, the analysis of within-case data data analysis software packages that aim to presents the first step and is clearly the less assist the researcher in sorting, coding, and formalized approach. Eisenhardt suggests that analyzing the data their data. While these it is imperative for the researcher to become software packages facilitate the process of intimately familiar with the case as a stand- drawing inferences and tend to make it more alone entity, for example by writing case his- systematic, they present but a partial solution tories, examining transcripts, or by collecting to the underlying problem of dealing with and tabulating quantitative data on a variety the complexity that configurations of factors of relevant aspects. Whatever approach the present. researcher chooses, the goal of this process At the other end of the methodological is to ‘allow the unique pattern of each case to spectrum lie quantitative methods for identi- emerge before investigators push to genera- fying and examining configurations. Among lize patterns across cases’ (Eisenhardt 1989, these, cluster analysis is probably the most p. 540). Connected to within-case analysis is popular one for distinguishing configurations, the search for patterns across cases to identify and has enjoyed a certain revival in the similarities and differences. This analysis can recent literature on organizational configu- take various forms, such as looking for within- rations, particularly in the field of business group similarity and inter-group differences, studies (e.g. Corso et al. 2003, Uhl-Bien examining pairs of phenomena, or comparing and Maslyn 2003, Desarbo et al. 2005, the evidence by data sources. Regardless of Malhotra et al. 2005, Lim et al. 2006, Marlin which particular from is chosen at this time, et al. 2007). Cluster analysis is attractive however, the general approach remains the for the study of configurations because it search for agreement and difference within offers various algorithms for grouping cases pattern of the evidence. that share similar features into respective While the advantage of a purely qualitative, clusters. As an exploratory tool for the hermeneutic analysis of the evidence lies in analysis of quantitative data, cluster analysis its ability to provide insights that are difficult can be used to discover structures in the to achieve using quantitative, statistically data without specifying a priori what those oriented methods, the disadvantage of this structures might be. As a result, a number approach is equally evident and in large of researchers have used cluster analysis to part stems from the researcher’s cogni- examine organizational configurations across tive limitations. Purely qualitative analysis a variety of levels of analysis (e.g. Hambrick quickly exhausts the levels of complexity in 1983, Cool and Schendel 1987, Ketchen patterns it can process (e.g. Stouffer 1941), et al. 1993, Bensaou and Venkatraman 1995, particularly if the concepts of interest are Dubbs et al. 2004, Moores and Yuen 2001; graded rather than binary in nature. Even if for reviews, see Ketchen and Shook 1996, our cognitive capabilities would allow us to Ketchen et al. 1997). consider the exponentially growing number of However, although cluster analysis allows configurations that emerge quickly even from the discovery of configurations of charac- only a few binary concepts, we would still be teristics that commonly occur together, it

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 424 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 425

