METHODOLOGICAL FIT IN MANAGEMENT FIELD … · METHODOLOGICAL FIT IN MANAGEMENT FIELD RESEARCH AMY...

METHODOLOGICAL FIT IN MANAGEMENTFIELD RESEARCH

AMY C. EDMONDSONHarvard Business School

STACY E. MCMANUSMonitor Executive Development

Methodological fit, an implicitly valued attribute of high-quality field research inorganizations, has received little attention in the management literature. Fit refers tointernal consistency among elements of a research project—research question, priorwork, research design, and theoretical contribution. We introduce a contingencyframework that relates prior work to the design of a research project, paying particularattention to the question of when to mix qualitative and quantitative data in a singleresearch paper. We discuss implications of the framework for educating new fieldresearchers.

To advance management theory, a growingnumber of scholars are engaging in field re-search, studying real people, real problems, andreal organizations. Although the potential rele-vance of field research is motivating, the re-search journey can be messy and inefficient,fraught with logistical hurdles and unexpectedevents. Researchers manage complex relation-ships with sites, cope with constraints on sam-ple selection and timing of data collection, andoften confront mid-project changes to plannedresearch designs. With these additional chal-lenges, the logic of a research design and how itsupports the development of a specific theoreti-cal contribution can be obscured or alteredalong the way in field research. Compared toexperimental studies, analyses of publisheddata sets, or computer simulations, achieving fitbetween the type of data collected in and thetheoretical contribution of a given field researchproject is a dynamic and challenging process.

This article introduces a framework for as-sessing and promoting methodological fit as anoverarching criterion for ensuring quality fieldresearch. We define methodological fit as inter-nal consistency among elements of a researchproject (see Table 1 for four key elements of fieldresearch). Although articles based on field re-search in leading academic journals usually ex-hibit a high degree of methodological fit, guide-lines for ensuring it are not readily available.Beyond the observation that qualitative data areappropriate for studying phenomena that arenot well understood (e.g., Barley, 1990; Bouchard,1976; Eisenhardt, 1989a), the relationship be-tween types of theoretical contributions andtypes of field research has received little ex-plicit attention. In particular, the conditions un-der which hybrid methods that mix qualitativeand quantitative data are most helpful in fieldresearch—a central focus of this paper—are notwidely recognized.

We define field research in management assystematic studies that rely on the collection oforiginal data—qualitative or quantitative—inreal organizations. The ideas in this paper arenot intended to generalize to all types of man-agement research but, rather, to help guide thedesign and development of research projectsthat centrally involve collecting data in fieldsites. We offer a framework that relates thestage of prior theory to research questions, typeof data collected and analyzed, and theoreticalcontributions—the elements shown in Table 1.

We thank David Ager, Jim Detert, Robin Ely, RichardHackman, Connie Hadley, Bertrand Moingeon, WendySmith, students in four years of the Design of Field ResearchMethods course at Harvard, seminar participants at the Uni-versity of Texas McCombs School, the MIT OrganizationStudies group, and the Kurt Lewin Institute in Amsterdam forvaluable feedback in the development of these ideas. Weare particularly grateful to Terrence Mitchell and the AMRreviewers for suggestions that improved the paper im-mensely. Harvard Business School Division of Research pro-vided the funding for this project.

� Academy of Management Review2007, Vol. 32, No. 4, 1155–1179.

1155Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyrightholder’s express written permission. Users may print, download, or email articles for individual use only.

In well-integrated field research the key ele-ments are congruent and mutually reinforcing.

The framework we present is unlikely to callfor changes in how accomplished field research-ers go about their work. Indeed, experiencedresearchers regularly implement the alignmentwe describe. However, new organizational re-searchers, or even accomplished experimental-ists or modelers who are new to field research,should benefit from an explicit discussion of themutually reinforcing relationships that promotemethodological fit.

The primary aim of this article, thus, is toprovide guidelines for helping new field re-searchers develop and hone their ability toalign theory and methods in field research.Because a key aspect of this is the ability toanticipate and detect problems that emergewhen fit is low, our discussion explores andcategorizes such problems. A second aim is tosuggest that methodological fit in field re-search is created through an iterative learningprocess that requires a mindset in which feed-back, rethinking, and revising are embracedas valued activities, and to discuss the impli-cations of this for educating new field re-searchers. To begin, in the next section wesituate our efforts in the broader methodolog-

ical literature and describe the sources thatinform our ideas.

BACKGROUND

Prior Work on Methodological Fit

The notion of methodological fit has deeproots in organizational research (e.g., Bouchard,1976; Campbell, Daft, & Hulin, 1982; Lee, Mitch-ell, & Sablynski, 1999; McGrath, 1964). Years ago,McGrath (1964) noted that the state of priorknowledge is a key determinant of appropriateresearch methodology. Pointing to a full spec-trum of research settings, ranging from field re-search to experimental simulations, laboratoryexperiments, and computer simulations, he pre-sented field studies as appropriate for explor-atory endeavors to stimulate new theoreticalideas and for cross-validation to assess whetheran established theory holds up in the real world.The other, non-field-based research settingswere presented as appropriate for advancingtheory. Understandably, given the era, McGrathdid not dig deeply into the full range of methodsthat have since been used within field researchalone.

Subsequently, Bouchard, focusing on how toimplement research techniques such as inter-

TABLE 1Four Key Elements of a Field Research Project

Element Description

Research question ● Focuses a study● Narrows the topic area to a meaningful, manageable size● Addresses issues of theoretical and practical significance● Points toward a viable research project—that is, the question can be

answered

Prior work ● The state of the literature● Existing theoretical and empirical research papers that pertain to the

topic of the current study● An aid in identifying unanswered questions, unexplored areas,

relevant constructs, and areas of low agreement

Research design ● Type of data to be collected● Data collection tools and procedures● Type of analysis planned● Finding/selection of sites for collecting data

Contribution to literature ● The theory developed as an outcome of the study● New ideas that contest conventional wisdom, challenge prior

assumptions, integrate prior streams of research to produce a newmodel, or refine understanding of a phenomenon

● Any practical insights drawn from the findings that may be suggestedby the researcher

1156 OctoberAcademy of Management Review

views, questionnaires, and observation, noted,“The key to good research lies not in choosingthe right method, but rather in asking the rightquestion and picking the most powerful methodfor answering that particular question” (1976:402). Others have issued cautions against as-suming the unilateral rightness of a method—wielding a hammer and treating everything asnails (e.g., Campbell et al., 1982). Yet all re-searchers are vulnerable to preferring thosehammers that we have learned to use well.Thus, we benefit from reminders that not alltools are appropriate for all situations. At thesame time, exactly how to determine the rightmethod for a given research question—particu-larly in the field—has not been as well speci-fied.

