Configuration research in strategic management: Key issues and suggestions

Journalof Management 1993. Vol. 19, No. 4. 77%795

Configuration Research in Strategic Management: Key Issues and Suggestions Gregory G. Dess The University of Texas at Arlington

Stephanie Newport Austin Peay State University

Abdul M. A. Rasheed The University of Texas at Arlington This paper discusses major theoretical and methodological issues that strategic management researchers must consider when developing and testing configuration theories. The theoretical issues include: (1) number of domains, (2) causality, and (3) temporal stability. The methodological issues are: (I) specification of key constructs, (2) effects of data aggregation, (3) the choice of unit of analysis, and (4) the appropriateness of research methodologies. Greater attention to these issues should result in more accurate findings and more meaningful interpretations.

The multidimensionality of constructs used to describe strategy phenomena has always posed a challenge for researchers. These challenges include not only developing meaningful constructs and relationships but also measuring them with a high degree of validity and reliability. Valid and reliable measurement is dependent on a unique, theory-based quantification of key factors such that each operationalization accurately reflects the phenomenon which it is supposed to measure (i.e., validity) and yields consistent results across multiple measurements (i.e., reliability) (Carmines & Zeller, 1979; Venkatraman & Grant, 1986; Chrisman, Hofer, & Bolton, 1988). The relationships among various constructs of interest to strategic management researchers have often been presented as gestalts, configurations, or archetypes. The widespread acceptance and increasing use of configurations in both normative and descriptive strategy research suggests a need to explore issues relevant to their appropriateness. (For a recent review of the basic premises and a historical overview of configuration research, refer to Mintzberg, 1990). A configuration represents a number of specific and separate attributes Direct all correspondence to: Gregory G. Dess, The University of Texas at Arlington, Administration, Department of Management, Arlington, TX 76019.

Copyright 0 1993 by JAI Press Inc. 0149-2063 775

College of Business

776

DESS, NEWPORT. AND RASHEED

which are meaningful collectively rather than individually (Rosenberg, 1968; Miller, 1987b). Configurations are finite in number and represent a unique, tightly integrated, and therefore relatively long-lived set of dynamics (Mintzberg, 1973; Miller & Mintzberg, 1983; Miller, 198 1; 1986; 1987b). The use of configurations in studies of organizations allows researchers to express complicated and interrelated relationships among many variables without resorting to artificial oversimplification of the phenomenon of interest. Configurations are a means of achieving parsimony while presenting rich, complex descriptions of organizations (Hambrick, 1983b; Miller & Friesen, 1977; 1984; Mintzberg, 1973; 1978). An important role of configuration research is the classification of organizations. By aggregating and organizing a large body of facts and data into a meaningful set, propositions and theories may be developed. Also, classification provides a means for further developing organization science to help ascertain and explain why some organizations are relatively unique and some have strong commonalities with others (McKelvey, 1982). Thus, classification systems provide a means for defining sets of homogeneous organizations which should significantly increase levels of explained variance of key variables across organizations (see Ulrich & McKelvey, 1990, for a recent discussion). Also, the determination and explanation of performance differences among organizations enhances the development of normative theory-traditionally a salient goal of strategic management research (Schendel & Hofer, 1979). For our purposes, we view the term “configuration” to be synonymous with both “gestalt” and “archetype.” This term has also been used to represent both taxonomies and typologies (Miller, 1986; 1987a; 1987b; Miller & Friesen, 1978). Distinctions between taxonomies and typologies are often blurred or dismissed in the literature (Hambrick, 1983a; Miller, 1987b; 1987~; 1988). However, we believe that there is a need to clarify exactly what constitutes a configuration. Are we to consider any classification scheme a configuration? If a typology constitutes a configuration and a typology can be defined as “any classification scheme” (Hambrick, 1983a, p. 214), then that is the only conclusion to be drawn. We suggest that the terms taxonomy and typology be used in accordance with their intrinsic meanings. That is, typology should refer only to theoretical/ conceptual classification schemes. Taxonomy, on the other hand, should refer only to empirically derived classifications. Further, the term “configuration” should be distinguished from taxonomy and typology. A typology or taxonomy contains elements or items that represent a single domain or an aspect of organizations, such as environment, structure, or strategy. Porter’s (1980) generic strategies of differentiation, overall low cost, and focus, for example, constitute a typology. A configuration contains relationships among elements or items representing multiple domains. Miles and Snow’s (1978) classification of organizations into defenders, prospectors, analyzers, and reactors uses multiple domains (strategy, structure, process, environment) and, therefore, clearly represents configurations. Miller (1987~) provides theoretical support for the existence of configurations based on what he terms the “imperatives’‘-environment, structure, leadership, strategy-that give rise to configurations. He argued that configuraJOURNAL OF MANAGEMENT.

VOL. 19. NO. 4. 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

777

tions exhibit great stability because of their “internal logic, integrity, and evolutionary momentum” (p. 697). Earlier work by Miller and his colleagues has empirically demonstrated various configurations or archetypes (e.g., Miller & Friesen, 1978). Chandler’s (1962) work can be viewed as involving the identification of configurations encompassing strategy and structure through rich, historical analysis. Yet another approach to the identification of configurations is suggested by Hinings and Greenwood (1988a). They view configurations as containing elements of both organizational structure and processes “which are strongly underpinned by provinces of meaning and interpretive schemes which bind them together in an institutionally derived normative order” (p. 295). The purpose of this article is to address what we believe is an important antecedent step in analyzing configurations. In the next two sections, we will address theoretical and methodological issues which must be considered by researchers who use configurations to develop normative and descriptive theory. Although, for convenience and organization these two types of issues will be discussed separately, we concur with Martin (1982) in recognizing their interdependence. Martin, drawing on Cohen, March, and Olsen’s (1972) seminal model of organizational decision making, contrasted the rational and garbage can models of the research process. While the rational normative model assumes, in part, that “the nature of the theoretical problem should determine the choice of the most appropriate research method” (p. 18), the garbage can model serves as a more accurate depiction of the actual research process. Here, theoretical problems, resources, methodologies, and the results of the research process are highly interdependent and mutually reinforcing. Theoretical Issues in Configuration

Research

Configuration researchers must select not only among multiple domains but also within each domain. They are further confronted with choosing among multiple conceptual schemes. Thus, it is particularly important to have a clear understanding of the conceptual building blocks of a meaningful configuration. Three issues that are especially relevant are: (1) number of domains: single versus multiple, (2) temporal stability of configurations, and (3) issues of causality in configurations. Number of Domains: Single versus Multiple An important issue in the development and further refinement of configuration research and theory building is the determination of the number of domains to be addressed. A majority of the strategic management research literature directed at developing typologies, taxonomies and configurations has used the conceptual domains or imperatives of strategy, structure, and environment. (See Miller [ 1987~1 for a recent review.) While some have focused their efforts on a single domain, others have included multiple domains. As noted by Weick (1979), the research process involves inevitable tradeoffs among generalizability, accuracy, and simplicity. In developing and testing theory on configurations, as one moves across a continuum from a single domain, to two JOURNAL OF MANAGEMENT. VOL. 19.NO. 4. 1993