also has significant weaknesses. Perhaps most the various characteristics included remains importantly, the exploratory nature of cluster largely unexamined. Although it is possible to analysis makes it unsuitable for testing theory. use regression analysis with dummy variables While it would of course be possible to for different configurations to examine the hypothesize the existence of different types relationship between these configurations and of clusters before actually conducting the an outcome of interest, the issue remains that analysis, it is not possible to test these these configurations need to be identified first, hypotheses because there is currently no test and the usual methods such as cluster analysis statistic for cluster membership. Although show significant weaknesses here. Further- one might compare the results of different more, clustering combined with correlational clustering algorithms or assess performance analysis would still not allow a researcher differences between different clusters, the to examine the effect of different levels of basic issue remains that cluster analysis will the variables that are joined in the dummy always result in some clustering and there for cluster membership. Such an approach is not test statistic to guide the analysis. would also not be able to address issues The fact that even with for the most part of equifinality, that is, situations where ‘a randomly distributed data some kind of cluster system can reach the same final state, from solution is likely to emerge has led a number different initial conditions and by a variety of of researchers to question the existence of different paths’ (Katz and Kahn 1978, p. 30). true underlying configurations, suggesting Equifinality und thus equal effectiveness of that e.g. strategic groups may be merely different configurations presents an important statistical artifacts (e.g. Hatten and Hatten theoretical issue for configurational thinking 1987, Thomas and Venkatraman1988, Barney on organizations, and methods that cannot and Hoskisson 1990). address this issue thus do not adequately Another problematic aspect of cluster match up with the theory (Fiss 2007). analysis is that its assignment of cases to Similar critiques also apply to approaches clusters is based on all the characteris- using deviation scores to determine the fit tics included regardless of the relationship between a hypothesized ideal and empir- between these characteristics and outcomes ically observed configuration. While such of interest, such as performance, reputation, approaches tend to be deductive in nature and deviance, or some other construct. Consider thus are theoretically more attractive than the for example a situation where not all char- largely inductive cluster analysis, deviation acteristics included in the analysis are in score approaches hold related problems such fact causally important regarding the outcome as limited insight into the relationship between (a situation that is likely to be the rule rather the different characteristics of the configura- than the exception). Here, cluster analysis tion and a considerable sample dependence is insensitive to the fact that some cases in how profiles are derived (e.g. Drazin and may be identical regarding a few causally Van de Ven 1985), thus making them quite important characteristics, but may be different sensitive to errors in estimating the ‘ideal’ along a large number of characteristics that configuration and reliability issues (Gupta and are irrelevant. From a causal point of view, Govindarajan 1993). such cases belong into the same category The recognition that organizational con- since they share the same causally important figurations can be nested across multiple characteristics. However, cluster analysis levels of analysis has led some researchers would usually place them in different clusters towards statistical modeling techniques that because they differ on many (irrelevant) allow assessing such multi-level effects. In characteristics. While cluster analysis thus particular the study of educational orga- allows the researcher quickly to determine nizations has employed hierarchical linear configurations in a dataset, the nature of these modeling (HLM) to take into account that configurations and the relationships between the achievement of individual students also