More recently, Lee et al. (1999: 163) tackled thechallenges of research in “natural settings” toexplicate strategies for effective qualitative or-ganizational and vocational research. Using ex-emplars, these authors showed that qualitativedata are useful for theory generation, elabora-tion, and even testing, in an effort to “inspire[other researchers] to seek opportunities to ex-pand their thinking and research” and to helpthem “learn from this larger and collective ex-perience and avoid misdirection” (1999: 161). Inadvocating the benefits of qualitative work fororganizational researchers, these authors pro-vide a helpful foundation for the present paper.We build on this work by distinguishing amongpurely qualitative, purely quantitative, and hy-brid designs, as well as by including a fullerrange of field research methods in a singleframework. The categories we develop allow amore fine-grained analysis of field research op-tions than offered previously.

A recent body of work debates the appropri-ateness of combining qualitative and quanti-tative methods within a single researchproject. Issues addressed in this debate in-clude whether qualitative and quantitativemethods investigate the same phenomena, arephilosophically consistent, and are paradigmsthat can reasonably be integrated within astudy (e.g., Greene, Caracelli, & Graham, 1989;Morgan & Smircich, 1980; Sale, Lohfeld, & Bra-zil, 2002; Yauch & Steudel, 2003). Consistentwith Yauch and Steudel (2003), who provide abrief review of the current thinking on thistopic, we propose that the two methods can becombined successfully in cases where the

goal is to increase validity of new measuresthrough triangulation1 and/or to generategreater understanding of the mechanisms un-derlying quantitative results in at least par-tially new territory. This paper complementsprior work on hybrid methods by addressinghow the state of current theory and literatureinfluences not only when hybrid researchstrategies are appropriate but also when othermethodological decisions are appropriate andhow different elements of research projects fittogether to form coherent wholes.

Sources for Understanding Fit in FieldResearch

Several sources have informed the ideas pre-sented in this paper. A long-standing interest inteaching field research methods fueled exten-sive note taking, reflection, and iterative modelbuilding over the past decade. In this reflectiveprocess we drew first from the many high-quality papers reporting on field research pub-lished in prominent journals; we use a few ofthese articles as exemplars to highlight and ex-plain our framework. Second, we drew from ourown experiences conducting field research,complete with missteps, feedback, and exten-sive refinement. Third, the first author’s experi-ence reviewing dozens of manuscripts reportingon field research submitted to academic jour-nals provided additional insight into both thepresence and absence of methodological fit.2

Unlike reading polished published articles, re-viewing offers the advantage of being able toobserve part of the research journey. Moreover,a reviewer’s reward is the opportunity to seehow other anonymous reviewers have evalu-ated the same manuscript—constituting an in-formal index of agreement among expert judges.Papers rejected or returned for extensive revi-sion because a poor match among prior work,

1 Triangulation is a process by which the same phenom-enon is assessed with different methods to determinewhether convergence across methods exists. See Jick (1979)for a thoughtful discussion.

2 These reviewing experiences were important inputs intothe framework in this paper; however, the confidentiality ofthe review process precluded using these cases as exam-ples. To illustrate poor fit and attempts to improve fit later inthe article, we resorted to drawing on our second primarysource—our own field research projects.

2007 1157Edmondson and McManus

research questions, and methods helped informour framework; agreement among expert re-viewers strengthens our confidence in theseideas.

This agreement is not explained by explicitinstruction. A glance at the current Academy ofManagement Journal and Administrative Sci-ence Quarterly checklists for reviewers revealsan emphasis on the quality of the individualelements of a submission—for example, “tech-nical adequacy”—without a formal criterion forevaluating fit among elements. Yet researchersmay employ a particular method exceptionallywell, without it being an effective approach tostudying the stated research question. This hap-pens, in part, because field research is oftenspurred by unexpected data collection opportu-nities. Responding to requests from contacts atcompanies, researchers may collect data drivenby company interests but not well matched toinitial research questions. For example, surveysmay be distributed that help the site but thathave limited connection to the researcher’s the-oretical goals. Similarly, interview data from aconsulting project may be reanalyzed for re-search, focusing on an area of theory not wellsuited to purely qualitative research. The oppor-tunistic aspect of field research is not in itself aweakness but may increase the chances of poormethodological fit when data collected for onereason are used without careful thought for an-other.

The experience of reviewing also highlightsthat a lack of methodological fit is easier todiscern in others’ field research than in one’sown. This motivated us to develop a formalframework to help researchers uncover areas ofpoor fit in their own field research earlier in theresearch journey, without waiting for externalreview.

Drawing on the above sources, we inductivelyderived the framework presented in this paper,revising it along the way, driven by each otherand by colleagues and reviewers both close anddistant. In exploring methodological fit, we areparticularly focused on how the state of currenttheory shapes other elements of a field researchproject. For clarity of illustration and compari-sons across diverse methods, we limit the sub-stantive topic of the research projects discussedto one area—organizational work teams.

In the next section we show that producingmethodological fit depends on the state of rele-

vant theory at the time the research is designedand executed. We use the state of prior theory asthe starting point in achieving methodologicalfit in field research because it serves as a given,reasonably fixed context in which new researchis developed: it is the one element over whichthe researcher has no control (i.e., the state ofextant theoretical development cannot be mod-ified to fit the current research project).

A CONTINGENCY FRAMEWORK FORMANAGEMENT FIELD RESEARCH

The State of Prior Theory

We suggest that theory in management re-search falls along a continuum, from mature tonascent. Mature theory presents well-developedconstructs and models that have been studiedover time with increasing precision by a varietyof scholars, resulting in a body of work consist-ing of points of broad agreement that representcumulative knowledge gained. Nascent theory,in contrast, proposes tentative answers to novelquestions of how and why, often merely sug-gesting new connections among phenomena. In-termediate theory, positioned between matureand nascent, presents provisional explanationsof phenomena, often introducing a new con-struct and proposing relationships between itand established constructs. Although the re-search questions may allow the development oftestable hypotheses, similar to mature theoryresearch, one or more of the constructs involvedis often still tentative, similar to nascent theoryresearch.