118

DESS, NEWPORT, AND RASHEED

domains, to multiple domains, more comprehensive and accurate depictions of organizational reality become possible. Also, one gains more confidence in positing normative implications of empirical findings. However, such “gains” in accuracy come at an inevitable cost in parsimony. Thus, configurations encompassing multiple domains necessitate “a smaller number of more encompassing conceptual categories with a broader range of generalizability” (Bourgeois, 1980b, p. 29). Although Bourgeois was addressing the construct of the product life cycle and its implications for business-level strategies, his point is germane to our argument: as the number of dimensions of a construct increases arithmetically, the number of combinations increases geometrically. Blalock (1969, p. 34), in his discussion of typology construction, argues that the theorist is forced to simplify in two ways: (1) restrict each variable to a dichotomy, and (2) assume a priori that certain combinations of variables do not exist in a sufficient number to merit inclusion. Developing and supporting such heuristics, of course, results in additional demands for theory construction. Thus, a benefit in single domain research is that one is better able to consider multiple levels of the construct of interest. An example is suggested by Bourgeois (1985), drawing on Hofer’s (1975) extension of the product life cycle concept. Since Hofer included substantially more constructs from multiple domains than earlier discussion of the concept, Bourgeois contends that Hofer had to conclude with only “the bare beginnings of an inventory of hypotheses that posit the ‘optimal’ business strategy for a given configuration of contingencies” (1985, p, 589). Therefore, if fewer domains are included in a configuration, the researcher is able to include more levels in each domain and still obtain a comparable level of parsimony. As stated earlier, research within a single domain cannot strictly be described as configurational in nature. However, such research provides the necessary building blocks for more complex and accurate depictions of organizational phenomena. They provide the means to create order out of chaos-that is, to address the question: how do variables and constructs relate to each other to help form meaningful wholes? Perhaps the framework that has fostered the most theoretical and empirical work in recent years has been Porter’s (1980) typology of three generic strategies: differentiation, overall low cost, and focus. For clarity and illustrative purposes, our discussion in this section will center on this seminal work. Many have argued that Porter’s typology is the most complete and theoretically sophisticated representation of the business-level strategy construct (White, 1986; Miller, 1986). A great deal of effort has been directed at further theoretical clarification (e.g., Chrisman, Hofer, & Boulton, 1988; Jones & Butler, 1988; Murray, 1988; Hill, 1988) and empirical analysis (e.g., Dess & Davis, 1984; Hawes & Crittenden, 1984; Kim & Lim, 1988; Miller & Dess, 1993; White, 1986; Wright, Kroll, Tu, & Helms, 1991) since the publication of Michael Porter’s original framework in 1980. Key research issues in such single domain studies typically center on combinations of or mutual exclusivity of Porter’s strategies and their relationships to performance. Porter’s framework has also been criticized for its emphasis on economic factors over political factors, its bias towards big business which limits JOURNAL

OF MANAGEMENT,

VOL. 19, NO. 4. 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

719

generalizability, and its narrow focus in terms of choice of a generic strategy in a generic environment (Mintzberg, 1990). Many have expressed dissatisfaction with universal propositions or “quick fixes” (cf. Hitt & Ireland, 1987; Jaedicke, 1989). In part, two domain research seeks to ameliorate such concerns by examining bivariate relationships. They also serve as a first step in discovering potential cause-effect relationships between variables of interest. One of the major advantages of research and theory building encompassing two domains is the potential to ascertain the performance implications of contingencies or “fits” among key constructs (Venkatraman & Camillus, 1984; Schoonhoven, 1981). Recent theoretical and methodological advances in conceptualizing and measuring fit (Venkatraman, 1989) can prove to be very helpful to researchers in choosing a perspective of fit that is appropriate for a specific research question. Although investigations of bivariate contingency relationships constitute an improvement over single domain research, they are often criticized for being too general and simplistic. The restriction of interest to two domains leads the researcher to leave out many relevant parameters, thereby failing to capture the essence of configuration. Also, the choice of just two domains from a multiplicity of domains is problematic since it requires the researcher to prioritize the domains to be studied. Much of the recent work using Porter’s framework is directed at exploring linkages between the generic strategies, environment, and organization structure. Exemplary contributions in this endeavor include Miller (1988) and Kim and Lim (1988). Miller (1988) studied strategy-environment and strategy-structure linkages. He found that successful firms pursuing a strategy of differentiation were more likely to compete in unstable environments and successful cost leaders tended to compete in stable environments. Further, he found that successful differentiators made extensive use of technocrats and liaison devices whereas successful cost leaders preferred formal control mechanisms. Kim and Lim (1988) explored the relationship between environmental dimensions and Porter’s generic strategies. They found that high-performing differentiators and high-performing cost leaders were more likely to compete in different types of environments. Multiple domain configuration studies enable researchers to examine complex multivariate relationships and establish the relative importance of fit among constructs in multiple domains. Along with such accuracy in the depiction of organizational phenomena comes a greater potential for findings that have normative implications. We concur with Lenz’s (1980) conclusion from his multivariate study of fifty savings and loan associations: “neither environment, strategy nor organization structure is sufficient to explain differences in performance ... organizational performance is determined, in part, by the particular coalignment administrators are able to achieve” (pp. 220-221). Given our earlier discussion of the presence, mutual exclusivity, combinations, and performance implications of Porter’s generic strategies, the challenge for theory building is self-evident. The demands are further intensified as one moves toward an exploration of multiple domains. An important beginning in this direction is Miller, Droge, and Toulouse’s (1988) study JOURNAL OF MANAGEMENT, VOL. 19.NO. 4. 1993

DESK NEWPORT, AND RASHEED

780

Table 1.

Single v. Multiple Domains: Advantages & Limitations

Illustrative Studies

Benefits

Limitations

Single Domain

Dess & Davis, 1984 White, 1986 Hambrick, 1983

- May use many levels of the construct of interest Helps to identify “building blocks” for multiple domain studies - Methodologically simple

May mask interactive relationships Little normative value May explain relatively little of the variance of the dependent variable.

TWO Domains

Kim& 1988

- Allows examination of bivariate relationships -Often provides first step to understanding cause-effect relationships Performance implications of fit between two constructs may be examined

May omit relevant parameters, i.e., essence of configuration Difficulty in prioritizing the domains to be studied may be - Bivariable contigency relationships too general and simplistic

Multiple Domains

Miller, Droge, & Toulouse. 1988 Miller, 1988

Greatly enhances prescriptive relevance Allows examination of multivariate relationships -Can lead to establishing relative importance of fit among multiple constructs _

- Aggregation Problems Methodologically complex May not be practical to consider several different levels of each of the constructs/ domains

Lim,

which found that strategy making process and strategy content mediate the relationship between organizational context and structure. Table 1 presents a summary of the major advantages and disadvantages associated with the choice of number of domains. Issues of Causality in Configurations Causation among the variables in a configuration must be approached more carefully than in studies that investigate bivariate relationships. Establishing causality requires that a specific temporal sequence be present-that is, the independent variable must precede the dependent variable. Additionally, the independent variable must be clearly beyond the influence of the dependent variablethat is, it must be permanent, or fixed (see Rosenberg, 1968, pp. 9-13). Typically, however, in the context of configuration research, such causal relationships among multiple variables are stipulated as reciprocal and mutually reinforcing. The strategy-structure relationship serves to illustrate difficulties inherent in espousing causal relationships. Much of the early work in strategic management (e.g., Chandler, 1962; Rumelt, 1974) posited a causal relationship between strategy and structure, that is, changes in a firm’s strategy (e.g., diversification) led to changes in a firm’s internal structure (e.g., decentralized, divisional organizational forms). Recently, several writers (Burgelman, 1983; Fredrickson, 1986; Hall & Saias, 1980) have argued that relationships between strategy and structure may also be characterized by reverse or mutual causality. The many rules associated with a high level of formalization, for example, may inhibit a firm’s ability to implement successfully a strategy characterized by innovation and creative activity. Also, a highly centralized organization may experience difficulty when diversifying into new product-market areas that require rapid responses to changes in the environment. When additional domains are added to the research questiona common attribute of configuration research-the difficulty in establishing causal JOURNAL

OF MANAGEMENT,

VOL.

