sign in sign up
<<
Home , Management

஽ Academy of Management Review 2010, Vol. 35, No. 3, 346–357.

EDITOR’S COMMENTS: CONSTRUCT CLARITY IN THEORIES OF MANAGEMENT AND

One of the more commonly cited reasons for narrowly constituted on empirical questions of rejecting a manuscript at AMR is that reviewers operationalization and . feel the submission lacks “construct clarity.” Yet Nor is my intent to discuss the broader ques- reviewers (and editors) often find it difficult to tion of what constitutes “good” theory. This topic articulate precisely what construct clarity is. In- has already received substantial prior, more deed, in contrast to other social , such skilled attention (i.e., Bacharach, 1989; Sutton & as and , where the nature Staw, 1995; Weick, 1989). While recognizing that and role of constructs are subjects of consider- strong, clear constructs contribute to good the- able debate, the field of management seems ory, my here is more modest. I simply in- unusually silent on the subject. The absence of tend to focus the discussion on why we need an open discussion about theoretical constructs clear constructs in developing theories of man- is somewhat surprising given their widespread agement and how best to accomplish this. use in and undeniable importance to manage- This essay proceeds in four parts. In the first I ment theory. discuss what constitutes a theoretical construct The purpose of this essay, thus, is twofold. My and how to best create clarity in our constructs. first objective is pragmatic. I hope to offer some Second, I outline why we need clear constructs degree of clarification about how the issue of in management theory. In the third part I outline construct clarity is dealt with at AMR.Idosoby how the term construct means different things to offering a review and synthesis of prior writing different kinds of researchers, and I explore how on the subject in management journals and in standards of construct clarity vary across epis- journals from related social disciplines. temological and methodological divisions. Fi- nally, I present a more normative argument Ideally, this will assist authors of prospective about the need for more open dialogue about the AMR manuscripts to improve the clarity of their role of constructs in our discipline. theoretical constructs. My second objective is less pragmatic but, arguably, more important. I hope to open a dialogue within the AMR com- munity about the role and use of constructs in WHAT ARE CONSTRUCTS ...AND WHAT IS developing theories. CONSTRUCT CLARITY? Before doing this, however, I should be clear Constructs are conceptual abstractions of about the scope of this essay. The intent is not to phenomena that cannot be directly observed discuss issues of construct validity. This is a (MacCorquodale & Meehl, 1948). Kerlinger de- subsidiary topic of high importance that has fines a construct as a concept that has “been received and continues to receive considerable deliberately and consciously invented or attention (i.e., Bagozzi & Edwards, 1998; Cook & adopted for a special scientific purpose” (1973: Campbell, 1979; Schwab, 1980). Questions of 29). Constructs are not reducible to specific ob- construct clarity and validity are quite distinct servations but, rather, are abstract statements of (Bacharach, 1989). Issues of construct validity, categories of observations (Priem & Butler, 2001). which flows from the ability to crisply and pre- Clear constructs are simply robust categories cisely describe theoretical constructs, are more that distill phenomena into sharp distinctions that are comprehensible to a community of researchers—that is, animal, mineral, or vege- table; gas, liquid, or solid. I thank Bob Gephart, Bob Hinings, Dave Whetten, and the editor and associate editors of the Academy of Management Constructs are the foundation of theory. Bach- Review for their helpful and stimulating comments on ear- arach defines theory as a “ of constructs lier versions of this essay. . . . in which the constructs are related to each

346 Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download, or email articles for individual use only. 2010 Editor’s Comments 347 other by propositions” (1989: 498). Just as con- Perhaps the most common definitional issue structs are the building blocks of strong theory, in manuscripts is that authors simply fail to clear and accurate terms are the fundament of define their constructs. Authors often use terms strong constructs. As Sutton and Staw (1995) re- described as constructs and assume that the mind us, constructs are not a substitute for the- reader understands the intended meaning. This ory. They are, however, essential to the process is clearly problematic since any word has both a of building strong theory. Constructs, therefore, denotative and connotative meaning. In The are a necessary but insufficient condition for Structure of Complex Words, famous literary theory. critic William Empson (1995/1951) demonstrates The essence of construct clarity comprises that even individual words like “knowledge” four basic elements. First, definitions are impor- and “honest” contain a complex “inner gram- tant. Construct clarity involves the skillful use of mar” that can generate multiple and sometimes language to persuasively create precise and contradictory interpretations of the same word. parsimonious categorical distinctions between Offering definitions of key terms and constructs, concepts. Second, construct clarity requires the thus, is a bare minimal standard of construct author to delineate the scope conditions or con- clarity. textual circumstances under which a construct A good definition should accomplish several will or will not apply. Third, not only must the tasks. First, the definition should effectively theorist offer clear conceptual distinctions, but capture the essential properties and character- he or she must also show their semantic rela- istics of the concept or phenomenon under con- tionship to other related constructs. Finally, the sideration. theorist must demonstrate a degree of coher- Second, a good definition should avoid tautol- ence or logical consistency of the construct in ogy or circularity. This occurs when a theorist relation to the overall theoretical argument he uses elements of the term being defined in the or she is trying to make. definition or incorporates antecedent or out- Reviewers are quick to reject a manuscript come variables as part of his or her definition. where the core constructs are weakly defined, where Thus, defining a “transformational leader” as a contextual conditions are not specified, or where “leader who transforms ” is an their connection to other constructs and to the empty definition because it uses the construct in overall theory is not clear. Unfortunately, the the definition. Similarly, defining “cognitive typical rejection