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 425 415–431 426 THE SAGE HANDBOOK OF CASE-BASED METHODS

depends on effects at the school and state or more relevant factors. And finally, because level, thus leading to three-level hierarchies they estimate a single equation, regression- (e.g. Raudenbusch et al. 1999). Similarly, based models are problematic when the goal HLM can be used to examine multilevel is to examine equifinal outcomes, that is, effects of organizational work groups, depart- situations where there are several path to ments, organizations, and environments (e.g. an outcome of interest, thus blocking the Hoffman 1997, Hoffman and Gavin 1998). empirical investigation of equifinal configu- However, while HLM presents a better rations (e.g. Gresov and Drazin 1997, Fiss methodological fit for multilevel theories of 2007). One might consider using ANOVA or organizations and allow the testing of more regression with dummy variables to examine complex effects, such modeling still has the relationship between e.g. membership in significant shortcomings regarding a truly a configuration and performance, but this of configurational understanding of cases and course does not address the main issue of how organizations. configurations and the relationships between Specifically, as linear models, HLM equa- their different characteristics were identified tions are based on a number of assumptions in the first place. A possible approach to that stand in contrast to configurational estimating these relationships would be the thinking. For instance, while configurational use of log-linear models (e.g. Knoke and approaches point to nonlinear, synergistic Burke 1980), but these make no distinction effects that can lead to equifinal configura- between dependent and independent variables tions, the econometric model that underlies and thus cannot determine the direction of HLM and regression methods more broadly causality within an relationship. In addition, assumes linear, additive effects that are log-linear models are useful for categorical unifinal. While ordinary regression analysis but not continuous variables, making them estimates one average net effect across a again rather unwieldy tools. whole population, HLM presents an improve- Some researchers have aimed to combine ment as it allows the net effect to be separated qualitative and quantitative analysis as a into level-specific effects. However, HLM promising way to better capture the complex still treats different variables as competing nature of configurations that marks case in explaining the variation of the outcome studies. For example, Siggelkow’s (2002) of interest, with a focus on the unique study of the configurations of a large US contribution of each variable while holding corporation combined in-depth qualitative all other effects constant. In contrast, config- interviews and the analysis of archival records urational thinking emphasizes that one has with network analysis methods.After estimat- to consider how causes combine rather than ing qualitatively the existence and strength of compete to create outcomes, and that the goal relationships between the different activities should be estimating this very relationship that the corporation engaged in, Siggelkow between different parts rather than aiming to then used network measures to determine the statistically control for such effects. centrality or ‘coreness’ of various organiza- Interaction effects present one way to take tional elements as well as identify the various these relationships into account and include patterns in the evolution of the organization. them into the statistical model, but for all Similarly, Black and Boal (1994) suggest the practical purposes such effects have been use of network analysis to capture the complex largely restricted to two-way interactions, as interdependencies between firm resources. three way interactions are exceedingly hard Treating the configuration as a network of to interpret and rarely appear in published interdependent characteristics thus offers in- research. Such a limitation to interaction triguing possibilities for future case-study effects between only two causal factors places research. an undue burden on the modeling of config- Both purely qualitative and quantitative urational effects that can easily include four approaches thus have considerable difficulties