19, NO. 4. 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

781

relationships is exacerbated. For example, Miller posited that environmental “uncertainty prompts the adoption of organic, decentralized, differentiated, and integrated structures and innovative and flexible strategies” (1987c, p. 688). Here, he was arguing for an environmental imperative, i.e., the need for organizations to adapt to their environments. However, one could argue that organic, decentralized, and differentiated structures and innovative strategies may lead to (i.e., cause) an increase in executives’ perceptions of environmental uncertainty. Such perceived uncertainty may be a result of the corresponding increase in boundary spanning activity and information processing that tends to be associated with such structures andstrategies. Thus, structures and strategies may serve to influence environmental perceptions. With respect to causality, two common problems that seem to characterize much of the organization research in general and configuration research in particular are (1) assuming a causal relationship, unidirectional or mutual, in the absence of adequate theoretical support for such an assumption, and (2) inferring causality even when the research design does not directly permit such an inference. The issue of causality is particularly important since there is often a tendency to advance prescriptive implications based upon the assumption of causality, when neither theory nor results of research justify such inferences (Rosenberg, 1968). In the investigation of any complex system of relationships, the concepts of antecedents and consequences are largely a matter of the researcher’s frame of reference and the focus of a particular research project (see Newcomb & Bentler, 1988, especially Chapter 2). For example: Is performance a cause or an efSect of a given configuration of strategy, structure, and environmental variables? If performance is found to be a cause instead of an outcome of a given configuration, there is clearly little to be provided in the way of normative guidance for practicing managers. Normative theory building is enhanced when the causal relationship between the performance construct and other constructs under consideration is supported. Findings of reverse or mutual causality will provide insight into the relevance of the “environmental determinism” versus “strategic choice” paradigms (c.f., Bourgeois, 1984) that is central to the strategic management discipline. Inferences of causality would also require one to move from the collection of crosssectional to longitudinal data and from correlational to causal research designs. This issue is discussed in greater detail under “Appropriateness of Research Methodologies.” Temporal Stability of Configurations Miller (1987~) suggests that configurations, because of the enduring themes that unify and organize them, are characterized by considerable temporal stability. In order to cause a change in a configuration, a ‘revolution’ would be necessary. Such a revolution may be triggered by a change in any one of four ‘imperatives’ that give rise to the prevailing configuration. Miller (1987~) suggests that these four imperatives include environment, organization structure, leadership, and strategy. For example, Zajac and Shortell (1989), using the Miles and Snow (1978) framework, studied how 570 hospitals responded to a dramatic environmental JOURNAL OF MANAGEMENT, VOL. 19,NO. 4, 1993

182

DESS, NEWPORT, AND RASHEED

event in 1983-the U.S. Government’s introduction of the Medicare Prospective Payment System (MPPS). The MPPS eliminated reimbursement to hospitals on a “cost plus” basis and forced hospitals to become much more cost effective and responsive to market conditions. Accordingly, they found that the majority of hospitals changed their strategic configurations-typically from that of a defender to that of a prospector or analyzer. Unsatisfactory or declining performance outcomes often force the organization to consider seriously major changes in strategy, structure, and leadership. However, such changes often prove to be difficult to implement due to strong inertial forces prevailing within the organization. Ginsberg (1988) points out that enduring external conditions (such as industry structure and stakeholder values) and enduring internal conditions (such as organizational belief systems and structure) represent significant resistance to change. In some cases, organizations might prefer a change but lack the resources and competencies required to accomplish it. Thus, over a period of time the incremental changes that the organization adopts become inadequate to maintain its alignment with external changes. This results in a gradually increasing ‘strategic drift’ (Johnson, 1988) which can be corrected only by non-incremental, revolutionary changes. This is consistent with several studies which found that organizational life is characterized by alternating short periods of “reorientation” (Tushman & Romanelli, 1985) or “revolution” (Miller, 1982) and relatively long periods of “convergence” (Tushman & Romanelli, 1985) or “evolution” (Miller, 1982). Miller (1987~) points out that organizations can and do change even when they are primarily driven by a single imperative. These changes, however, are normally in conformity with the original theme of the configuration. Using terminology suggested by Ginsberg (1988), we may describe changes within a configuration as changes in ‘magnitude’ whereas a change from one particular configuration to another would be described as a change in ‘pattern’. Thus, one of the major issues that needs to be addressed while studying the stability of configurations is: How do we develop and validate operationalizable criteria to distinguish between changes that occur within a configuration as a part of normal adaptive behavior and changes that represent a transition from one configuration to another? Snow and Hamb~ck (1980) provide some insightful criteria. They suggest that strategic change occurs when an organization (1) undergoes a major change in its alignment with its environment and (2) substantial alterations in its internal features-technology, structure, and process-take place to fit its new alignment. Configurations are relatively stable (Miller & Mintzberg, 1983), but this stability is not to be viewed as static equilibrium. A necessary first step in understanding the evolutionary dynamics of configurations is the identification of external and internal forces or imperatives that disturb the stability of a configuration and cause it to change. This needs to be followed by a corresponding identi~cation of external and internal factors that resist change. At any given point in time, a configuration experiences both forces for change and resistance to change. However, most of the time, the inertial forces tend to be stronger and the configuration exhibits stability. Occasionally, either the forces for change become JOURNAL

OF MANAGEMENT,

VOL. 19, NO. 4, 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

783

too strong or resistance to change becomes comparatively weak. Under such conditions, the organization can bring itself back to stability only through a change in the configuration. Since it is theoretically possible to have more than one successful organizational configuration, even within an industry, an interesting research issue would be: Are certain types of transitions easier for organizations to accomplish than others and, if so, what are they? In other words, longitudinal studies may reveal certain patterns or favored paths that organizations follow as part of their evolutionary dynamics. Currently, research that has tracked patterns of change in configurations is limited (Hinings & Greenwood, 1988b). Most studies rely largely on archival data or published case histories in order to collect sufficient data over a ten or twenty year period (Miller & Friesen, 1980; Eisenhardt & Schoonhoven, 1990). There has been substantial theoretical development in this area (Greiner, 1972; Tushman, Newman, & Romanelli, 1986; Miller, 1987~). However, empirical investigation of the causes and consequences of change in configurations has lagged the development of theory due to difficulties in data collection and analysis. Configuration research requires data on a variety of variables, all of which are not available from published data sources. It is possible to collect cross-sectional data on structure and process variables through survey methods, but to do this over a period of ten or twenty years is extremely difficult. Some of these problems can be partially overcome by combining both survey and secondary data sources to understand what causes configurations to change over time as illustrated by Zajac and Shortell (1989). Their study of hospitals found that configurational change is influenced not only by environmental change but also by the organization’s prior configuration. Summary Table 2 contains an illustrative listing of major theoretical questions to consider in order to improve configuration research. These questions are arranged by the three sets of issues that we have discussed in this section. Methodological Issues in Configuration Research Research on configurations is primarily inductive in nature-a process of observation and description. In the pursuit of discovery, configurations are a valuable tool insofar as they offer a way of modeling the complexity inherent in strategy phenomena. In order to advance research and theory development, researchers must address not only considerations of terminology and conceptual development, but also issues regarding measurement and analysis. Accordingly, the following sections discuss: (1) specification of key constructs, (2) effects of data aggregation, (3) unit of analysis, and 4) appropriateness of research methodologies. JOURNAL OF MANAGEMENT, VOL. 19,NO. 4, 1993

784

DESS, NEWPORT,

Table 2.

Key Theoretical

Kev Issues A. Single/Multiple Domains

AND RASHEED

Questions in Configuration

Research

Imnortant Ouestions to Consider

I. Is the number of domains selected appropriate to the research question? 2. How have the complexity, simplicity. generalizability tradeoffs been approached? 3. For single domain, what other, interactive relationships may be of interest? Should richer configuration descriptions be developed? 4. For multiple domains. what is the appropriate measure of ‘fit’?