letter offers little space to con- ability” as “a capability that enables people to textualize or elaborate these conditions. How do learn more effectively in contexts that are dy- these constituent elements contribute to clear namic or complex” creates confusion because it construct development? How do they contribute incorporates, as part of the definition, anteced- to theory? More significantly, what can I, as an ent variables (i.e., complex and dynamic con- author, do to ensure that the constructs used in texts) that are likely causally related to the con- my theoretical argument meet the requisite struct being defined. standard for clarity and precision? My intent is Third, a good definition should be parsimoni- to address these questions in the balance of this ous. That is, it should try to capture as concisely section. I begin by elaborating each of the four as possible the essential characteristics of a subcomponents of construct clarity described phenomenon or concept. The challenge here is above under the following four headings; defi- twofold. On the one hand, the definition should nitions, scope conditions, relations between focus the meaning of the term as narrowly as constructs and coherence. possible. On the other hand, there is a danger of overshooting the mark—offering a construct definition that is so narrow it lacks relevance Definitions and cannot be generalized. Theory construction relies on the ability of These three characteristics of a good defini- theorists to accurately abstract empirical phe- tion are intended to help fix the meaning of a nomena into robust conceptual generalizations. theoretical term. Meanings, however, are notori- Accomplishing this requires an unusual skill in ously difficult to specify, for a variety of reasons. translating abstract concepts into crisply de- One reason is that the meanings of words are fined theoretical constructs. never fixed or permanent. When different re- 348 Academy of Management Review July searchers apply an existing construct to a new inition—which they then recalibrate by trim- empirical context, they often change the mean- ming away the surplus meaning of prior uses ing of the term, however slightly. Over time and and introducing their own three salient at- over multiple empirical applications, the defini- tributes (power, legitimacy, and urgency). In this tion of a construct tends to drift—that is, it ac- way the authors demonstrate their command of quires substantial “surplus meaning” (MacCor- the relevant literature by illustrating the prior quodale & Meehl, 1948) or meaning beyond the accumulation of surplus meaning in a term and parameters of its original intended definition. then impose some order on the construct by in- Therefore, it is incumbent on the theorist to first troducing a new, refined definition of the term. demonstrate the prior uses of the term and then The most common error in developing con- to illustrate, as exhaustively as possible, prior structs is making them too general. There is, variation in how the term has been used. however, considerable danger in extending this Similarly, new constructs are often given logic too far. That is, on occasion constructs can names used in common speech—for example, be presented too narrowly. Recall that a key “organizational performance.” In this case the function of constructs is to create robust catego- term performance has acquired substantial sur- rizations of phenomena. If the categories are plus meaning as a result of its use in literature expressed too narrowly, the theoretical rele- and advertising, as well as its everyday use. The vance of a construct will be compromised (Ast- term has a depth of connotation, some of which ley, 1985). The creative capacity of theoretical works well for the intended theory and some of constructs rests on the tension between defini- which does not. As a result, researchers working tional accuracy and ideational scope. Effective at both the organizational (Hansen & Wernerfeldt, constructs create broad categories and, thus, 1989) and individual (Rogers & Wright, 1998) levels should not be reducible to narrow empirical ob- of analysis have expressed concern about the dif- servations. Some degree of linguistic ambiguity ficulty of constraining the definition of perfor- is therefore a useful component of any theoreti- mance as a construct. The construct of “family cal construct (Astley & Zammuto, 1992). The chal- ” also suffers from definitional clarity lenge is to create constructs that are sufficiently as a result of its surplus meaning acquired from narrow enough to strip away unintended conno- everyday use. Sharma, Chrisman, and Chua tations and surplus meaning but are conceptu- (1996) found thirty-four different definitional ally broad enough to capture the underlying es- uses of the term in management literature. sence of the phenomenon. In any case, it is critically important for the theorist to attempt to strip away the extraneous Scope Conditions meaning that has become attached to a con- struct. He or she can accomplish this by offering In contrast to the physical sciences, few con- a contextually specific and clear definition of structs in organization theory have universal the term. But achieving this is no simple task, application. Rather, organizational constructs particularly where, as in most instances of the- tend to be highly sensitive to and contingent on ory development, there is no clear agreement on contextual conditions. So, for example, con- the substantive definitional content of a structs developed from research on large, pub- construct. licly traded may have little rele- There are, however, excellent illustrative ex- vance for closely held corporations (Shane & amples of how this can be done. Consider, for Venkataraman, 2000) or family . Sim- example, how Mitchell, Agle, and Wood (1997) ilarly, organizational constructs are highly cul- develop their definition of the term turally sensitive. Constructs formulated by in their groundbreaking AMR article. They begin studying North American corporations may not by acknowledging the “maddening variety” of exhibit the same characteristics in Asian orga- uses of the term. They then catalog these defi- nizations (Gibson & Zellmer-Bruhn, 2001; Shen- nitions across various theories, including kar & von Glinow, 1994; White, 2002). agency, behavioral, ecological, institutional, re- There is also a noted tendency within man- source dependence, and transaction cost theo- agement scholarship for researchers to “borrow” ries of the firm. The authors go on to offer a new concepts from other disciplines, such as psy- definition of the term—a purposefully broad def- chology or biology. In addition, organizational 2010 Editor’s Comments 349 researchers often take constructs developed at level of analysis under which a proposed con- one level of analysis, such as