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 426 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 427

regarding the configurational nature of the the case-study approach, and the potential case study. Whereas qualitative approaches for significant methodological improvements are flexible, they quickly exhaust the levels along these traditional lines of inquiry seems of complexity they can handle in a rigorous limited at this time. However, due its con- way. Quantitative approaches allow for the figurational nature, set-theoretic methods in analysis of a large number of data points, general and QCA in particular offer a way but they either allow little insight into how to conduct case-study research that is both configurations emerge or are rooted in the methodologically rigorous and able to offer linear model that is frequently not useful new and different insights than traditional in examining configurational arguments. As methods. These methods can be used in a I have argued elsewhere (Fiss 2007), this dis- variety of ways. For example, they can be used connect between configurational arguments to analyze quantitative data on organizational and empirical methods in organization and structure, strategy, and the environment. How- management theory has emerged as signif- ever, set-theoretic methods can also be used icant hindrance to the further development to examine qualitative evidence contained of a configurational understanding of organi- in the narratives that commonly accompany zations. In a similar manner, the case-study case studies, allowing for a more rigorous approach, with its configurational nature, examination than is usually feasible with would likewise benefit from employing more purely qualitative approaches. Additionally, frequently a methodology better in line with QCA can be used as a meta-analysis tool to its assumptions of complex causal interdepen- examine case studies. An intriguing example dencies. Specifically, set-theoretic methods of this approach is offered by Hodson and such as QCA (e.g. Ragin 1987, 2000) provide Roscigno (2004), who combine a content an attractive alternative here. Rooted in the analysis of 204 organizational ethnographies comparative methods between qualitative and with QCA analysis to determine the causal variable-based approaches, QCA is deeply configurations that lead to organizational configurational in its understanding of how success such as employee involvement and causes combine to create outcomes, making competent management. Furthermore, while it a particularly useful tool for the case-study QCA can now be successfully applied to approach. QCA is able to handle considerable large-N analyses (e.g. Ragin and Fiss 2007), amounts of causal complexity while retaining it was originally designed to handle small-N the holistic quality of the phenomenon under situations, making it particularly attractive to study – two issues that are essential for the a case-study approach that requires consid- case-study approach. erable in-depth knowledge about cases and thus places certain constraints on the number of cases that can be explored in sufficient AN AGENDA FOR FUTURE RESEARCH detail. ON CONFIGURATIONAL PHENOMENA Beyond the use of tools that speak to the configurational nature of the case study, an The case study still presents one of the most important way forward lies with the study of attractive research strategies for understand- configurations across levels of analysis. While ing life in and around organizations. The a number of studies have already explored configurational nature of the case study in configurations reaching across the individual, particular aligns well with the configurational organizational, and supra-organizational lev- thinking that underlies much of management els, such research is still the exception rather and organization theory and is typical of the than the norm. Given the interconnectedness social world more broadly. As I have argued, of many organizational phenomena, much however, many of the current data analysis more research is needed along these lines. Fur- approaches used with a case study do not thermore, such analyses need not be restricted fully speak to this configurational nature of to the traditional constructs of e.g. strategy,