B. Issues of Causality in Configurations

I. Does the research design permit strong causal inferences? 2. Is there strong theoretical support for reverse or mutual causality? 3. How might the implications for researchers and practitioners be affected by reverse or mutual causality? 4. Has the feasibility for the collection and analysis of longitudinal data been explored?

C. Temporal Stability

I. Have theoretical and operational distinctions between changes within a configuration and transition from one configuration to another been developed? 2. Have the internal and external factors that disturb the stability of a configuration been identified? 3. Have the internal and external factors that resist the forces of change been identified? 4. Are certain tvI)es of transitions easier than others? i.e.. are there favored Daths?

Specification of Key Constructs Configurations are inherently multidimensional entities in which key attributes are tightly interrelated and mutually reinforcing. The researcher’s primary task involves disentangling these complex relationships and isolating key constructs. Churchill (1979) suggests that this process begins by specifying the domain of the construct through a review of the relevant literature. Our review identified several constructs that have been used regularly as key components of configurations. These constructs can be grouped into four major classifications: (1) environment, (2) structure, (3) process, and (4) content. However, determining the frequency of the use of constructs such as environment, structure, and strategy process and content in configuration research is quite problematic given the vast literature on such constructs as well as the interdisciplinary nature of strategic management. For example, Meyer (199 1) has recently commented on strategy’s “pluralistic processes enabled by the field’s low entry barriers and fluid boundaries” (p. 83 1). Similarly, Summer, Bettis, Duhaime, Grant, Hambrick, Snow, and Zeithaml have argued that “it is difficult to see the goals of the field and its boundaries” (1990, p. 363). Mintzberg (1990), in his recent and insightful review of configuration theory and research, provides further evidence of the interdisciplinary nature of the strategic management field as well as the value and potential contributions of the configuration approach “in all of the social sciences” (p. 180). Considerable progress has been made in recent years in identifying and operationalizing the salient dimensions of the constructs of environment, structure, content and process. In identifying the dimensions of organizational environments, researchers regularly use such concepts as uncertainty, complexity, dynamism, and hostility or munificence (Duncan, 1972; Tung, 1979; Miller & Friesen, 1983; Dess & Beard, 1984; Miller, 1988). Structure dimensions frequently include complexity, integration, differentiation, formalization, and centralization (Lawrence & Lorsch, JOURNAL

OF MANAGEMENT,

VOL.

19. NO. 4. 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

785

1967; Pugh, Hickson, Hinings, & Turner, 1969; Fredrickson, 1986). Strategy process dimensions have been identified as rational, adaptive, entrepreneurial, innovative, bargaining, and proactive (Mintzberg, 1973; 1978; Miller & Friesen, 1983; Hart, 1992). Strategy content constructs include cost leadership, differentiation, and focus (Porter, 1980). The development of a consistent set of constructs allows researchers to compare results across studies and benefit from the richness of the descriptions they provide. When a construct or set of dimensions exhibits strong content validity, there may be a tendency among researchers to accept them without rigorous testing of other components of validity such as convergent, discriminant, or nomological (predictive) validity (Venkatraman & Grant, 1986). Similarly, there has been a corresponding lack of attention to the various components of reliability such as testretest, alternate form, internal comparison, and scorer reliability (Anastasi, 1988; Carmines & Zeller, 1979). Although increasing attention has been paid to some aspects of reliability in recent years, relatively little effort has been devoted to assessment of validity. For example, Podsakoff and Dalton (1987), in a recent study of 193 empirical articles in the field of organizational studies, lamented the lack of evidence for the assessment of the discriminant and convergent validities of the measures employed. These authors found that two-thirds of the studies reported reliabilities (primarily Cronbach’s alpha) and approximately twenty percent used factor analysis to assess the dimensional~ty of the items that constituted the construct scale. However, “factor analysis and reports of reliability are not adequate, in and of themselves, for establishing the validity of a construct” (1987, p. 430). Churchill’s (1979) “Domain Sampling Paradigm” suggests that the use of multiple item measures, multiple respondents, and purification of the measurement instrument are essential steps in establishing the reliability and validity of key constructs. The research stream that investigates configurations illustrates how constructs can be refined and improved through such careful operationalization. In empirical studies of configurations, Miller (1986; 1987b; 1988) and Miller & Friesen (1980; 1983) use cross-validation of responses by comparisons between different groups of executives and comparisons with alternate measures derived from secondary data sources. This helps to assess interrater reliability and convergent validity, respectively. One of the most common means by which the operationalization of constructs has improved is the use of multiple item variables which are subsequently simplified through factor analysis (Lenz, 1980; Snow & Hrebiniak, 1980; Miller, 1986; 1988). This also helps to assess the unidimensionality of the constructs. This work has led to the identification of several subdimensions of important strategy constructs and a richer, although somewhat less parsimonious, set of descriptive measures. Porter’s original three generic strategies, for example, were expanded to include variations on differentiation (i.e., complex product innovation and marketing differentiation) and focus (i.e. breadthinnovation and breadth- stability) (Miller, 1988). These variations represent more complex descriptions of activities undertaken by organizations in pursuing given strategies. This illustrates the value of carefully adapting measurement scales to JOURNAL OF MANAGEMENT, VOL. 19, NO. 4, 1993

DESS, NEWPORT, AND RASHEED

786

the requirements of specific research questions. Failure to do this may lead to inappropriate or inaccurate measurement and inferences. Effects of Data Aggregation With the exception of fine-grained research methodologies (Harrigan, 1983) such as single case studies, the analysis and interpretation of research is dependent upon the aggregation of data collected from many participants across firms. When such data are aggregated, the uniqueness or “richness” of each firm is compromised. A different type of aggregation problem arises when data are collected from more than one informant in an organization. Although this enhances the validity of the study, the tendency to average the responses from multiple respondents within a firm may result in misleading findings and interpretations if there is wide intrafirm variation. (For an insightful discussion of the tradeoffs involved in choosing between single versus multiple informants, refer to Glick, Huber, Miller, Doty, and Sutcliffe, 1990, p. 304). When deciding how to conceptualize inherently complex organizational phenomena, researchers must recognize the “postulate of commensurate complexity” (Thorngate, 1976) which posits that any theory of social behavior cannot simultaneously achieve the goals of generalizability, accuracy, and simplicity. Therefore, researchers are faced with these inevitable design tradeoffs (Weick, 1979). Typically, in configuration research, the “cost” in terms of accuracy is rather severe because data are not only aggregated across firms, but also variables are either factor analyzed or subjected to cluster analysis techniques. This, in essence, creates new constructs altogether. These new constructs are composites of the original variables, and the raw data (i.e., respondents’ input) become scores on one or more multivariate indices. Thus, there is substantial “information loss” and the researcher must implicitly assume that the empirical results are still reasonably accurate portrayals of the phenomena under investigation. The potential “gain” that is sought from such aggregation is twofold. First, the appropriateness of generalizing beyond one firm or one type of firm is enhanced when relationships are observed using data from multiple sources. Second, parsimony (or simplicity) is achieved through a reduction in the number of variables with which the researcher must contend. An important potential advantage of configuration research lies in the “richness” and parsimony implicit in the composite formed by many interrelated variables. If such variables are sufficiently interrelated to form a composite, the problem of multicollinearity can obfuscate the interpretability of the results of a classical linear regression model which assumes independence among predictor variables. Thus, the multivariate techniques (e.g., factor analysis, cluster analysis) common in configuration research may enhance the analysis and interpretation of data. The proper recognition and control for industry effects in configuration research has important implications for the interpretation of research findings. This is a particularly important consideration since many studies consist of samples of firms from a wide variety of industries in which no one industry typically accounts for more than a small percentage of the entire sample (Hambrick, 1983b; Miller & JOURNAL