the individual, struct will apply (Rousseau, 1985). As Klein, and apply them to another level of analysis, Dansereau, and Hall remind us, such as the group, , or organization (Floyd, No construct is level free. Every construct is tied 2009). While the practice of borrowing constructs to one or more organizational levels or entities, can be beneficial, it is often done unreflectively, that is, individuals, dyads, groups, organizations, without considering how the borrowed construct industries, markets, and so on. To examine orga- might vary as a result of the distinctive nature of nizational phenomena is thus to encounter levels issues. Levels issues create particular problems organizations (Whetten, Felin, & King, 2009). In when the level of theory, the level of measure- the process of borrowing constructs, researchers ment, and/or the level of statistical analysis are often assume universality and neglect to clarify incongruent (1994: 198). whether the essential characteristics of a con- So, for example, employee performance is a con- struct that may have been present in the origi- struct that is highly dependent on the level of nal context are equally present in the new one. analysis within the organization where observa- Because organizational constructs lack uni- tions are made. An individual might be perform- versality, it is very important for theorists to ing extremely well in reference to his or her past spell out the contextual conditions under which performance (individual level) but below aver- a proposed construct will or will not adhere (Du- age relative to group performance and merely bin, 1969). Failure to specify the “boundary lim- average at the organizational level. its” or “scope conditions” of a construct exposes Organizational constructs are subject to con- one’s theoretical argument to almost certain re- straints of time because organizational phenom- jection. When an author claims universal appli- ena tend to be temporal, and as a result, cation of a construct, it is usually very easy for a changes in time may affect the expression of reviewer to identify at least one exception to the any construct (Avital, 2000; Zaheer, Albert, & Za- hypothetical abstraction. Indeed, as Walker and heer, 1999). Critics, however, have observed that Cohen observe, “One can easily find exceptions management theorists tend to ignore the tempo- to most of the propositions which are advanced ral boundaries of phenomena and assume in- as general . . . principles” (1985: 288). variance over time in key constructs. George Finding a single exception is often fatal to a and Jones offer two key examples: construct because it implies that any proposi- tion associated with the construct is false. Re- So, for example, any conceptualization of motiva- viewers may take this position even in cases tion must contain reference to its inherently sub- where there is substantial positive empirical jective and changing nature and definitions should not be constrained by viewing it through support for a construct, largely because most the lens of standard time. As another example, it reviewers have been oversocialized to accept should be recognized that the desire to act oppor- falsification as the basis of scientific truth. An tunistically can be viewed as a state of mind that easy resolution to this problem, however, is sim- can change, often quickly; however opportunism ply to avoid overgeneralizing the nature of your is often treated as a stable tendency that exists across people and situations (2000: 667). constructs by placing scope conditions on them—that is, carefully outlining the contextual Another example, recounted by Zaheer et al. conditions under which the constructs will or (1999: 726), is that the antecedents and nature of will not apply. trust seem to vary over different time scales. There are three general categories or types of When observed over short time frames, trust ap- scope conditions: space, time, and values (Bach- pears to be based on stereotypical features of a arach, 1989). The first two are relatively straight- potential alliance partner. But when measured forward and easy to address. Constraints of over longer time frames, trust is based on more space, discussed above, refer to the fact that specific or individualized elements of potential constructs may apply differently in different partners. types of organizations, at different levels of or- Often, organizational constructs implicitly as- ganizational analysis, under different cultural sume boundary conditions of time, without mak- conditions, or in varying environmental circum- ing them explicit. George and Jones (2000: 662) stances. Perhaps the most common omission in point to research on job satisfaction as an illus- theory manuscripts is a failure to specify the tration of this, where research typically mea- 350 Academy of Management Review July sures job satisfaction at time 1 and then mea- should provide answers to what the constructs sures absenteeism at time 2—usually one year are, how and why they are related, who the later. The methodology implies certain assump- constructs apply to, and when and where they tions about the temporal scope conditions of job are applicable. satisfaction as a construct—that is, that job sat- isfaction is stable over that time period but ab- Relationships Between Constructs senteeism is incremental. Job stress, similarly, has been viewed as a construct with both incre- With apologies to John Donne, no construct is mental temporal scope conditions—it increases an island. Constructs exist only in referential incrementally as an employee encounters in- relationships, either explicit or implicit, with creasing levels of stressors—and discontinuous other constructs and with the phenomena they temporal scope conditions—a specific event in- are designed to represent. New constructs are creases stress temporarily but the stress then rarely created de novo. Rather, they are usually subsides. Both conceptualizations of the con- the result of creative building upon preexisting struct of stress are accurate, but they operate constructs, which themselves refer to other ex- under different boundary conditions of time. The tant constructs, in an ongoing web of referential onus is on the researcher to clearly state the relationships. Constructs, thus, are the outcome temporal scope conditions under which he or of a semantic network of conceptual connections she assumes the construct to operate. to other prior constructs. Psychologists refer to Constraints of value are more complex and this as the nomological network (Cronbach & arguably more difficult to address. Constraints Meehl, 1955) and semioticians as the system of of value refer to scope conditions of a theoretical signification (Saussure, 2000). While these two construct that arise as a result of the assump- groups of researchers may not