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 427 415–431 428 THE SAGE HANDBOOK OF CASE-BASED METHODS

structure, and environment. For example, Finally, the majority of prior research on Fombrun (1989) suggests three levels of con- organizational configurations has focused on straint that are likely to affect organizational static rather than dynamic configurations. configurations: the infrastructure of interde- Again, a case-study approach can contribute pendencies, the sociostructure of exchange here by allowing for a more dynamic under- relations, and the superstructure of symbolic standing of organizational configurations. representations. However, while most studies Tracking configurations over time is method- have so far focused on the infrastructure of ological challenging, but certainly not infeasi- largely economic and technological interde- ble and would significantly enhance our way pendencies, much remains to be explored of thinking about configurational phenomena regarding the sociopolitical and symbolic- in and around organizations. cultural side of configurations. The processes creating organizational configurations operate at many levels, including competitive and NOTES evolutionary processes, but also sociopolitical and cultural-symbolic ones, as suggested 1. For an example of this process, see Ragin’s by the institutional theory (e.g. Lounsbury (1992) analysis of Wieviorka’s (1992) ‘casing’ as and Ventresca 2002). Industry-level ‘recipes’ evidenced in treatment of terrorist groups as social about how to compete and what is successful movements. in organizing may be powerful scripts leading to specific organizational forms that either are configurations or lead to the formation REFERENCES of specific configurations because of ensuing economic and technical inderdependencies. Abbott, A. 1992 ‘What do cases do? Some notes Accordingly, a true understanding of orga- on activity in sociological analysis’ in C. Ragin and nizational configurations must go beyond H. Becker (Eds) What Is a Case? Exploring the merely technical interdependencies to include Foundations of Social Inquiry. Cambridge: Cambridge the interaction between these and sociopoliti- University Press, pp. 53–82. cal and cultural-symbolic factors. Allison, G.T. 1971 Essence of decision: Explaining the The goal of extending case-study ap- Cuban missile crisis. Boston: Little, Brown. proaches to examine phenomena across more Althusser, L.K. 1972 For Marx. London: NLB. levels of analysis and phenomena furthermore Barney, J.E. and Hoskisson, R.E. 1990 ‘Strategic points to the fact that many organizational groups: untested assertions and research proposals’ Managerial and Decision Economics 11: 187–198. phenomena are essentially constituted by Bensaou, M. and Venkatraman, N. 1995 ‘Configurations configurations of configurations. This phe- of interorganizational relationships: A comparison nomenon – where the whole takes the same between U.S. and Japanese automakers’ Manage- shape or form as its parts – is known as self- ment Science 41: 1471–1492. similarity in complexity theory. While some Black, J.A. and Boal, K. 1994 ‘Strategic resources: traits, researchers have argued that organization configurations and paths to sustainable competi- theory has much to gain from connecting more tive advantage’ Strategic Management Journal 15: closely to the insights of complexity theory 131–148. (e.g. Brown and Eisenhardt 1997, Levinthal Blau, P. 1955 The Dynamics of Bureaucracy. Chicago: 1997, Rivkin and Siggelkow 2003), so far Chicago University Press. these connections have not been explored Brown, S.L. and Eisenhardt, K.M. 1997 ‘The art of continuous change: linking complexity theory in detail. Furthermore, most of the research and time-paced evolution in relentlessly shifting in this vein has employed computational organizations’ Administrative Science Quarterly 42: modeling as a research strategy, suggesting 1–34. that much could be gained by connecting to Chandler, A.D. 1962 Strategy and Structure: Chapters more empirically oriented research such as the in the History of the Industrial Enterprise. Cambridge, case-study approach. MA: MIT Press.