OF MANAGEMENT,

VOL. 19, NO. 4, 1993

CONFIGURATION

RESEARCH IN STRATEGIC MANAGEMENT

781

Friesen, 1983; Miller, 1988; Miller, Droge, & Toulouse, 1988; Segev, 1987). There is a substantial body of literature which contends that industry factors can severely affect statistical results and interpretations. Levels of industry profitability have been shown to vary across industries (Hirsch, 1975; Hatten, Schendel, & Cooper, 1978; Beard & Dess, 1981; Porter, 1980). Thus, researchers should be sensitive to the need for controlling for important industry variables as well as other relevant constructs which may be sensitive to industry conditions (Dess, Ireland, & Hitt, 1990). Such careful assessment of variables should aid researchers in avoiding the error of attributing organizational performance to only constructs included in the research design (e.g., distinctive competence, planning systems). An organization’s level of performance, in fact, may be more accurately attributed to a higher rate of growth or profitability of the industry within which it competes. Such erroneous interpretations may lead not only to inaccurate descriptions and interpretations of observed relationships but also to misleading normative implications. Unit of Analysis The specification of a level of analysis has been labeled a “fundamental” question that determines the context for substantive questions in strategy research (Freeman & Lorange, 1985). Choices regarding the unit of analysis have important theoretical and empirical implications. From a theoretical perspective, it could lead to what is often referred to as “ecological fallacy”, i.e., attempting to make inferences at a specific level of theorizing on the basis of data obtained and analyzed at a different level of aggregation (Datta, 1980). Operationally, the types and sources of data used in empirical research could vary significantly based on the unit of analysis chosen. These problems are of particular importance in research using the configuration approach since such research has to link constructs belonging to multiple domains such as environment, content of strategies, strategy making processes and organizational structures. First, a choice must be made between the industry as a whole and the individual firm as the unit of analysis. Since generalizations are often made about groups of firms which may or may not belong to the same industry, this represents not merely a methodological problem, but a theoretical one as well. At the firm level, a further choice must be made as to whether research will focus on the corporate level or the business unit level. After all, a diversified company may compete in many different environments. Differing levels of analysis within the organization may also require different effectiveness criteria weightings depending upon subunit technologies, objectives, or other contingencies (Bedeian, 1986). Similarly, when strategy processes or firm perceptions are being studied, there are additional problems in deciding upon the unit of analysis. The choices here are among surveying CEOs, the top management team (TMT), or a much larger collectivity including middle level managers. Research indicating that the characteristics of the TMT significantly differ from that of the CEO (Norbum, 1989) and that strategy processes are subject to considerable influence by middle level managers (Schilit & Paine, 1988) point to the difficulties involved in making a clear choice. JOURNAL OF MANAGEMENT,

VOL. 19, NO. 4, 1993

788

DESS, NEWPORT, AND RASHEED

The problematic nature of the issue of unit of analysis issue can be illustrated with studies which measure constructs such as structure, strategy, and environmental perceptions on the basis of the questionnaire responses of multiple respondents from sample firms. Such constructs are typically assumed to represent organizational level dynamics (Venkatraman & Grant, 1986). However, as suggested by Rousseau, “when individual data are aggregated to the unit level to measure characteristics or attributes of the unit, it is important to establish the extent to which unit members agree on their descriptions of the unit prior to aggregation” (1985, p. 31). She suggests that indicies of interrater agreement are useful in assessing the extent of wit~n-unit consensus. Low interrater agreement raises important unit of analysis concerns and demonstrates the need to assess reliability beyond interitem reliability (i.e., Cronbach’s alphas) as argued by Podsakoff and Dalton (1987). There are several potential reasons why, in a specific study, there may not be statistically significant differences between within-firm and between-firm variance in the responses to questionnaire items. First, the results could be due to strong differences in the intraorganizational status of the organizations. Second, although multiple respondents from the same firm participate in a study, the participants may not comprise the entire top management team. A representative team of executives from multiple levels within each organization may help to provide a more complete and cohesive picture of the constructs of interest. Other research has indicated that the assumption of agreement among members of the top management team may be erroneous due to the tendency of executives to attend to different sectors of the environment and to perceive different levels of uncertainty, depending on their functional orientations (Bourgeois, 1980a; 1985; Ireland, Hitt, Bettis, & Auld de Porras, 1987; Priem, 1990). In some instances, a balance of points of view or functional biases may be desirable (Huber & Power, 1985), but in others such balance can obscure the evidence of within-firm agreement that is essential to establishing an organizational-level construct. Appropriateness

ofResearch Methodologies

Choosing research methodologies that are appropriate for configuration research presents several challenges. These challenges arise from two basic differences between con~gurations and more traditional types of research that have been discussed already. First, configuration research involves investigation of relationships among constructs in multiple domains. Secondly, configuration research does not necessarily assume unidirectional causal relationships. The methodological implications of these two distinctive characteristics of configuration research are discussed in this section. When two domains are included in the research design, simple correlational methods may generally prove to be adequate. When more than two domains are included in the research design, canonical correlation may be more suitable. Canonical correlation provides a means for analyzing multiple dependent and multiple independent variables simultaneously. As with simple correlational methods, no causal direction is implied. It is possible to use factor analysis in conjunction with canonical correlation to obtain meaningful results in research employing mulJOURNAL

OF MANAGEME~,

VOL.

19, NO. 4, 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

789

tiple domains (Newport, 1990). The investigation of complex constructs benefits considerably from the data reduction provided by this technique. Factor analysis helps to identify relevant dimensions of major constructs and establish their discriminant validity. It also provides continuous measures for variables of interest (Snow & Hrebiniak, 1980; Lenz, 1980). Canonical correlation can be used to analyze relationships between either continuous or discrete variable measures. Factor scores, because they are continuous, are generally preferred over categorical measures and can be used as input in the canonical correlation procedure. Other techniques that have been used to identify and analyze configurations include q-factor analysis (Miller & Friesen, 1984) and cluster analysis (Hambrick, 1983a). However, researchers need to keep in mind that cluster analysis is a “structure-seeking” procedure. In order to ensure that the clusters identified are more than mere statistical artifacts, the cluster solution requires further validation. Aldenderfer and Blashfield (1984) suggest five techniques for validating cluster analysis solutions. They are (1) the cophenetic correlation, (2) significance tests on variables used to create clusters, (3) replication, (4) significance tests on independent variables, and (5) Monte Carlo procedures. The statistical methods used in configuration research must recognize the difficulty in establishing causal relationships. Typically, configuration studies use simple correlation, multiple regression and some multivariate techniques-most notably factor and cluster analysis, for the identification of key variables and their interrelationships. The use of multiple regression, though common, may be less than appropriate since it implies a causal direction in the relationships under examination that may be, at best, tenuous, such as the aforementioned strategy-structure relationship. Thus, where the causality issue is ambiguous, and where research models often tend to be underspecified, regression clearly may not be the most appropriate analytical method. Establishing causal relationships in configuration research increases the demands for rigorous theory building and research designs. Longitudinal research designs or causal modeling techniques such as LISREL can be helpful in providing insights into multivariate relationships as has been demonstrated by Miller, Droge, and Toulouse (1986). This method provides researchers with a means for exploring complex interrelationships characterized by simultaneity and feedback. From a methodological standpoint, LISREL enables researchers to control for random and systematic measurement error and to model the effects of external confounding factors such as methods errors and spurious causation (Bagozzi & Philips, 1982). Methods such as cluster analysis, canonical correlation analysis, and q-factor analysis represent a move away from bivariate reductionism to a more holistic perspective on configurations (Venkatraman & Prescott, 1990). The use of these methods may lead to the empirical discovery of multivariate interconnections among strategy, structure, process and environment which capture organizational reality in a more meaningful way. Meyer ( 199 1) has recently addressed methodological contributions of strategic management research, many of which have important implications for configuration research. These include refinements in the conceptualization and JOURNAL OF MANAGEMENT, VOL. 19. NO. 4, 1993