This continuum is perhaps best understood asa social construction that allows the develop-ment of archetypes. Consequently, it is not al-ways easy to determine the extent of theory de-velopment informing a potential researchquestion.3 We propose a continuum rather than

3 We thank an anonymous reviewer for pointing this outand Terry Mitchell for suggesting how we might address thisissue. To gain insight into raters’ agreement on this catego-rization approach, we prepared short descriptions of four-teen research questions that each began with a brief sum-mary of the state of prior work on the topic. The fourteencases included the articles described in this paper, alongwith a few additional field research studies. We then askedfour organizational researchers to categorize them accord-ing to definitions of the three stages of theory we provided.The average overall agreement with our intended classifi-


clear stages to acknowledge that the categorieswe suggest are not obvious or inviolable and torecognize the potential for debate on the statusof prior work related to a given research ques-tion. In short, our aim is to help field researchersthink about methodological fit in a more ex-plicit, systematic way, using exemplars from theorganizational literature to illustrate how thestate of current theory informs methodologicaldecisions.

Developing Sensible Connections to PriorWork

In a given field study, the four elements inTable 1 should be influenced by the stage ofdevelopment of the current literature at the timeof the research. In general, the less known abouta specific topic, the more open-ended the re-search questions, requiring methods that allowdata collected in the field to strongly shape theresearcher’s developing understanding of thephenomenon (e.g., Barley, 1990). In contrast,when a topic of interest has been studied exten-sively, researchers can use prior literature toidentify critical independent, dependent, andcontrol variables and to explain general mech-anisms underlying the phenomenon. Leverag-ing prior work allows a new study to addressissues that refine the field’s knowledge, such asidentifying moderators or mediators that affecta documented causal relationship. Finally,when theory is in an intermediate stage of de-velopment—by nature a period of transition—anew study can test hypotheses and simulta-neously allow openness to unexpected insightsfrom qualitative data. Broadly, patterns of fitamong research components can be summa-rized as in Table 2.

We begin our more detailed exploration of fitbetween theory and method with a discussion ofmature theory, because it conforms to tradi-tional models of research methodology and soserves as a conceptual base with which to com-pare the other two categories. By drawing pri-marily on the topic of work teams, we demon-strate that the state of prior knowledge for

specific research questions within one broadtopic can vary from mature to nascent.

Mature Theory Research

Mature theory encompasses precise models,supported by extensive research on a set of re-lated questions in varied settings. Maturitystimulates research that leads to further refine-ments within a growing body of interrelated the-ories. The research is often elegant, complex,and logically rigorous, addressing issues thatother researchers would agree from the outsetare worthy of study. Research questions tend tofocus on elaborating, clarifying, or challengingspecific aspects of existing theories. A re-searcher might, for example, test a theory in anew setting, identify or clarify the boundaries ofa theory, examine a mediating mechanism, orprovide new support for or against previouswork.

Specific testable hypotheses are developedthrough logical argument that builds on priorwork. Researchers draw from the literature toargue the need for a new study and to developthe logic underlying the hypotheses they willtest. This hypothesis-testing approach examinesrelationships between previously developedconstructs (and variables) to produce variancetheory (an increase in some X is associated withan increase in some Y; Mohr, 1982). Although themost compelling test of a theory may be exper-imental (e.g., Campbell & Stanley, 1963), fieldresearchers usually cannot manipulate inde-pendent variables randomly across units. Re-search questions and designs thus utilize corre-lation-based analyses consistent with causalinferences supported by logic (e.g., while a per-son’s sex may predict salary level, it would benonsensical to assert the reverse). These studiesrely heavily on statistical analyses and infer-ences to support new theoretical propositions.4

Many excellent examples of published workcould be used to illustrate fit in mature theoryresearch. Stewart and Barrick’s (2000) research

cation was 86 percent, with seven of the fourteen researchquestions achieving 100 percent accuracy and agreement;the raters also had 86 percent overall agreement with eachother.

4 Research explaining team effectiveness, boasting manyempirical studies providing statistical support for consistentexplanatory models, fits this category. See, for example,Hackman (1987). Multiple empirical studies lend support tothe basic model (e.g., Campion, Medsker, & Higgs, 1993;Cohen & Ledford, 1994; Goodman, Devadas, & Hughson, 1988;Wageman, 2001).


serves as a recent exemplar in the area of teameffectiveness. The researchers asked whetherthe relationship between team structure andteam performance changes as a function of tasktype and whether intrateam processes mediatethe structure-performance relationship. The firstquestion gave rise to hypotheses about moder-ators of the relationship between structural in-puts and performance outcomes (notably, whenteam task is conceptual, the relationship be-tween team interdependence and performancewill be stronger than when team task is behav-ioral). These hypotheses were inspired by incon-

sistent findings within a large body of previouswork that had identified relationships betweenfacets of team structure (such as interdepen-dence) and team effectiveness. Because theseinconsistencies suggested the presence of amoderator, the researchers investigatedwhether differences in task type might ac-count for differences in the relationship be-tween team structure and effectiveness. Thesecond question addressed an untested as-sumption in the literature—that inputs such asteam structure affect team processes, which,in turn, explain team effectiveness (McGrath’s

TABLE 2Three Archetypes of Methodological Fit in Field Research

State of Prior Theoryand Research Nascent Intermediate Mature

Research questions Open-ended inquiryabout a phenomenonof interest

Proposed relationshipsbetween new andestablishedconstructs

Focused questionsand/or hypothesesrelating existingconstructs

Type of data collected Qualitative, initiallyopen-ended data thatneed to beinterpreted formeaning

Hybrid (bothqualitative andquantitative)

Quantitative data;focused measureswhere extent oramount is meaningful

Illustrative methods forcollecting data

Interviews;observations;obtaining documentsor other materialfrom field sitesrelevant to thephenomena ofinterest