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 428 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 429

Chatman, J.A. 1989 ‘Improving Interactional Organiza- Fombrun, C.J. 1989 ‘Convergent dynamics in the tional Research: A Model of Person-Organization Fit’ production of organizational configurations’ Journal Academy of Management Review 14: 333–439. of Management Studies 26: 439–458. Child, J. 2002 ‘A Configurational Analysis of Inter- Foreman, P.B. 1948 ‘The theory of case studies’ Social national Joint Ventures’ Organization Studies 23: Forces 26: 408–419. 781–815. Gephart, R.P. 2004 ‘Qualitative research and the Cool, K. and Schendel, D.E. 1987 ‘Strategic group academy of management journal’ Academy of formation and performance: The case of the US Management Journal 47: 454–462. pharmaceutical industry, 1963–1982’ Management Gerring, J. 2004 ‘What is a case study and what is it good Science 33: 1102–1124. for?’ American Political Science Review 98: 341–354. Corso, M., Martini, A., Pellegrini, L. and Paolucci, E. Gerring, J. 2007 Case Study Research: Principles and 2003 ‘Technological and organizational tools for Practices. Cambridge: Cambridge University Press. knowledge management: in search of configurations’ Gouldner, A.W. 1954 Patterns of Industrial Bureaucracy. Small Business Economics 21: 397–408. Glencoe, IL: Free Press. Crozier, M. 1964 The Bureaucratic Phenomenon. Greckhamer, T., Misangyi, V.F., Elms, H. and Lacey, R. London: Tavistock Publications. 2008 ‘Using qualitative comparative analysis in Dalton, Melville. 1959 Men Who Manage. New York: strategic management research: an examination of Wiley. combinations of industry, corporate, and business- Delery, J. E. and Doty, D.H. 1996 ‘Modes of theorizing unit effects’ Organizational Research Methods in strategic human resource management: tests Special Issue on Research Methods in Strategic of universalistic, contingency, and configurational Management. performance predictions’ Academy of Management Gresov, C. and Drazin, R. 1997 ‘Equifinality: Functional Journal 39: 802–835. equivalence in organization design’ Academy of Desarbo, W.S., Di Benedetto, C.A., Song, M. and Sinha, I. Management Review 22: 403–428. 2005 ‘Revisiting the miles and snow strategic frame- Gupta, A.K. and Govindarajan, V. 1993 ‘Methodological work: uncovering interrelationships between strategic issues in testing contingency theories: an assessment types, capabilities, environmental uncertainty, and of alternative approaches’ in Y. Ijiri (Ed.) Creative and firm performance’ Strategic Management Journal 26: Innovative Approaches to the Science of Manage- 47–74. ment. Westport, CT: Quorum Books, pp. 453–471. Dess, G.G. and Davis, P.S. 1984 ‘Porter’s (1980) Hall, P.A. and Soskice, D. (Eds) 2001 Varieties generic strategies as determinants of strategic of Capitalism: The Institutional Foundations of group membership and organizational performance’ Competitive Advantage. Oxford: Oxford University Academy of Management Journal 27: 467–488. Press. Doty, D.H., Glick, W.H. and Huber, G.P. 1993 ‘Fit, Hambrick, D.C. 1983 ‘An empirical typology of equifinality, and organizational effectiveness: a mature industrial-product environments’ Academy of test of two configurational theories’ Academy of Management Journal 26: 213–230. Management Journal 36: 1196–1250. Hambrick, D.C. 1984 ‘Taxonomic approaches to study- Drazin, R. and Van de Ven, A.H. 1985 ‘Alternative forms ing strategy: Some conceptual and methodological of fit in contingency theory’ Administrative Science issues’ Journal of Management 10: 27–41. Quarterly 30: 514–539. Hatten, K.J. and Hatten, M.L. 1987 ‘Strategic groups, Dubbs, N.L., Bazzoli, G.J., Shortell, S.M. and Kralovec, asymmetrical mobility barriers and contestability’ P.D. 2004 ‘Reexamining organizational configura- Strategic Management Journal 8: 329–342. tions: an update, validation, and expansion of the Hedström, P. and Swedberg, R. 1998 Social Mech- taxonomy of health networks and systems’ Health anisms: An Analytical Approach to Social Theory. Services Research 39: 207–220. Cambridge: Cambridge University Press. Eisenhardt, K.M. 1989 ‘Building theories from case Heise, D. 1989 ‘Modeling event structures’ Journal of study research’ Academy of Management Review 14: Mathematical Sociology 14: 139–169. 532–550. Hodson, R. and Roscigno, V.J. 2004 ‘Organizational Emerson, R.M., Fretz, R.I. and Shaw, L.L. 1995 success and worker dignity: complementary or Writing Ethnographic Fieldnotes. Chicago: University contradictory?’ American Journal of Sociology 110: of Chicago Press. 672–708. Fiss, P.C. 2007 ‘A set-theoretic approach to organi- Hoffman, D.A. 1997 ‘An overview of the logic and zational configurations’ Academy of Management rationale of hierarchical linear models’ Journal of Review 32: 1180–1198. Management 23: 723–744.