790

DESS, NEWPORT, AND RASHEED

measurement of firm-level performance which enables researchers to examine the consequences of the formation and adaptation of organizational configurations over time. Also, the use of sophisticated analytical techniques and “fashioning of objective measures of industries and organizations” (p. 826) by employing archival data bases help both to capture the complexity of organizational phenomena and to validate self-report data. Researchers need not be confined to quantitative methodologies in studying configurations. Longitudinal qualitative analysis of organizations can provide meaningful insights about the evolution of configurations as well as the specific relationships among the constructs within a configuration. Through careful comparison of in-depth case studies, it is possible to arrive inductively at relationships among environment, strategy, structure, processes and outcomes (Eisenhardt, 1989; Bourgeois & Eisenhardt, 1988). Despite their ability to overcome many of the problems associated with conventional statistical analysis, qualitative studies are extremely labor intensive and subject to potential problems such as researcher bias and non-replicability (Van de Ven & Huber, 1990). However, it is possible to overcome some of these problems through methodologies such as case replication (Leonard-Barton, 1990) and retrospective event histories (Glick et al., 1990). Conclusion The configuration approach to research in strategic management offers much potential for solving some of the difficulties inherent in accurately articulating multifaceted constructs and relationships among them. In order to realize this potential, several theoretical and methodological issues that we discuss in this paper need to be addressed. The major theoretical issues include (1) determining the number of domains necessary to answer the research question under consideration, (2) marshalling theoretical support to reflect adequately a complex pattern of reverse and mutual causal relationships among constructs of interest, and (3) developing a conceptual framework to explain and predict patterns of stability and change. Important methodological issues include (1) specifying of key constructs to satisfy the criteria for reliability and validity, (2) employing techniques to reduce the possibility of erroneous inferences due to data aggregation, (3) deciding on the appropriate unit of analysis, and (4) selecting methodologies that are consistent with the theoretical question under examination and the number of domains in the study. Although, for purposes of clarity and convenience, we have addressed theoretical and methodological issues separately, we recognize their strong interdependence. Consistent with Martin’s (1982) depiction of the research process, theoretical issues cannot be divorced from their methodological considerations and vice versa. We conclude by discussing three examples of such interdependence. The specific research question one wishes to explore has implications for not only the choice of domains but also the selection of data sources. For example, when investigating strategy process constructs (Fredrickson, 1984; Segev, 1987) researchers tend to rely on the use of managerial perceptions, whereas strategy JOURNAL

OF MANAGEMENT,

VOL. 19. NO. 4, 1993

CONFIGURATION RESEARCH IN STRATEGIC MANAGEMENT

791

content or realized strategies (Mintzberg, 1978) is typically studied using secondary data sources. Since configuration research uses constructs from multiple domains, there is a corresponding need to combine the use of primary and secondary sources in studying configurations. Further, convergent validity in configuration research can be enhanced by combining primary and secondary sources of data when operationalizing a given construct. For example, Zajac and Shortell (1989) combined both managerial self-reports and archival data in their operation~ization of the Miles and Snow (1978) framework. The theoretical underpinnings of a configuration study often strongly influence the emphasis placed on a particular domain. For example, a study that adopts a perspective of environmental determinism (Bourgeois, 1985) may place primary emphasis on the env~onmental imperative, while a study consistent with the strategic choice perspective (Child, 1972) would stress the importance of the strategy imperative. The former would focus on such variables as entry, exit, and mobility barriers (Porter, 1980; Oster, 1982) and environmental attributes such as munificence (Castrogiovanni, 1991), whereas the latter would emphasize variables that reflect corporate-level and business-level strategic positioning and resource allocations. As Hrebeniak and Joyce (1985) point out, these two perspectives are not mutually exclusive and organizational phenomena could be more meaningfully explored by synthesizing these two seemingly contradictory perspectives. Such a synthesis of perspectives requires the inclusion of several constructs from multiple domains. However, an increase in the number of domains calls for a corresponding increase in the sophistication of research methodologies and data analytic procedures involved. The use of data reduction techniques such as interitem analysis and factor analysis can help achieve parsimony in the depiction of individual constructs by identifying their salient dimensions. Similarly, the use of techniques such as factor analysis and cluster analysis also help in achieving the simplification or “lumping” which is a basic characteristic of configuration research (Mintzberg, 1990). Exploring the nature of organizational change can often benefit from configuration research. As noted earlier, Snow and Hambrick (1980) suggest that strategic change (as opposed to incremental adaptation) requires major changes in both a firms’s strategic direction as well as its internal structure and process. Thus, studies exploring changes in configurations can benefit from the inclusion of multiple domains such as strategy, environment, process and structure to depict both the magnitude and direction of change. Configurations, as we discussed earlier, are relatively stable. Change in configurations tend to be slow and gradual (except in the case of revolutionary changes). The study of gradual change spanning long periods of time has methodological implications. Chandler’s (1962) classic study used historical analysis to explore the relationships among the strategies of expansion, vertical integration, and diversification as well as the functional and divisional structures in the context of broad environmental changes. The usefulness of historical analysis in understanding organizational change is further evident from studies such as ~in~berg and Waters (1985) and Smith and Grimm (1987). JOURNAL OF MANAGEMENT, VOL. 19.NO. 4, 1993

192

DESS, NEWPORT, AND RASHEED

In conclusion, the configuration approach holds considerable promise for investigating and explaining complex organizational phenomena. As we have discussed in this article, the configuration approach is compatible with a variety of theoretical perspectives on organizations. It can be used to depict both static and dynamic aspects of organizations and their environments. Further, the configuration approach can be used with various data analytic methodologies. Given these advantages, it is not surprising that in recent years, there has been increasing recognition of the benefits of the configuration approach. A~knowIedg~ent: An earlier version of this paper was presented at the 1990 Academy of Management Meetings, San Francisco, CA. We would like to acknowledge the helpful comments of David Arnott, Brian Boyd, Jane Dutton, David Harrison, and Richard Priem on an earlier draft. References Aldenderfer, M. S. & Blashfield, R. K. (1984). Cluster anofysis. Beverly Hills: Sage. Anastasi, A. (1988). Psychological testing, 6th ed. New York: Macmillan. Bagozzi, R. P. & Phillips, L. W. (1982). Representing and testing organizational theories: A holistic model. Administrative Science Quarterly, 27: 459-489. Beard, D. & Dess, G. G. f 1981). Corporate-level and business-level strategy and firm performance. Academy of Monugemen~ Journal. 24: 663-688. Bedeian, A. G. (1986). Contemporary challenges in the study of organizations. Journal of Management. 12: i 85201. Blalock, H. (1969). Theory construction: From verbal to mathemntical formulations. Englewood Cliffs, NJ: Prentice-Hall. Bourgeois. L. J., 111.( 1980af. Performance and consensus. Strategic ~o~ogerne~t Jourrxzt, I: 227-248. f 1980b). Strategy and enviro~ent: A conceptual integration, Academy of~u~~ogemenf Review, 5: 2% --__, 39. (1984). Strategic management and determinism. Academy of Management Review 9: 586-596. ____. (1985). Strategic goals, perceived uncertainty, and economic performance in volatile environments. _____. Academy of Management Journal. 28: 548-573. Bourgeois, L. 3. & Eisenhardt, K. (1988). Strategic decision processes in high velocity environments: Four cases in the microcomputer industry. Monogeme~t Science, _W: 816-835. Burgeiman, R. A. (1983). A model of the interaction of strategic behavior, corporate context, and the concept of strategy. Academy of Management Review, 8: 6 I-70. Carmines, E. G. & Zeller, R. A. (1979). Reliability and validity assessment. Beverly Hills: Sage. Castrogiovanni, C. J. (199 I). Environmental munificence: A theoretical assessment. Academy of Management Review 16: 542-565. Chandfer, A. D. (1962). Strategy and strucfure: Chapers in the [listor?, qf A~?~eri~o~ i~zd~~friaf enterprise. Cambridge: MIT Press. Child, J. (1972). Organizational structure, environment, and performance: The role of strategic choice. Sociology, 6: l-22. Chrisman, J. J., Hofer, C. W., & Bolton, W. R. (1988). Toward a system for classifying business strategies. Academy of Management Review, 13: 4 l3- 428. Cohen, D., March, J. G., & Olsen, J. P. (1982). A garbage can model of org~ization~ choice. Adrn~nj~~ru~~~,e Science Quarter@ 17: l-25. Churchill, G. A., Jr. (1979). A paradigm for developing better measures of marketing constructs. Journal of Marketing Research, 16: 64-73. Datta, Y. (1980). New directions for research in business strategy. Journal of General Management. 6: 48-60. Dess, G. G & Beard, D. (1984). Dimensions of organizational task environments. Administrative Science Quarter+ 29: 52-73. Dess, G. G. & Davis, P. (1984). Porter’s (1980) generic strategies as determinants of strategic group membership and organizational performance. Academy of Manogemenr Journal, 27: 467-488. Dess, G. G.. Ireland, R. D., & Hitt. M. A. (1990). Industry effects and strategic management research. Journal of Management, 16: 7-21.