Interviews;observations;surveys; obtainingmaterial from fieldsites relevant to thephenomena ofinterest

Surveys; interviews orobservations designedto be systematicallycoded and quantified;obtaining data fromfield sites thatmeasure the extent oramount of salientconstructs

Constructs andmeasures

Typically newconstructs, fewformal measures

Typically one or morenew constructsand/or newmeasures

Typically relyingheavily on existingconstructs andmeasures

Goal of data analyses Pattern identification Preliminary orexploratory testingof new propositionsand/or newconstructs

Formal hypothesistesting

Data analysis methods Thematic contentanalysis coding forevidence ofconstructs

Content analysis,exploratorystatistics, andpreliminary tests

Statistical inference,standard statisticalanalyses

Theoreticalcontribution

A suggestive theory,often an invitation forfurther work on theissue or set of issuesopened up by thestudy

A provisional theory,often one thatintegratespreviously separatebodies of work

A supported theory thatmay add specificity,new mechanisms, ornew boundaries toexisting theories


[1984] input-process-output model)—to test a pre-cise specification of team process as a mediatorbetween team interdependence and performance.

Nine hypotheses were developed from thesetwo questions, using constructs specified byprior work. For instance, Stewart and Barrick didnot need to observe teams to determine whattype of tasks teams performed; instead, they re-viewed the literature and identified a distinc-tion between conceptual and behavioral teamtasks. Similarly, prior work had identified teamstructural elements and clarified structures thatappeared most related to team effectiveness (in-terdependence and team self-leadership). Stew-art and Barrick could then draw on this work tofurther specify the conditions under which thoserelationships were present.

The researchers used a cross-sectional de-sign, collecting quantitative survey data fromforty-five manufacturing teams in three plants.This methodology was appropriate because theconstructs themselves were well understood.Reliable, valid measures of them existed in theliterature, and quantitative data were needed totest the hypotheses. Data analyses began withstatistical tests to ascertain whether data aggre-gation from the individual to the team level ofanalysis was justified,5 and standard reliabilityanalyses were conducted to ensure convergentand discriminant validity of the measures. Hy-potheses were then tested with regression anal-yses, using a quadratic term to test the hypoth-esized curvilinear relationship—that high levelsof team performance would be observed at highand low levels of team interdependence,whereas low levels of team performance wouldbe observed at moderate levels of team interde-pendence.6 Data analyses also tested for mod-eration and mediation effects.7

The contributions to the literature were a morerefined specification of factors that enhanceteam effectiveness, a clarification of task typeas a moderator, and tests of process mediators.The authors suggested that the input-process-output model of teams is more useful in explain-ing the relationship between interdependenceand performance when team tasks are concep-tual and less useful when the tasks are behav-ioral. Including task type as a boundary condi-tion helped refine team effectiveness theory.Finally, the study showed that team process me-diates the relationship between team structureand team effectiveness, providing empirical ev-idence for assumptions made in prior theoreti-cal work.

As this example demonstrates, precise mod-els, supported by quantitative data, are charac-teristic of effective field research in areas ofmature theory. Other examples in team researchinclude work by Wageman (2001) and by Chenand Klimoski (2003). Table 3 compares thesethree studies to highlight commonalities,thereby summarizing basic attributes of fieldresearch that achieves methodological fit withinmature theory related to work teams.

The examples in Table 3 are not intended tosuggest that there is never a benefit in revisitingwell-trodden theoretical territory with a com-pletely open mind. In the sections that follow,we show how researchers can—given certainconditions—develop greater understanding ofexisting relationships or mechanisms by em-bracing a qualitative or hybrid approach. Next,we turn to exploratory methods appropriate forunderstanding phenomena still in early stagesof theory development.

Nascent Theory Research

On the other end of the continuum is nascenttheory—topics for which little or no previoustheory exists. These topics have attracted littleresearch or formal theorizing to date, or elsethey represent new phenomena in the world(e.g., “virtual” or geographically dispersed workteams). The types of research questions condu-cive to inductive theory development includeunderstanding how a process unfolds, develop-

5 Two commonly used statistics for determining whether itis appropriate to aggregate individual responses to team-leveldata are the intraclass correlation coefficient, or ICC (James,1982), and Rwg (James, Demaree, & Wolf, 1984). Others have alsoused the eta-squared statistic (Georgopolous, 1986).

6 Hierarchical regression analyses testing for a curvilin-ear relationship proceed by regressing the dependent vari-able in step one and then adding the independent-variable-squared term in step two. A significant increase in theamount of variance accounted for by the second equation(i.e., a significant increase in R-squared) supports the exis-tence of a curvilinear relationship.

7 See Baron and Kenny (1986) for details on conductingmoderator and mediator analyses. Other useful sources for

deciding on appropriate statistical tests include Cohen andCohen (1983), Keppel (1991), Klein and Kozlowski (2000), Ped-hazur (1982), and Tabachnick and Fidell (1989).


ing insight about a novel or unusual phenome-non, digging into a paradox, and explaining theoccurrence of a surprising event. Interest inthese problems can arise from unexpected find-ings in the field, from questioning assumptionsor accepted wisdom promulgated in the extantliterature, and from identifying and addressinggaps in existing theory. The research questionsare more open-ended than those used to furtherknowledge in mature areas of the literature. Instudies where theory is nascent or immature,researchers do not know what issues mayemerge from the data and so avoid hypothesiz-ing specific relationships between variables.

Because little is known, rich, detailed, andevocative data are needed to shed light on the

phenomenon. Interviews, observations, open-ended questions, and longitudinal investiga-tions are methods for learning with an openmind. Openness to input from the field helpsensure that researchers identify and investigatekey variables over the course of the study. Datacollection may involve the full immersion of eth-nography8 or, more simply, exploratory inter-views with organizational informants.