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 429 415–431 430 THE SAGE HANDBOOK OF CASE-BASED METHODS

Hoffman, D.A. and Gavin, M.B. 1998 ‘Centering gearing for partner-enabled market knowledge decisions in hierarchical linear models: implications creation’ MIS Quarterly 29: 145–187. for research in organizations’ Journal of Management Maoz, Z. 2002 ‘Case study methodology in international 24: 623–641. studies: from storytelling to hypothesis testing’ in F.P. Kahneman, D. and Tversky, A. 1973 ‘On the Harvey and M. Brecher (Eds) Evaluating Methodology of prediction’ Psychological Review 80: 237–251. in International Studies: Millennial Reflections on Kanter, R.M. 1977 Men and Women of the Corporation. International Studies. Ann Arbor, MI: University of New York: Basic Books. Michigan Press. Katz, D. and Kahn, R.L. 1978 The social psychology of Marlin, D., Ketchen, D.J. and Lamont, B. 2007 ‘Equifinal- organizations (second edition). New York: Wiley. ity and the strategic group-performance relationship’ Ketchen, D.J. and Shook, C.L. 1996 ‘The application of Journal of Managerial Issues 19: 208–232. cluster analysis in strategic management research: an Mauffette-Leender, L.A., Erskine, J.A. and Leender, M.R. analysis and critique’ Strategic Management Journal 2001 Learning with Cases (second edition). London, 17: 441–485. Ontario: Ivey Publishing. Ketchen, D.J., Thomas, J.B. and Snow, C.C. 1993 McNamara, G., Deephouse, D.L. and Luce, R.A. ‘Organizational configurations and performance: a 2003 ‘Competitive positioning witching and across comparison of theoretical approaches’ Academy of strategic groups: the performance of core, subsidiary, Management Journal 36: 1278–1313. and solitary firms’ Strategic Management Journal 24: Ketchen, D.J., Combs, J.G., Russel, C.J., Shook, C., 161–181. Dean, M.A., Runge, J., Lohrke, F., Naumann, S., McPhee, R.D. and Poole, M.S. 2001 ‘Organizational Haptonstahl, D.E., Baker, R., Beckstein, B.A., structures and configurations’ in F. Jablin and Handlers, C., Honig, H. and Lamoureux, S. 1997 L. Putnam (Eds) The New Handbook of Organizational ‘Organizational configurations and performance: a Communication: Advances in Theory, Research, and meta-analysis’ Academy of Management Journal 40: Methods. Thousand Oaks, CA: Sage, pp. 503–543. 223–240. Meyer, A.D., Tsui, A.S. and Hinings, C.R. 1993 ‘Con- King, G., Keohane, R.O. and Verba, S. 1994 Designing figurational approaches to organizational analysis’ Social Inquiry: Scientific Inference in Qualitative Academy of Management Journal 36: 1175–1195. Research. Princeton, NJ: Princeton University Press. Miles, M.B. and Huberman, M. 1984 Qualitative Data Knoke, D. and Burke, P.J. 1980 Log-Linear Models. Analysis: A Sourcebook of New Methods. Beverly Beverly Hills, CA: Sage. Hills, CA: Sage. Kogut, B. and Ragin, C.C. 2006 ‘Exploring complexity Miles, R.E. and Snow, C.C. 1978 Organization strategy, when diversity is limited: institutional complementar- structure, and process. New York: McGraw-Hill. ity in theories of rule of law and national systems Milgrom, P. and Roberts, J. 1995 ‘Complementarities revisited’ European Management Review 3: 44–59. and fit: strategy, structure, and organizational Kunda, G. 1992 Engineering Culture: Control and Com- change in manufacturing.’ Journal of Accounting and mitment in a High-Tech Corporation. Philadelphia: Economics 19: 179–208. Temple University Press. Mill, J.S. 1843/2002 A System of Logic: Ratiocinative Levinthal, D. 1997 ‘Adaptation on rugged landscapes’ and Inductive. Honolulu, HI: University Press of the Management Science 43: 34–950. Pacific. Lim, L.K.S., Acito, F. and Rusetski, A. 2006 ‘Development Miller, D. 1986 ‘Configurations of strategy and structure: of archetypes of international marketing strategy’ a synthesis’ Strategic Management Journal 7: Journal of International Business Studies 37: 233–249. 499–524. Miller, D. 1990 ‘Organizational configurations: cohe- Lounsbury, M. and Ventresca, M.J. 2002 ‘Social sion, change, and prediction’ Human Relations 43: structure and organizations revisited.’ Research in the 771–789. Sociology of Organizations 19: 3–36. Miller, D. and Friesen, P.H. 1978 ‘Archetypes of strategy Lundberg, G.A. 1941 ‘Case-studies vs. statistical formulation’ Management Science 24: 921–933. methods: an issue based on misunderstanding’ Miller, D. and Friesen, P.H. 1980 ‘Archetypes of Sociometry 4: 379–383. organizational transition’ Administrative Science MacKenzie, S.B. 2003 ‘The dangers of poor construct Quarterly 25: 268–299. conceptualization’ Journal of the Academy of Moores, K. and Yuen, S. 2001 ‘Management accounting Marketing Science 31: 323–326. systems and organizational configurations: a life-cycle Malhotra, A., Gosain, S. and El Sawy, O.A. 2005 perspective’ Accounting, Organizations, and ‘Absorptive capacity configurations in supply chains: 26: 351–389.

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 430 415–431 CASE STUDIES AND THE CONFIGURATIONAL ANALYSIS OF ORGANIZATIONAL PHENOMENA 431