share much in tions or world view of the researcher. So, for terms of epistemology, they both seem to under- example, Pierce, Gardner, Cummings, and Dun- stand that theoretical constructs are suspended ham (1989) note that most of the constructs de- in a complex web of references to and relation- veloped by human theorists, such as ships with other constructs. turnover, climate, and citizenship, tend to “priv- Part of the task in demonstrating construct ilege” or adopt the point of view of the employee clarity, thus, is to draw out these relationships and his or her role within the organization. in a that the reader can understand. These authors point out the implications of Describing the historical relationships between these value assumptions and how they might the proposed new construct and the prior histor- limit, theoretically, the depth of each construct. ical constructs on which it was built is a critical They go on to suggest that researchers might component of the literature review of any theo- correct for their “employee bias” by adopting a retical manuscript. Theorists need to “acknowl- broader set of assumptions that are “anchored edge the stream of logic on which they are draw- in an organizational frame of reference” (Pierce ing and to which they are contributing” (Sutton et al., 1989: 624). & Staw, 1995: 372). Similarly, theorists also need Because constructs are subject to conditions of to carefully describe the logical connections be- value, researchers must make their best efforts tween the proposed new construct and other ex- to explicate the hidden assumptions that they tant constructs, a process Bacharach (1989) sug- bring to the theorization of a construct. As orga- gests is usually carried out in the form of nizational theorists, we must adopt an ongoing propositions. position of critical reflexivity about how our in- The key observation here is that the clarity of dividual point of view, our often taken-for- a construct is only partly achieved by the preci- granted assumptions, and our institutional biog- sion of its definition. The notion of clarity ex- raphy might introduce bias and distortion into tends beyond this to include clarity in how the how we conceptualize and abstract reality. theorist describes the complex relationships In sum, clearly stating the scope conditions of that exist between constructs. One effective way constructs contributes directly to building of addressing this is for the theorist to demon- strong theory. Whetten (1989) summarizes this strate the historical lineage of a new construct relationship neatly in his description of the four and position that construct on the horizon of essential conditions of a strong theory. A theory extant related constructs. 2010 Editor’s Comments 351

Constructs may be “relational” in a different ship to other constructs must all make sense. sense, however. Some constructs are relational That is, they must all cohere or “hang to- not just because they are derived from other gether” in a logically consistent manner. In constructs but because they are embedded in part, the need for coherence derives from the processes that involve other constructs. That is, inherently multidimensional nature of man- some constructs are processual in nature be- agement research. Most of the constructs we cause they are derived from process data (Lang- use are highly contextually sensitive, and over ley, 1999). Constructs in process theory are qual- time constructs developed in management re- itatively different from constructs derived from search tend to consist of a number of interre- variance theory (Mohr, 1982). Process data are lated attributes or dimensions that may vary inherently “messy,” in part because they are somewhat in different organizational contexts collected in real time through direct observa- but still meaningfully capture a comprehen- tions in the field (Langley, 1999). Constructs de- sive element of organizational experience. As rived from processes, therefore, tend to be rela- a result, constructs often become multidimen- tional inasmuch as they focus on events that are sional. That is, they describe abstract con- multidimensional, temporally embedded, and cepts that are themselves composed of multi- often spanning multiple levels of analysis (Lang- ple attributes. ley, 1999; Van de Ven, 1992). So, for example, So, for example, organizational citizenship some theorists have identified stories or narra- behavior (OCB) is a recognized construct that tives as distinct constructs in process theory has been based on five distinct foundational (Pentland, 1999). elements: civic virtue, sportsmanship, altru- A critical question that arises when assessing ism, conscientiousness, and courtesy (, the clarity of constructs used in process theory is Wong, & Mobley, 1998). Each of these founda- what the tipping point is at which complex pro- tional elements is based on distinct measures cess constructs should be broken down into and may vary somewhat across different or- more concise theoretical units. Addressing this ganizational contexts in terms of its propor- issue is difficult in that it illustrates, in part, how tionate contribution to the “umbrella con- different epistemological assumptions produce struct” of OCB. However, the umbrella different standards of construct clarity (dis- construct retains an overall coherence or con- cussed in more detail below). Some general sistency that is more than the sum of its foun- principles of construct clarity, however, may dational parts. Law et al. (1998) describe this shed some insight here. Notwithstanding the internal coherence of the umbrella construct complexity and multidimensionality of con- as a “latent model” and the summated ele- structs used in process theory, authors should ments as an “aggregate model.” They also use still strive to capture the essential characteris- the term profile model to describe the different tics of phenomena with constructs that balance profiles that occur when the elements vary as accuracy (comprehensiveness) with simplicity a result of different contextual conditions. The (parsimony) and generality (Langley, 1999). Con- key element of their argument, however, is structs derived from process theory might well the understanding that the core construct is be more densely embedded in relation to other greater—that is, more resilient—than its foun- constructs. The goal of the researcher, however, dational elements. This perhaps best illus- should still be to strive for clarity, parsimony, trates the notion of coherence in that a true and precision in capturing the essential ele- multidimensional construct demonstrates ments of the construct and in mapping out the greater resilience than its component ele- relationships between the focal construct and ments. other constructs within which the focal construct Often, the issue presented