8 Ethnography is the “written representation of culture (orselected aspects of a culture)” (Van Maanen, 1988: 1) thatoften reveals not only the inner workings of the culture butalso the context in which the culture exists, and how theculture both affects and is affected by the context (see alsoDenzin & Lincoln, 2000). Organizational ethnographies study

TABLE 3Similarities Among Mature Theory Studies

Element Stewart and Barrick (2000) Wageman (2001) Chen and Klimoski (2003)

Nature of the researchquestion

Testing theory-drivenhypotheses that therelationship between teamstructure and teamperformance changes as afunction of task type andthat intrateam processesmediate the structure-performance relationship

Testing theory-drivenhypotheses about thecontributions of teamleader coaching and teamdesign to the effectivenessof self-managed teams

Testing theory-drivenhypotheses that individualdifferences along withmotivational andinterpersonal processespredict role performance inindividuals who are new toproject teams engaged inknowledge work

Primary method of datacollection

A survey instrument thatyields quantitativemeasures of team process,task type, and otherestablished constructs inthe team effectivenessliterature

An interview protocol forteam members, withresultant qualitative datalater systematically codedto produce quantitativemeasures of leadercoaching, team design, andother established constructsin the team effectivenessliterature

A survey instrument thatyields quantitativemeasures of empowerment,role performance, and otherestablished constructs inthe team effectivenessliterature; created twomeasures of establishedconstructs in order toassess them appropriatelyfor the given sample

Data analysis Statistical tests: teamagreement tests (ICCs),followed by correlation andregression

Statistical tests: correlationand regression

Statistical tests: teamagreement tests (ICCs),followed by hierarchicalregression and structuralequations modeling

Contribution A precise model: teamprocess mediates theeffects of team structure onteam effectiveness; tasktype alters relationshipbetween team structure andteam effectiveness

A precise model: team designaffects team effectivenessmore than team leadercoaching: design andcoaching interact topositively impact teameffectiveness

A precise model: self-efficacyand self-expectations affectteam newcomers’ roleperformance throughmotivational processes,while prior experience andothers’ expectations affectteam newcomers’ roleperformance throughinterpersonal processes


Researchers frequently use a grounded theoryapproach to connect these data to existing andsuggestive new theory (Glaser & Strauss, 1967).Instead of a sequential process in which hypoth-eses are formed and data are collected and thenanalyzed, data analyses often alternate and it-erate with the data collection process. Contentanalyses help reveal themes and issues thatrecur and need further exploration. Through thisiterative process, theoretical categories emergefrom evidence and shape further data collection(Eisenhardt, 1989a; Glaser & Strauss, 1967). Inthis analytic journey, both the organization ofqualitative data into coherent stories of experi-ence and sensemaking processes are essentialanalytic activities.

Working within the nascent theory arena re-quires an intense learning orientation andadaptability to follow the data in inductivelyfiguring out what is important. Effective paperspresent a strong, well-written story to makesense of compelling field data. The essentialnature of the contribution of this type of work isproviding a suggestive theory of the phenome-non that forms a basis for further inquiry.

To continue our focus on work teams, wechose Barker’s (1993) paper as an exemplar of fitin nascent theory. Barker investigated how indi-viduals handled the transition from working in abureaucratic organization to working togetherin self-managed teams in a production environ-ment without formal control systems. Prior liter-ature maintained that organizations had movedover time from simple control (e.g., direct, au-thoritarian) to technological control (e.g., assem-bly lines) to bureaucratic control (e.g., hierar-chies, rules), with each new form of controlcreated to overcome the problems of the earlierform.

Many drawbacks had been identified with bu-reaucratic control systems, such as endless “redtape” that made it difficult to accomplish simpletasks. Weber (1958) had called the resulting sys-tem of rules, regulations, and rigid structure an“iron cage” that trapped employees in its imper-sonal grasp. To overcome the stifling nature ofhighly developed bureaucracies, firms beganimplementing self-managed teams to allow em-ployees greater discretion, empowerment, lati-

tude, and personal control over their work lives.The new system relied on “concertive control,”where team members worked together to nego-tiate behavioral norms. Yet little formal theoryand research existed to understand how thisprocess worked. Barker’s question thus focusedon the nature of concertive control, its develop-ment over time in a single organization, andwhether such a system truly represented a steptoward greater personal freedom compared to abureaucratic control system.

This research question was well matched toan in-depth qualitative study of newly formedself-managed work teams in a small manufac-turing firm. Barker’s immersion in the settingallowed him to gain detailed data on people’sexperiences over time, and thus to develop anunderstanding of how teams cope with the in-terpersonal challenges of self-management.Data collection spanned two years, includingsix months of weekly half-day plant visits; infor-mal conversations and interviews with manu-facturing workers and other employees; andconsiderable observation of teams working,meeting, and interacting informally. These first-hand data were supplemented by company doc-uments and surveys. Finally, Barker followedone team closely for four months.

Throughout the fieldwork, Barker engaged inan iterative process of analyzing data, writingup his understanding of the situations andevents, and developing new questions to shapesubsequent data collection. Because his re-search question focused on understanding dif-ferences between control practices in the newself-managed teams and those in the old bu-reaucratic system, Barker elicited control-related themes that he refined as he collectednew data. He used “sensitizing concepts” (Jor-gensen, 1989) drawn from previous research onvalue-based control (Giddens, 1984; Tompkins &Cheney, 1985) to guide his work. His work illus-trates how sensitizing concepts can be a valu-able tool in nascent theory research to guidequestions and help identify key themes. More-over, as Barker reports, reviewing emergentideas with colleagues who lack prior knowledgeof the firm or its teams is also an importantsource of feedback.9

the cultures of firms or groups within firms using in-depthqualitative field research.

9 See Adler and Adler (1987) for a discussion of the valueof feedback from research project outsiders.


As this study illustrates, when researchers donot know in advance what the key processesand constructs are, as they could if mature the-ory on their topic were available, they must beguided by and open to emergent themes andissues in their data. Iterating between data col-lection and analysis provides the flexibilityneeded to follow up on promising leads and toabandon lines of inquiry that prove fruitless.The results of Barker’s application of this in-vestigative process showed how what beganas a challenging and engaging process foremployees shifting to self-managed workteams degenerated into a stressful, fear-inducing, ever-tighter iron cage. He explainedhow concertive control in teams could be as ormore restrictive than a hierarchical bureau-cracy. Barker’s process description, told in acompelling narrative form, shed new theoreti-cal light on a previously obscure construct.