JOURNAL

OF MANAGEMENT.

VOL. 19, NO. 4. 1993

CONFIGURATION

RESEARCH

IN STRATEGIC MANAGEMENT

793

Duncan, R. B. (1972). Characteristics of organizational environments and perceived environmental uncertainty. Administrative Science Quarterly, 17: 3 13-321. Eisenhardt, K. (1989). Building theories from case study research. Academy of Management Review, 14: 532550. Eisenhardt, K. & Schoonhoven, C. B. (1990). Organizational growth: Linking founding team, strategy, environment, and growth among U.S. semiconductor ventures, 1978- 1988. Administrative Science Quarterly, 35: 504-529. Fredrickson, J. W. (1984). The comprehensiveness of strategic decision processes: Extension, observations, and future directions. Academy of Management Journal, 27: 399-423. . (1986). The strategic decision process and organizationa structure. Academy of~anugeme~t Review II: 280-297. Freeman, R. E. & Lorange, P (1985). Theory building in strategic management. Pp. 9-38 in R. Lamb & P Shrivastava (Eds.), Advances in Strategic Management, Vol. 3. Greenwich, CT: JAI. Ginsberg, A. (1988). Measuring and modelling changes in strategy: Theoretical foundations and empirical directions. Strategic Management Journal, 9: 559-576. Glick, W. H., Huber, G. P., Miller, C. C, Doty, D. H., & Sutcliffe, K. M. (1990). Studying changes in organizational design and effectiveness: Retrospective event histories and periodic assessments. Organization Science. I: 293-3 12. Greiner, L. (1972). Evolution and revolution as organizations grow. Harvard Business Review: JO: 37-46. Hall, D. J. & Saias, M. A. (1980). Strategy follows structure! Strategic Management Journal, I: 149-163. Hambrick, D. C. (1983a). An empirical typology of mature industrial-product environments. Academy of Management Journal, 26: 213-230. (I 983b). Some tests of the effectiveness and functional attributes of Miles and Snow’s strategic types. ----I Academy of Management Journal, 26: 5-26. Harrigan. K. R. (1983). Research methodologies for contingency approaches to business strategy. Academy of Management Review. 8: 398405. Hart, S. J. (19921. An integrative framework for strategy-making processes. Academy of Management Review, I7: 327-351. Hatteo, K. J., Schendel, D. E., & Cooper. A. C. (1978). A strategic model of the U.S. brewing industry. Academy ofManagement Journal, 21: 592- 610. Hawes, J. M. & Crittenden, W. F. (1984). A taxonomy of competitive retailing strategies. Strategic Management Journal. 5: 275-281. Hill, C. W. L. (1988). Differentiation versus low cost or differentiation and low cost: A contingency framework. Acudemy ofManagemenf Review, I3: 401-422. Hinings, C. R. & Greenwood, R. (1988a). Organizational design types, tracks, and the dynamics of strategic change. Organization Studies. 9: 293-316. (1988b). The tracks and dynamics afstrategic change. Oxford: Blackwell. ----. Hirsch, P M. (1975). Organizational effectiveness and the institutional environment. Administrative Science Quarteriy. 20: 327-344. Hitt, M. E. & Ireland, R. D. (1987). Peters and Wate~an revisited: The unended quest for excellence. Academy ofManagement Executive. I: 91-98. Hofer, C. W. (1975). Toward a contingency theory of business strategy. Academy ofManagement Journal. 18: 784-810. Huber, C. P. & Power, D. J. (1985). Retrospective reports of strategic-level managers: Guidelines for increasing their accuracy. Strategic Managemenf Jaumal, 6: I7 l- 180. Hrebiniak, L. G. & Joyce, W. F. (1985). Orga~zationai adaptation: Strategic choice and enviro~enta1 determinism. Administrative Science Quarterly, 30: 336-349. Ireland, R. D., Hitt, M. A., Bettis. R. A., & Auld de Porras, D. (1987). Strategy formulation processes: Differences in perceptions of strength and weakness indicators and environmental uncertainty by managerial level. Strategic Management Journal, 8: 469-485. Jaedicke, R. K. (1989). f~side~~aIu~ress at the f989AACSB annual meeting, Montreal. April 16-19. Johnson, G. f 1988). Rethinking incrementalism. Strategic Ma~ge~neat Jaurnai, 9: 75-91. Jones, G. R. & Butler, J. E. (1988). Costs. revenues, and business-level strategy. Academy of Management Review 13: 202-213. Kim, L. & Lim, Y. (1988). Environment, generic strategies and performance in a rapidly developing country: A taxonomic approach. Academy of Management Journal, 31: 802-827. Lawrence, P R. & Lorsch. J. W. (19673. Organization and environment. Boston: Harvard University Press. Lenz, R. T. (1980).Enviro~ent, strategy, o~a~~tional structure, and performance: Patterns in one industry. Strategic Management JotcrnuI. I: 209-226.