by questions of co- is embedded. herence is the ability of the theorist to use the constructs to create logically consistent and the- oretically integrated arguments. As Sutton and Coherence Staw (1995) have observed, theorists will regu- A final component of construct clarity is the larly offer up complex schematics or intricate notion that the construct, its definition, its process flow diagrams in lieu of coherence. The scope conditions, its lineage, and its relation- diagrams are a good start, but ultimately, the 352 Academy of Management Review July power of coherence of such an assembly of con- search. I elaborate each of these points in the structs can only be provided by a compelling balance of this section. and coherent explanation. Coherence is a difficult characteristic of con- struct clarity to explain, in part because it re- Clarity Facilitates veals the recursive or dialectical relationship that exists between constructs and theory. That Construct clarity allows us to build on prior is, in large part constructs gain their coherence, research by providing the research community both internally and in relation to other con- with a common language. A common language structs, as a result of the theory in which they is an essential prerequisite for a community of are embedded. It is difficult to understand the scholars interested in the same or similar phe- construct of legitimacy independent of one’s nomena to exchange ideas and build knowl- knowledge of institutional theory (Suchman, edge. The ability to precisely articulate the key 1995). Kaplan calls this the “paradox of concep- elements that underpin an idea helps us to un- tualization,” noting that “proper concepts are derstand the degree to which ideas overlap or needed to formulate good theory, but we need a differ. Moreover, the advancement of theory and good theory to arrive at proper concepts” (1964: knowledge relies on the ability of new research- 501). ers to build on the of prior researchers. If Coherence, thus, is a somewhat intuitive as- new and old researchers cannot agree on or sessment of whether the various attributes of a communicate the basic elements of a phenome- phenomenon are adequately contained within a non, the accumulation of knowledge cannot construct—that is, do these attributes hang to- occur. gether in a logical and empirically convincing The ability to precisely define the essence of way? Is the construct plausible, given one’s ex- an abstraction in such a way that differentiates perience in similar contexts (Weick, 1989)? Are it from other similar abstractions produces seri- the relationships described or implied by the ous advantages for a scholarly community. construct plausible? Does the construct make Foremost, it avoids the proliferation of different sense? terms and labels for similar phenomena—a Collectively, these four characteristics (defini- problem that is often colloquially described as tions, scope conditions, relationships between putting “old wine in new bottles.” Truman constructs, and coherence) capture the essential Kelley, one of the founders of the Stanford elements of construct clarity. It is perhaps trite Achievement Test, called the proliferation of dif- to note that the characteristics mutually rein- ferent labels for the same underlying construct force each other. It is difficult, for example, to the “jangle fallacy.” Kelley (1927) observed that demonstrate scope conditions without first pro- when researchers used different words—“intel- viding a sound definition. Similarly, coherence ligence” or “achievement”—to describe the is highly dependent on clearly stated scope con- same underlying construct of general intelli- ditions. Construct clarity requires considerable gence, there was a tendency to start treating the skill in crafting each of these essential terms as completely different constructs, even elements. though the overlap in individual differences that underpin the two terms was over 90 percent (Lubinski, 2004). When researchers use different terms for sim- WHY DO WE NEED CONSTRUCT CLARITY? ilar phenomena, it produces confusion—“con- There are three main justifications for clear founding effects”—that impede the ability of and concise constructs, each of which builds on members of a research community to communi- the core idea that such clarity is critical to the cate with each other or to accumulate knowl- accumulation of knowledge. First, clear con- edge. The creation of a common vocabulary structs facilitate communication between schol- avoids the “Tower of Babel” effect, in which sub- ars. Second, improved clarity of constructs en- communities of researchers have no common hances researchers’ ability to empirically means of communication. In the absence of com- explore phenomena. Third, clear constructs al- mon and well-articulated constructs, the bound- low for greater creativity and in re- aries between subcommunities become more 2010 Editor’s Comments 353 sharply defined and organizational knowledge sence of empiricism is the ability to create clear becomes increasingly fragmented. classifications of phenomena that structure ex- Clear constructs can and should also extend perience into meaningful categories (Hacking, the scope of knowledge beyond the academic 1975). Clear constructs are simply robust catego- community to include practitioners. Manage- ries that organize experience. Moreover, clear ment scholars have expressed considerable dis- constructs help researchers identify anomalies may about the failure of academic research to or phenomena that defy categories and force penetrate the practitioner community (Rynes, researchers to reevaluate their theories. 2007; Rynes, Bartunek, & Daft, 2001). In part, this is the result of weakly articulated constructs or Construct Clarity Enhances Creativity constructs that are so narrowly defined they lack relevance. Effective constructs, however, Clearly defined theoretical constructs serve a can help immensely in bridging that gap. As creative heuristic purpose in the elaboration of Astley and Zammuto note, it is at the abstract theory. Like metaphors, a well-crafted construct level of ideational constructs “where much of can capture the essential elements or character- the between scientific and istics of a phenomenon and, simultaneously, practitioner domains occurs” (1992: 444). An ef- highlight both its similarities to and differences fective construct, thus, navigates a narrow path from related phenomena. Constructs are care- between definitional accuracy and communica- fully articulated abstractions that, if effectively ble generality—that is, it is precisely and ac- crafted, expand the range of phenomena and curately constructed, but in a way that broad relationships they capture. Effective constructs, audiences can understand and participate in thus, can enhance research creativity by “allow- the