Peteraf, M. and Shanley, M. 1997 ‘Getting to know Siggelkow, N. 2002 ‘Evolution towards fit’ Administra- you: a theory of strategic group identity’ Strategic tive Science Quarterly 47: 25–159. Management Journal Summer Special Issue 18: Stouffer, S.A. 1941 ‘Notes on the case-study and the 165–186. unique case’ Sociometry 4: 349–357. Pettigrew, A. 1990 ‘Longitudinal field research on Suddaby, R. 2006 ‘What grounded theory is not’ change: theory and practice’ Organization Science 1: Academy of Management Journal 49: 633–642. 267–292. Thomas, H. and Venkatraman, N. 1988 ‘Research on Ragin, C.C. 1987 The Comparative Method: Moving strategic groups: and prognosis’ Journal of Beyond Qualitative and Quantitative Strategies. Management Studies 25: 537–555. Berkeley, CA: University of California Press. Tsui, A.S. and O’Reilly, C.A. 1989 ‘Beyond simple Ragin, C.C. 1992a ‘Introduction: cases of “what is a demographic effects: the importance of relational case?”’ in C. Ragin and H. Becker (Eds) What Is a demography in superior-subordinate dyads’ Academy Case? Exploring the Foundations of Social Inquiry. of Management Journal 32: 402–423. Cambridge: Cambridge University Press, pp. 1–17. Uhl-Bien, M. and Maslyn, J.M. 2003 ‘Reciprocity Ragin, C.C. 1992b ‘“Casing” and the process of social in manager-subordinate relationships: components, inquiry’ in C. Ragin and H. Becker (Eds) What configurations, and outcomes’ Journal of Manage- Is a Case? Exploring the Foundations of Social ment 29: 511–532. Inquiry. Cambridge: Cambridge University Press, Van Maanen, J. 1988 Tales of the Field: On Writing pp. 217–226. Ethnography. Chicago: University of Chicago Press. Ragin, C.C. 1994 Constructing Social Research: The Van Maanen, J. 1998 ‘Different strokes: qualitative Unity and Diversity of Method. Thousand Oaks, CA: research in the administrative science quarterly from Pine Forge Press. 1956–1996’ in J. Van Maanen (Ed.) Qualitative Ragin, C.C. 2000 Fuzzy Set Social Science. Chicago: Studies of Organizations. Thousand Oakes: CA: Sage, University of Chicago Press. pp. ix–xxxii. Ragin, C.C. and Becker, H.S. (Eds) 1992 What Is a Walton, J. 1992 ‘Making the theoretical case’ in C. Ragin Case? Exploring the Foundations of Social Inquiry. and H. Becker (Eds) What Is a Case? Exploring the Cambridge: Cambridge University Press. Foundations of Social Inquiry. Cambridge: Cambridge Ragin, C.C. and Fiss, P.C. 2007 Fuzzy Set Policy Analysis. University Press, pp. 121–137. Manuscript. University of Arizona, Tucson, Arizona. Wason, P.C. 1960 ‘On the failure to eliminate Raudenbush, S.W., Fotiu, R.P. and Cheong, Y.F. 1999 hypotheses in a conceptual task’ Quarterly Journal ‘Synthesizing results from the trial state assessment’ of Experimental Psychology 12: 129–140. Journal of Educational and Behavioral Statistics 24(4): Wieviorka, M. 1992 ‘Case studies: history or soci- 413–438. ology?’ in C. Ragin and H. Becker (Eds) What Rivkin, J. and Siggelkow, N. 2003 ‘Balancing search and Is a Case? Exploring the Foundations of Social stability: interdependencies among elements of organi- Inquiry. Cambridge: Cambridge University Press, zational design’ Management Science 49: 290–311. pp. 159–172. Scott, W.R. 1995 and Organizations. Whitley, R. 1999 Divergent Capitalisms: The Social Thousand Oaks, CA: Sage. Structuring and Change of Business Systems. Oxford: Selznick, P. 1949 T.V.A. and the Grass Roots. Berkeley, Oxford University Press. CA: University of California Press. Yin, R.K. 1981 ‘The case study crisis: some answers’ Siggelkow, N. 2001 ‘Change in the presence of fit: the Administrative Science Quarterly 26: 58–65. rise, the fall, and the of Liz Claiborne’ Yin, R. 1984 Case Study Research: Design and Methods. Academy of Management Journal 49: 838–857. Thousand Oaks: CA: Sage.

[12:25 24/11/2008 5247-Byrne-Ch24.tex] Paper: a4 Job No: 5247 Byrne: The SAGE Handbook of Case-Based Methods Page: 431 415–431