JOURNAL

OF MANAGEMENT,

VOL. 19, NO. 4, 1993

794

DESS, NEWPORT, AND RASHEED

Leonard-Barton, D. (1990). A dual meth~oIogy for case studies: Synergistic use of a longitu~nal single site with replicated multiple sites. ~~~~zatio~ Science, I: 248-266. Martin, J. (1982). A garbage can model of the research process. Pp. 17-39 in J. E. McGrath, J. Martin, & R. A. Kulka (Eds.), Judgment calls in research. Beverly Hills: Sage. McKelvey, B. (1982). Organizational systematics. Berkeley, CA: University of California Press. Meyer, A. D. (1991). What is strategy’s distinctive competence? Journal of Management, 17: 821-833. Miles, R. E. & Snow, C. C. (1978). Organization sfrutegy, structure, undptvcess. New York: McGraw,-Hill. Miller, A. Br Dess, G. G. (1993). Assessing Porter’s (1980) model in terms of its g~eralizability, ao&acy, and simplicity. Journal of Management &dies. Miller, D. (198 I). Toward a new contingency approach: The search for organizational gestalts. Journal of Management Studies. 18: l-26. (1982). Evolution and revolution: a quantum view of structural change in organizations. Journal of _---I ~a~ageme~t Studies, 19: 13 I - IS I. (1986). ~o~~rations of strategy and structure: Towards a synthesis. Srrafegic Ma~gemenf Journaf, ____. 7: 233-249. (1987a). The structural and enviroumental correlates of business strategy. Strategic Management ----. Journal. 8: 55-76. (1987b). Strategy making and structure: Analysis and implications for performance. Academy of ----. Management Journal. 30: 7-32. (1987~). The genesis of co~guration. Academy ofM~ageme~f Review, 12: 686-701. ____. (1988). Relating Porter’s business strategies to environment and structure: Analysis and performance _---. implications. Academy of Management Journal. 31: 280-308. Miller, D., Droge, C., & Toulouse J. (1988). Strategic process and content as mediators between organizational context and structures. Academy of Management Journal, 31: 544-569. Miller, D. & Friesen, P H. (1977). Strategy making in context: Ten empirical archetypes. Journal of Management Studies, I& 255-280. (1978). Archetypes of strategy formulation. Management Science, 24: 92 l-933. _---. (1980). Archetypes of organizational transition. Administrative Science Quarterly, 25: 269-299. --_-. (1982). The longitudinal analysis of organizations: A methodological perspective. Management ____. Science. 28: 1013-1034. (1983). Strategy-making and environment: The third link. Strategic Management Journal. 4: 221-23.5. _---. (1984). O~a~~zatjoh~ A 4~nf~ view. Englewood Cliffs, NJ: Prentice-Hall. ____. Miller, D. & Mint&erg, H. (1983). The case for configuration. Pp. 57-73 in G. Morgan (Ed.), Beyond method. Beverly Hills, CA: Sage. Mintzberg, H. (1973). Strategy-making in three modes. California Management Review, 16: 44-53. f 1978). Patterns in strategy formation. Management Science. 24: 934-948. ----. (1990). Strategy formation: Schools of thought. Pp. 10.5-236 in J. W. Fredrickson (Ed.), Perspectives ____I on strategic m~agement. New York: Harper Business. Mintzberg, H. & Waters, J. A. (1985). Of strategies, deliberate and emergent. Strategic Management Journaf. 6: 257-272. Murray, A. I. (1988). A contingency view of Porter’s generic strategies. Academy of Management Review, 13: 390-400. Newcomb, M. D. & Bentler, P. M. (1988). Consequences of adolescent drug use: lmpuct on the lives ofyoung ad&s. Beverly Hills: Sage. Newport, S. (1990). Stakeholder influence on strategic configurations in the heaith care industry. Unpublished doctoral dissertation, University of Texas, Arlington. Norbum, D. (1989). The chief executive: A breed apart. Strafegic Management Journal, 10: I-15. Oster, S. (1982). lntra-industry structure and the ease of strategic change. Review of Economics and Statistics. 64: 376-383. Pod&off, P M. & Dalton, D. R. (1987). Research methodology in orga~zati~ studies. Jo~r~~ of Management, 13: 419-441. Porter, M. E. (1980). Compefitive strategy. New York: Free Press. Priem, R. L. (1990). Top management team group factors, consensus, and firm performance. Strafegic Management Journal, I I : 469-478. Pugh, D. S., Hickson, D. J., Hinings, C. R., & Turner, C. (1969). The context of organizational structures. Administrative Science Quarterly, 14: 9 I- I 14. Rosenberg, M. (1968). The logic of survey analysis. New York: Basic Books. Rousseau, D. M. (1985). Issues of level in organizational research: Multi-level and cross-level perspectives. Pp. l-37 in L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior, Vol. 7. Greenwich, CT: JAI.

JOURNAL

OF MANAGEMENT,

VOL. 19, NO, 4, 1993

CONFIGURATION

RESEARCH

IN STRATEGIC MANAGEMENT

195

Rumelt, R. P (1974). Strategy, structure and economic performance. Cambridge, MA: Harvard University Press. Schendel, D. & Hofer, C. W. (1979). Strategic management. Boston: Little, Brown. Schilit, D. K. & Paine, F. T. (1988). An examination of the underlying dynamics of strategic decisions subject to upward influence activity. Journal of Management Studies, 24: 161- 188. Schoonhoven, C. B. (1981). Problems with contingency theory: Testing assumptions hidden within the language of contingency ‘theory’. Administrative Science Quarterly. 26: 349-377. Segev, E. (1987). Strategy, strategy making, and performance: An empirical investigation. Management Science, 33: 258-269.

Smith, K. G. &Grimm, C. M. (1987). Environmental variation, strategic change, and firm performance: A study of railroad deregulation. Strategic Management Journal, 8: 363-376. Snow, C. C. & Hambrick, D. C. (1980). Measuring organizational strategies: Some theoretical and methodological problems. Academy of Management Review, 5: 527-538. Snow, C. C. & Hrebiniak, L. G. (1980). Strategy, distinctive competence, and organizational performance. Administrative

Science Quarterly, 25: 317-336.

Summer, C. E., Bettis, R. A., Duhaime, 1. M., Grant, J. H., Hambrick, D. C., Snow, C. C., & Zeithaml, C. P. (1990). Doctoral education in the field of business policy and strategy. Journal of Management, 16: 361398.

Thorngate,

W. (1976).

“In general”

vs. “it depends”:

Some comments

on the Gergen-Schlenker

debate.

Personality and Social Psychology Bulletin. 2: 404410.

Tung, R. L. (1979). Dimensions of organizational environments: An exploratory study of their impact on organizational structure. Academy of Management Journal. 22: 672-693. Tushman, M. L., Newman, W. H., & Romanelli, E. (1986). Convergence and upheaval: Managing the unsteady pace of organizational evolution. California Management Review, 29: 29-44. Tushman, M. L. & Romanelli, E. (1985). Organization evolution: A metamorphosis model of convergence and reorientation. Pp. 171-222 in B. Staw & L. L. Cummings (Eds.), Research in organizational behavior, Vol. 7. Greenwich, CT: JAI. Ulrich, D. & McKelvey, V. (1990). General organizational classifications: An empirical test using the United States and Japanese electronics industry. Organization Science, I: 99- 118. Van de Ven, A. H. & Huber, G. P. (1990). Longitudinal field research methods for studying processes of organizational change. Organization Science, I: 213-219. Venkatraman, N. (1989). The concept of fit in strategy research: Toward verbal and statistical correspondence. Academy of Management Review, 14: 423-444.

Venkatraman,

N. & Camillus, J. C. (1984). Exploring

the concept of ‘fit’ in strategic management.

Academy of

Management Review, 9: 5 13-525.

Venkatraman, N. & Grant, J. H. (1986). Construct measurement in organizational strategy research: A critique and proposal. Academy of Management Review, II: 7 I-87. Venkatraman, N. & Prescott, J. E. (1990). Environment-strategy coalignment: An empirical test of its performance implications. Strategic Management Journal. I I: I-24. Weick, K. (1979). The social psychology of organizing, 2nd ed. Reading, MA: Addison-Wesley. White, R. E. (1986). Generic business strategies, organizational context and performance: An empirical investigation. Strategic Management Journal. 7: 217-23 1. Wright, P., Kroll, M., Tu, H., & Helms, M. (1991). Generic strategies and business performance: An empirical study of the screw machine products industry. British Journal of Management, 2: 1-9. Zajac, E. J. & Shortell, S. M. (1989). Changing generic strategies: Likelihood, direction, and performance implications. Strategic Management Journal. IO: 413-430.

JOURNAL

OF MANAGEMENT,

VOL. 19, NO. 4, 1993