process of empirical elaboration and ing managers to redefine problems in ways that exploration. are more amenable to resolution” (Astley & Zam- muto, 1992: 455). Constructs are conceptual frames, and clear constructs expose a phenom- Clarity Assists Empirical Analysis enon to multiple perspectives. Construct clarity aids in the empirical appli- A clear construct, thus, not only serves as a cation of theory. For positivists, construct clarity useful means of description but can stimulate helps them test theory, since precisely defined insights into additional possible relationships, constructs are easier to operationalize and test related constructs, and often related theories. (Schwab, 1980) and it is easier for researchers to Like a well-chosen metaphor, a carefully crafted compare and contrast results (Bagozzi & Ed- construct is a powerful creative tool that en- wards, 1998). For constructivists, construct clar- hances theory development. ity is not intended to lead to precise operation- alization and measurement but is still critical HOW DOES CONSTRUCT CLARITY VARY? (perhaps even more essential) for capturing and communicating with precision the often subjective Thus far, I have presented the notion of con- meaning and interpretation of an abstraction by struct clarity in a somewhat catholic fashion— individual subjects. Berger and Luckmann, for that is, with an implicit assumption that its example, argue that construct clarity helps con- importance and basic characteristics are uni- structivist researchers avoid the positivist di- versally accepted across the various epistemo- lemma of reification or “confusing its own con- logical and ontological regimes that comprise ceptualizations with the of the universe” the Academy of Management scholars. Clearly, (1967: 187). Similarly, the success of grounded that is not the case, and at various points in the theory research lies, in large part, in the ability discussion above, I have made some effort to of researchers to clearly identify and express foreshadow that, for example, positivists and “concepts or constructs that are grounded in ac- social constructionists might hold different tors’ meaning-in-use, rather than categories or views as to what might constitute a good defi- constructs that are imposed by the researcher” nition. Indeed, the term construct itself is likely (Bob Gephart, personal communication). to be contested by nonpositivists, based on the In sum, careful description of phenomena is connotations of hypothesis testing and opera- the fundament of empirical exploration. The es- tionalization typically associated with the term. 354 Academy of Management Review July

For researchers using a constructivist perspec- disconfirm or build upon existing theories....We tive, concept might be a more acceptable value- argue that the classic case study approach has been extremely powerful because these authors neutral term. have described general phenomena so well that Different traditions of research have very dif- others have little difficulty seeing the same phe- ferent understandings of what construct clarity nomena in their own experience and research. is and how constructs might best be used in We return to the classics because they are good building theory. Let me illustrate through two stories, not because they are merely clear state- ments of a construct. Indeed the very clarity of the examples. The first comes from Eisenhardt’s constructs stems from the story that supports and (1989) classic paper on how to use case studies demonstrates them (1991: 617). to build theory. Here Eisenhardt adopts a very positivist view of what constructs are and how On its surface, the debate between Eisenhardt they should be used in building theory: and Dyer and Wilkins seems to reflect two op- posing and irreconcilable epistemological posi- A priori specification of constructs can also help to shape the initial design of theory-building re- tions. On closer examination, however, they search. Although this type of specification is not simply illustrate different assumptions regard- common in theory-building studies to date, it is ing the role of constructs in the research process. valuable because it permits researchers to mea- Eisenhardt (1989) sees constructs as lenses sure constructs more accurately. If these con- through which data can be analyzed in the the- structs prove important as the study progresses, then researchers have a firmer empirical ground- ory-building process. Dyer and Wilkins (1991) ing for the emergent theory (1989: 536). see constructs as emerging from the data. Both, however, seem to acknowledge the need for Note that Eisenhardt sees constructs as essen- clear constructs; they simply differ on their role tial to theory building but acknowledges that in the process of building theory. the researcher will bring preexisting constructs In fact, Eisenhardt (1989) and Dyer and Wilkins into the research to be “tested” through empiri- (1991) are assuming two distinct roles in an on- cal application. Eisenhardt qualifies this posi- going dialectic or tension over theoretical con- tion somewhat with an admonition to the re- structs that Hirsch and Levin (1999) describe as searcher to keep an open mind about the the “umbrella advocates” versus the “validity possibility of refining the construct in accor- police.” The term umbrella advocates refers to dance with the data as the research progresses: those researchers who argue that constructs Although early identification of the research should be viewed as large buckets or broad con- question and possible constructs is helpful, it is cepts loosely defined because this better cap- equally important to recognize that both are ten- tures the inherent complexity and messiness of tative in this type of research. No construct is the empirical world we study. The term validity guaranteed a place in the resultant theory, no matter how well it is measured (1989: 536). police refers to those researchers who argue that constructs should be small buckets nar- Eisenhardt’s (1989) view of the role of con- rowly defined in order to bring more scientific structs in theory building is not universally ac- rigor and validity to the study of organizations. cepted and was subsequently challenged by re- Hirsch and Levin (1999) argue that the tension searchers who thought that entering the field between these two regimes creates a distinct with narrowly defined constructs would inter- life cycle for theoretical constructs in organiza- fere with the researcher’s ability to create new tion studies, where umbrella advocates first constructs or enrich our understanding of exist- introduce a new construct, which then succumbs ing ones. Drawing from a research tradition that to demands from the validity police to “clean up encourages more creativity and flexibility in the the concept” (for a recent