In addition to Barker’s work, Table 4 summa-rizes attributes of two more research projectsthat demonstrate methodological fit for nascenttheory. The research questions guiding thesefield studies were exploratory, designed to gen-erate new theory or propositions. Gersick (1988)explored how temporary project groups developover time, and Maznevski and Chudoba (2000)explored processes that allow geographicallydispersed industrial technology teams to effec-tively interact and produce successful results. Ineach case, theory relevant to the topic existedbut either failed to fit with observed processes(Gersick, 1988) or was not well enough devel-oped to motivate testable hypotheses related tothe particular question (Maznevski & Chudoba,2000). Including Barker (1993), each investigatorchose to collect qualitative data, through open-ended interview questions, observations ofmeetings, and review of archival qualitative

TABLE 4Similarities Among Nascent Theory Studies

Element Gersick (1988) Barker (1993)Maznevski and Chudoba(2000)


Exploring how short-termproject groups develop overtime and howdevelopmental shifts ingroups are triggered

Exploring how the controlsystems of self-managedteams emerge, areexperienced by teammembers, and differ frombureaucratic controlsystems

Exploring factors andprocesses that allow globalvirtual teams to operateeffectively

Primary method ofdata collection

Observation of all groupmeetings, supplemented byinterviews of half the studysample, that yieldedqualitative data about grouptask strategies, actions,changes, and taskcompletion; longitudinaldata collection (ranging fromseven days to six months)

Observation, conversations,and in-depth interviewsthat yielded qualitativedata about teaminteractions and controlsystem development;longitudinal data collection(over two years)

Observation of meetings,semi-structured andunstructured interviews,communication logs,questionnaires, and accessto company documentationthat yielded qualitativedata about team interactionmethods, timing, andcommunication content;longitudinal data collection(twenty-one months)

Data analysis Iterative, exploratory contentanalysis

Iterative, exploratory contentanalysis

Iterative, exploratory contentanalysis

Contribution An importation of a newconstruct—punctuatedequilibrium—and asuggestive model of thetemporary project group lifecycle

A new construct—concertivecontrol—and a suggestivemodel of how teams movefrom values to norms torules that become binding,limiting, and invisible

A suggestive model of howvirtual teams managesocial interactions and anew emphasis on therhythmic pacing of teammember encounters overtime to create effectiveoutcomes


data. Almost all data were collected longitudi-nally, with researchers spending anywhere fromseven days with a single group (Gersick, 1988) toover two years in the field (Barker, 1993).

These papers introduced or elaborated con-structs—punctuated equilibrium, concertivecontrol, and temporal rhythms—that could befurther developed in subsequent studies. Allpresented process models supported by repeat-ing patterns across data sources (or cases), high-lighting similarities of procedural stages orphases across units. Instead of reasonably con-clusive results, each study provided suggestivetheoretical insights to inform and inspire futureresearch on an interesting phenomenon.

Intermediate Theory Research

Intermediate theory research draws from priorwork— often from separate bodies of litera-ture—to propose new constructs and/or provi-sional theoretical relationships. The resultingpapers may present promising new measures,along with data consistent with the provisionaltheory presented. Such studies frequently inte-grate qualitative and quantitative data to helpestablish the external and construct validity ofnew measures through triangulation (Jick, 1979).Careful analysis of both qualitative and quan-titative data increases confidence that the re-searchers’ explanations of the phenomena aremore plausible than alternative interpretations.

One trigger for developing intermediate the-ory is the desire to reinvestigate a theory orconstruct that sits within a mature stream ofresearch in order to challenge or modify priorwork. For example, Edmondson (1999) marriedinsights from organizational learning research(tacit beliefs impede learning) with theory onteam effectiveness (structural differences acrossteams explain performance) to propose a provi-sional explanatory model of team learning thatfocused on how differences in interpersonal cli-mate across teams affected both team learningand performance.

Research questions conducive to developingintermediate theory include initial tests of hy-potheses enabled by prior theory (e.g., Edmond-son, 1999) and focused exploration that gener-ates theoretical propositions as output (e.g.,Eisenhardt, 1989b). The latter may include verypreliminary quantitative analysis to reinforcethe logic underlying the qualitatively induced

propositions. A single study may describe pat-terns that suggest both variance theories (anincrease in X leads to an increase in Y) andprocess theories (how a phenomenon works, howa process unfolds), although effective paperstend to emphasize one over the other (Mohr,1982). Just as quantitative methods are appropri-ate for mature theory and qualitative methodsfor nascent theory, intermediate theory is wellserved by a blend of both. This blend works tosupport provisional theoretical models. Thecombination of qualitative data to help elabo-rate a phenomenon and quantitative data toprovide preliminary tests of relationships canpromote both insight and rigor—when appropri-ately applied (e.g., Jick, 1979; Yauch & Steudel,2003). At the same time, integrating qualitativeand quantitative data effectively can be difficult(e.g., Greene et al., 1989), and there is a risk oflosing the strengths of either approach on itsown.

Examples of research achieving methodolog-ical fit within intermediate theory are growingin number, although there are fewer than thetwo more familiar categories. To continue ourfocus on teams, we use Edmondson (1999) toillustrate this category. This field study intro-duced a new construct, team psychologicalsafety, and investigated its effect on team learn-ing and performance. The ideas were groundedin two reasonably mature but separate theoret-ical perspectives—team effectiveness and orga-nizational learning—and included eight hypoth-eses about factors that enhance or inhibit teamlearning and performance. The design inte-grated qualitative and quantitative data, pro-viding an explicit rationale for doing so:

Most organizational learning research has reliedon qualitative studies that provide rich detailabout cognitive and interpersonal processes butdo not allow explicit hypothesis testing. . . . Manyteam studies, on the other hand, utilize largesamples and quantitative data but have not ex-amined antecedents and consequences of learn-ing behavior. . . . I propose that, to understandlearning behavior in teams, team structures andshared beliefs must be investigated jointly, usingboth quantitative and qualitative methods (Ed-mondson, 1999: 351).

Data were collected in a company whereteamwork and collective learning were salientand varied across teams. Variance was essen-tial for addressing the research question ofwhether psychological safety predicted team


performance; the salience of teams and learningfor the company was helpful in ensuring thatinformants took the topic seriously and providedcareful, informed reports of their experiences.The study involved three stages, starting withobservations and interviews with eight teams todevelop new survey measures to supplementexisting team measures and to further the re-searcher’s understanding of both psychologicalsafety and team learning processes in a busi-ness setting.