example of this stage, research process and one that might adopt the see Briner, Denyer, & Rousseau, 2009). In some term concept in place of construct, in their re- instances constructs become so clearly defined, joinder to Eisenhardt, Dyer and Wilkins suggest measurable, and operationalized over time that that researchers ought to aim for “good stories” they lose relevance with the empirical world rather than “good constructs”: and, ultimately, reappear under a different Eisenhardt’s approach [is] to start with . . . con- name. Hirsch and Levin (1999) illustrate this phe- structs and measurement instruments. Such an nomenon through an analysis of the emergence, approach leads the case researcher to confirm, clarification, and disappearance of the theoret- 2010 Editor’s Comments 355 ical construct “organizational effectiveness.” lined in this essay still applies. Phenomenolo- Hirsch and Levin (1999) remind us that this gists may not agree with the notion that con- construct was itself a replacement for a prior structs must be made measureable, but they similar construct called “organizational would not argue with the notion that the con- performance.” cepts they derive from their research must be In contrast to the evident disagreement in the communicated clearly, with appropriate limit- debate between Eisenhardt (1989) and Dyer and ing conditions and assumptions and with some Wilkins (1991) about the proper role of theoreti- explanation of how these concepts fit in relation cal constructs, Hirsch and Levin argue that this to other concepts used in similar research. Sim- tension between broad and narrow interpreta- ilarly, while a researcher using tions of constructs is not only healthy but is might be striving for novel insights from his or necessary for the advancement of knowledge: her data, when writing his or her theory, the researcher must still bear the burden of demon- Though each of us may have his or her own lean- strating how his or her insights fit on the horizon ings, the field as a whole probably needs both broad (umbrella) and narrow (policing) perspec- of prior knowledge of the subject under study, tives, for this dialectic can be useful for under- even those drawn from other research traditions. standing and explaining the underlying issues of organizational life. This struggle thus enables the field as a whole to balance its competing WHY WE NEED AN ONGOING needs to be both scientific and relevant (1999: CONVERSATION ABOUT THEORETICAL 209). CONSTRUCTS IN ORGANIZATION STUDIES The standards for the meaning and use of con- I have tried to demonstrate how construct clar- structs, thus, appear to vary considerably across ity lies at the heart of theory building. Clearly different research traditions, epistemologies, defined conceptual categories encourage re- and ontological positions within organization searchers to generate more effective research studies. questions, apply appropriate and epistemologi- While different research traditions may have cally consistent methods, and identify excep- different interpretations of how constructs are tions to the categories that open opportunities constituted and how they should be used in re- for future research. All of this serves to search, the need for clarity and precision in the strengthen our understanding of phenomena. description of constructs remains intact. For pos- Construct clarity also aids in the communica- itivists, precise language is necessary to cap- tion and accumulation of knowledge. Clear con- ture, as effectively as possible, the essence of ceptual categories can help overcome fragmen- the subject matter under study. The challenge is tation in the field, make our research relevant to to use language to create constructs that accu- broader audiences, and enhance the legitimacy rately represent reality. For nonpositivists, pre- of management as a research discipline. cise language is equally important, not to cap- What surprises me is how unusually mute our ture reality or to enable the measurement of discipline seems to be on so important a subject. constructs but, rather, to recognize that linguis- While management journals devote some space tic constructs are themselves the embodiment of to discussions of constructs and their role in knowledge (Gergen, 1982). As Astley observes, theory development (i.e., Astley & Zammuto, for nonpositivists, “Language is not simply a 1992; Hirsch & Levin, 1999; Preim & Butler, 2001), vehicle for transmitting information. Rather it is the coverage seems disproportionate to the im- the very embodiment of truth.... Scientific portance of the topic. This anomaly is reflected, fields are word created and maintained somewhat, in how we train graduate students, through a process of negotiation between adher- where considerable time is devoted to under- ents to alternative theoretical languages” (1985: standinghowconstructsaremeasuredandopera- 499). tionalized but substantially less time is devoted So, while the interpretation of what a con- to understanding how constructs are created struct means might vary across subdisciplines and used in the research process. of organizational research, I believe that the One clear conclusion from this essay is that requirement for clarity of description as well as construct clarity is highly dependent on a theo- the four key elements of construct clarity out- rist’s facility with language. Good constructs ef- 356 Academy of Management Review July fectively balance some competing (if not contra- Design and analysis issues for field setting. Chicago: dictory) tensions. So, for example, constructs Rand McNally. must strip away surplus meaning but not be Cronbach, L. J., & Meehl, P. E. 1955. Construct validity in made too narrow. Constructs should offer clear psychological tests. Psychological Bulletin, 52: 281–302. boundaries and scope conditions but also be Dubin, R. 1969. Theory building. New York: Free Press. sufficiently “linguistically ambiguous” to spark Dyer, W. G., & Wilkins, A. L. 1991. Better stories, not better new connotative relationships. These are no constructs, to generate better theory: A rejoinder to small tasks, even for those trained in the skillful Eisenhardt. Academy of Management Review, 16: 613– use of language. Yet even though we under- 619. stand that language matters to effective theory Eisenhardt, K. 1989. Building theory from case study re- development in the same way it matters to phi- search. Academy of Management Review, 14: 532–550. losophy (Hacking, 1975), we devote considerably Empson, W. 1995. (First published in 1951.) The structure of less time to training new researchers in under- complex words. London: Penguin Books. standing the nuances of language than we do to Floyd, S. W. 2009. Borrowing theory: What does this mean understanding the nuances of statistical mea- and when does it make sense in management scholar- surement. ship? Journal of Management Studies, 46: 1059–1075. Our silence on the subject of use of constructs George, J. M., & Jones, G. R. 2000. The role of time in theory in management may be a pragmatic effort to and theory building. Journal of Management, 26: 657– avoid the “paradigm wars” of the past, or it may 684. simply reflect the ongoing fragmentation of our Gergen, K. J. 1982. Toward transformation in social knowl- field. My hope is that this essay will not only edge. New York: Springer. help aspiring theorists understand how to more Gibson, C. B., & Zellmer-Bruhn, M. E. 2001. Metaphors and effectively develop constructs but will also re- meaning: An intercultural analysis of the concept of new the conversation on constructs in manage- teamwork. Administrative Science Quarterly, 46: 274– 303. ment theories and focus discussion on tech- niques for improving their clarity, increasing Hacking, I. 1975. Why does language matter to philosophy? Cambridge: Cambridge University Press. our understanding of their role in bridging re- search disciplines, and improving both the rel- Hansen, G. S., & Wernerfeldt, B. 1989. Determinants of firm evance and rigor of organizational research. performance: The relative importance of economic and organizational factors. Journal, 10: 399–411. REFERENCES Hirsch, P. M., & Levin, D. Z. 1999. Umbrella advocates versus validity police: A life-cycle model. Organization Sci- Astley, W. G. 1985. Administrative science as socially con- ence, 10: 199–212. structed truth. Administrative Science Quarterly, 30: 497– Kaplan, A. 1964. The conduct of inquiry. San Francisco: 513. Chandler. Astley, W. G., & Zammuto, R. F. 1992. Organization science, Kelley, T. L. 1927. Interpretation of educational measure- managers and language games. Organization Science, ments. New York: World Book. 3: 443–460. Kerlinger, F. N. 1973. Foundations of behavioral research. Avital, M. 2000. Dealing with time in social inquiry: A tension New York: Reinhart & Winston. between method and lived experience. Organization Science, 11: 665–673. Klein, K. J., Dansereau, F., & Hall, R. J. 1994. Levels issues in Bacharach, S. B. 1989. Organizational theories: Some criteria theory development, data collection, and analysis. for . Academy of Management Journal, 14: Academy of Management Review, 19: 195–229. 496–515. Langley, A. 1999. Strategies for theorizing from process data. Bagozzi, R. P., & Edwards, J. R. 1998. A general approach for Academy of Management Review, 24: 691–710. representing constructs in organizational research. Or- Law, K. S., Wong, C., & Mobley, W. H. 1998. Toward a taxon- ganizational Research Methods, 1: 45–87. omy of multidimensional constructs. Academy of Man- Berger, P. L., & Luckmann, T. 1967. The social construction of agement Review, 23: 741–755. reality: A treatise on the sociology of knowledge. New Lubinski, D. 2004. Introduction to the special section on cog- York: Anchor Books. nitive abilities: 100 years after Spearman’s (1904) “Gen- Briner, R. B., Denyer, D., & Rousseau, D. M. 2009. Evidence- eral intelligence,” objectively determined and mea- based management: Concept cleanup time? Academy of sured. Journal of Personality and , 86: Management Perspectives, 23(4): 19–32. 96–111. Cook, T. D., & Campbell, D. T. 1979. Quasi-experimentation: MacCorquodale, K., & Meehl, P. E. 1948. On a distinction 2010 Editor’s Comments 357

between hypothetical constructs and intervening vari- Shane, S., & Venkataraman, S. 2000. The promise of entre- ables. Psychological Review, 55: 95–107. preneurship as a field of research. Academy of Manage- Mitchell, R. K., Agle, B. R., & Wood, D. J. 1997. Toward a theory ment Review, 25: 217–226. of stakeholder identification and salience: Defining the Sharma, P., Chrisman, J. J., & Chua, J. H. 1996. A review and principle of who and what really counts. Academy of annotated bibliography of family . Bos- Management Review, 22: 853–886. ton: Kluwer Academic. Mohr, L. B. 1982. Explaining . San Shenkar, O., & von Glinow, M. A. 1994. Paradoxes of orga- Francisco: Jossey-Bass. nizational theory and research: Using the case of China Pentland, B. T. 1999. Building process theory with narrative: to illustrate national contingency. Management Sci- From description to explanation. Academy of Manage- ence, 40: 56–69. ment Review, 24: 711–724. Suchman, M. C. 1995. Managing legitimacy: Strategic and Pierce, J. L., Gardner, D. G., Cummings, L. L., & Dunham, R. B. institutional approaches. Academy of Management Re- 1989. Organization-based self-esteem: Construct defini- view, 20: 571–610. tion, measurement, and validation. Academy of Man- Sutton, R. I., & Staw, B. M. 1995. What theory is not. Admin- agement Journal, 32: 622–648. istrative Science Quarterly, 40: 371–384. Priem, R. L., & Butler, J. E. 2001. Tautology in the - Van de ven, A. H. 1992. Suggestions for studying strategy based view and the implications of externally deter- process: A research note. Strategic Management Jour- mined resource value: Further comments. Academy of nal, 13: 169–188. Management Review, 26: 57–66. Walker, H. A., & Cohen, B. P. 1985. Scope statements: Imper- Rogers, E. W., & Wright, P. M. 1998. Measuring organization- atives for evaluating theory. American Sociological Re- al performance in strategic human resource manage- view, 50: 288–301. ment: Problems, prospects and performance informa- Weick, K. E. 1989. Theory construction as disciplined imagi- tion markets. Human Review, nation. Academy of Management Review, 14: 516–531. 8: 311–331. Whetten, D. A. 1989. What constitutes a theoretical contribu- Rousseau, D. M. 1985. Issues of level in organizational tion? Academy of Management Review, 14: 490–495. research: Multi-level and cross level perspectives. Re- search in Organizational Behavior, 7: 1–37. Whetten, D. A., Felin, T., & King, B. 2009. The practice of theory borrowing in : Current issues and Rynes, S. L. 2007. Let’s create a tipping point: What academ- future directions. Journal of Management, 35: 537–563. ics and practitioners can do, alone and together. Acad- emy of Management Journal, 50: 1046–1054. White, S. 2002. Rigor and relevance in Asian management research: Where are we and where can we go? Asia- Rynes, S. L., Bartunek, J. M., & Daft, R. L. 2001. Across the Pacific Journal of Management, 19: 287–352. great divide: Knowledge creation and transfer between practitioners and academics. Academy of Management Zaheer, S., Albert, S., & Zaheer, A. 1999. Time scales and Journal, 44: 340–355. . Academy of Management Re- view, 24: 725–741. Saussure, F. 2000. Course in general . Peru, IL: Open Court. Schwab, D. P. 1980. Construct validity in organizational be- Roy Suddaby havior. Research in Organizational Behavior,2:3–43. Associate Editor Copyright of Academy of Management Review is the property of Academy of Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.