In the second phase a team survey10 was dis-tributed to 496 members of 53 teams in the firm.An additional survey was given to two or threeinternal customers of each team’s work to pro-vide data on the team’s learning behavior andperformance. To promote confidence in thequantitative measures, additional data fromother sources were collected. For example, aresearch assistant, blind to the study’s hypoth-eses, collected additional structured interviewdata from managers familiar with one or more ofthe teams to generate independent quantitativemeasures of four team design variables11 so asto mitigate common method bias. Last, the studyused an extreme-case-comparison technique,contrasting high- and low-learning teams, to un-derstand how they differed and how these dif-ferences were related to team performance.

Standard statistical analyses were used to an-alyze the quantitative data,12 and the resultsgenerally supported the hypotheses. The find-ings were enriched by supplemental qualitativedata to help explain the quantitative findings—shedding light on how those teams worked to-gether. These comparisons allowed a more fine-grained analysis of what was occurring “behindthe numbers” within the teams. The results of

the study broaden our understanding of teameffectiveness from a structural emphasis to in-clude interpersonal factors such as team psy-chological safety.

Other studies illustrating fit in intermediatetheory include Eisenhardt (1989b) and Allmen-dinger and Hackman (1996), as shown in Table 5to summarize basic attributes of methodologicalfit in this category. Blending qualitative andquantitative methods occurs in two basic waysin these studies. One approach supplementsqualitative work with quantitative data, allow-ing researchers to discern unexpected relation-ships, to check their interpretation of qualitativedata, and to strengthen their confidence in qual-itatively based conclusions when the two typesof data converge (e.g., Eisenhardt, 1989b). Theother approach supplements quantitative testswith qualitative data that enable a fuller expla-nation of statistical relationships between vari-ables, ensuring in particular that the proposedtheory constitutes a valid analysis of the phe-nomenon rather than artifacts of measurement.This approach also provides a deeper under-standing of and rationale for a proposed newconstruct (e.g., Edmondson, 1999). In summary,hybrid strategies allow researchers to test asso-ciations between variables with quantitativedata and to explain and illuminate novel con-structs and relationships with qualitative data(Yauch & Steudel, 2003).

Intermediate theory research sheds light onhow theory in management moves from the nas-cent stage toward maturity. Scholars have longadvocated cycling between inductive theory cre-ation processes and deductive theory-testingstrategies to produce and develop useful theory(e.g., Cialdini, 1980; Fine & Elsbach, 2000; Weick,1979). As our examples illustrate, theory in or-ganizational research rarely marches steadilyforward from nascent to mature, instead spawn-ing tangent studies that both build and diverge.Although some studies build on prior theory toelaborate and specify models more precisely(e.g., Stewart & Barrick, 2000), others use priorwork to inspire investigations in a brand newdirection (e.g., Barker, 1993). Intermediate theorydescribes a zone in which enough is known tosuggest formal hypotheses, but not enough isknown to do so with numbers alone or at a safedistance from the phenomenon (e.g., Edmond-son, 1999). In summary, intermediate theorystudies propose provisional models that ad-

10 Analyses of qualitative data in Phase I included exam-ining fieldnotes and interview transcripts to identify vari-ables of interest and to assess differences between teams onthose variables, as well as to shape the development of newsurvey measures through empathic design (Alderfer &Brown, 1972).

11 The four team design variables were the extent towhich a clear goal was present, the extent to which theteam’s task was interdependent, the extent to which theteam composition was appropriate, and the amount of con-text support each team received.

12 These analyses included tests of internal consistencyreliability, discriminant validity (e.g., Campbell & Fiske,1959), group-level variables (Kenny & LaVoie, 1985), regres-sion analyses, and GLM analyses.


dress both variance- and process-oriented re-search questions. Using both qualitative andquantitative data, these studies can identify keyprocess variables, introduce new constructs, re-conceptualize explanatory frameworks, andidentify new relationships among variables.

Mean Tendencies and Off-DiagonalOpportunities

Above we presented a pattern in which thematurity of theory and research in a given nar-row area strongly influences the design of fieldresearch conducted in that area. To show howmethods vary in form across a theoretical con-

tinuum, we drew from a range of articles pro-duced from original data collected in real orga-nizations. We chose studies that concentratedon teams to enable focused comparisons, with-out varying too many factors at once. In sum-mary, mature theory spawns precise, quantita-tive research designs, maturing or intermediatetheory benefits from a mix of quantitative andqualitative data to accomplish its dual aims,and nascent theory involves exploring phenom-ena through qualitative data. These archetypalcategories of organizational field research canbe positioned along the diagonal in Figure 1.

Congruence among the state of prior theory,the research question, and the research design

TABLE 5Similarities Among Intermediate Theory Studies

Element Eisenhardt (1989b)Allmendinger and Hackman(1996) Edmondson (1999)


Generating testable researchpropositions about howdifferent variables wererelated to strategic decisionspeed; exploring how firmsmake fast decisionseffectively

Assessing orchestracharacteristics undercontrasting conditions ofcontextual change;exploring factors thatdistinguished successfullyadapting groups from others

Preliminary tests of theory-driven hypotheses abouthow team structure andteam beliefs affect teamlearning and performance;exploring howpsychological safety andlearning behavior in teamsare related

Primary method ofdata collection

An interview protocol withdirect observations thatyielded qualitative dataabout decision making infast-paced environments;also archival data andindustry reports; quantitativedata about firm performance

Archival and interview datathat yielded qualitative dataabout orchestras’environmental context andhistories; group surveyswith established constructsyielding quantitative data,along with interviews andobservations, to assessrelationships amongvariables in this unusualgroup context

An interview protocol thatyielded qualitative data,followed by the creation ofan empathically developedquestionnaire used tocollect quantitative datafor main analyses,supplemented by newqualitative data to explainquantitative relationships

Data analysis Content analysis of qualitativedata; pair-wise comparisonsof stories between cases;quantitative analysesmentioned as supportive ofqualitative data

Content analysis ofqualitative data; statisticalanalyses to assessdifferences in theoreticallyrelevant variables acrossorchestra types and contexts

Content analysis ofqualitative data for inputto questionnairedevelopment; statisticalanalyses as initial tests;qualitative analysis fordeeper understanding

Contribution Iteratively developed researchpropositions and aprovisional model of howfirms make fast decisions

Provisional contingencyframework of when and whypreviously establishedtheoretical models areuseful for understandinghow groups respond toenvironmental change

New construct incorporatedinto a provisional modelwith roots in a maturetheoretical model, and newintegration of theoreticalperspectives


help a new field study make a compelling newcontribution to the literature. As illustrated inthe preceding pages and tables, the nature ofthis contribution varies as research travelsalong the diagonal, from a suggestive new the-ory that invites further research to a provisional,partially supported theory that may introducenew constructs or integrate previously disparatebodies of literature to a precise theory that addsnew specificity to the existing theoretical mod-els in a given body of literature.

This pattern of archetypes cleanly situatedalong the diagonal represents a mean tendencyin effective field research, but by no means doesit comprise a rigid rule. First, the oval shape ofthe diagonal line is intended to suggest leewayin research design. For instance, as notedabove, intermediate theory may draw primarilyfrom qualitative data, with minimal quantita-tive data in the background, or it may relyextensively on quantitative data, with supple-mentary qualitative data to shed light on mech-anisms. Second, off-diagonal opportunities existwhen—with awareness of the literature on aparticular topic—a study’s focus is reframedfrom the broad to the narrow. In his study ofself-managed teams, for example, Barker (1993)did not ask what makes self-managed teamseffective but, rather, how team members createand cope with the social pressures of self-management. Thus, despite the maturity of re-search on self-managed work teams, Barkerused qualitative data to suggest compellingnew theory with evocative case descriptions ofreal work teams. Methodological fit in this ex-ample was created in an initially off-diagonallocation by framing the study’s focus narrowly

and examining an area where theory no longercould be categorized as mature.

Perlow’s (1999) ethnographic investigation ofhow people use their time at work provides an-other illustration of this approach. Contemplat-ing a relatively mature body of research onwork/life balance and time management, Per-low saw unanswered questions about people’sday-to-day experience of time constraints. Sheset out to understand how—and why—peoplereally used their time at work, as well aswhether their time usage patterns were effectivefor both themselves and their workgroups. Herqualitative study of seventeen engineers in asoftware development group in a Fortune 500company revealed patterns of work interruptionthat greatly limited individual and group pro-ductivity, increasing the engineers’ work hours.The second phase of the study included a smallexperiment imposing “quiet time” to amelioratethe counterproductive pattern, improving pro-ductivity briefly until old habits prevailed afterthe researcher’s departure. From these findings,Perlow (1999) suggested a need for a “sociologyof time” to recognize the interdependence of so-cial and temporal contexts at work. In sum, shestarted with a more mature area of research butdiverged from there to explore a key phenome-non—interactions among individuals’ timemanagement—to suggest new theory to inspireand inform future discussions in this area.

These two examples can be located conceptu-ally at the intersection of initially mature theoryand qualitative data marked by B in Figure 1. Incontrast, we consider the intersection of nascenttheory and quantitative data, marked by A inFigure 1, an approach that is more difficult tojustify. For instance, a strategy of collecting ex-tensive quantitative data to explore for statisti-cal associations runs the risk of finding signifi-cance by chance, merely because of the largenumber of potential relationships (Rosenthal &Rosnow, 1975). Moreover, because data collec-tion in organizational field research is expen-sive and often moderately intrusive, it should becollected with care for a deliberate purpose. Thespace below the diagonal in Figure 1, therefore,may present creative opportunities for theoreti-cal contributions, whereas work in the spaceabove is not likely to produce compelling fieldresearch.

Finally, sometimes an initial diagnosis ofstudy type must be revised because of unex-

FIGURE 1Methodological Fit As a Mean Tendency


pected findings. For example, a research projectmight start in the upper right-hand corner andmigrate down to intermediate status after a sur-prising quantitative finding seems worth inves-tigating further. Such a journey was describedin an initially mature theory study of nursingteam effectiveness using medical error rates asa dependent variable (Edmondson, 1996). Star-tled to discover that team effectiveness andteam leader coaching were correlated withhigher, not lower, detected error rates, the re-searcher suspected that differential reportingclimates accounted for the unexpected result. Toexplore this possibility, a research assistant,blind to the quantitative data and to the newhypothesis, explored how each team worked asa social system. This additional qualitative dataprovided tentative support for interpersonal cli-mate as a hidden variable accounting for theunexpected result, reclassifying the study as ahybrid design working within intermediate the-ory.

DISCUSSION

We argue that methodological fit promotesthe development of rigorous and compellingfield research. We delineate archetypes of meth-odological fit in field research, in which threelevels of prior work (nascent, mature, and inter-mediate) correspond to three methodologicalapproaches (qualitative, quantitative, and hy-brid). Our framework is not intended as an in-flexible set of rules but, rather, as a clarifyingheuristic that articulates tacit principles embed-ded in effective field research and that builds onmethodological rules and guidelines coveredelsewhere (e.g., Bouchard, 1976; Lee et al, 1999;McGrath, 1964).

Some problems of poor fit can be solved byreframing a paper or reanalyzing qualitativedata; others may require new data or a freshstart. Our framework—with both on- and off-diagonal opportunities—is intended to help re-searchers (and reviewers) ascertain which casesfall into the former category, as well as whenand how to shape and reshape a researchproject and its outputs in such a way that theconclusions are compelling. Off-diagonal oppor-tunities exist when a researcher intentionallyopens a new area of focused inquiry within abroadly familiar topic, thereby pursuing a newtopic related to an old phenomenon (e.g., concer-

tive control in self-managed teams). More com-monly, however, researchers who stray from thediagonal are unaware of doing so—in part ow-ing to the complexity of field data—and may runinto a small number of predictable problems.

Problems Created by Poor Fit

For each of the three levels of prior work ar-ticulated above, we suggest that using either ofthe alternative (off-diagonal) methodologies cre-ates problems that diminish the effectiveness ofthe research products. Table 6 summarizes thesix problems and their three essential outcomes.

Two types of poor fit in areas of mature theory.When prior work related to a research question(e.g., what explains team effectiveness?) hasproduced some reasonably robust findings, astudy that relies on purely qualitative data risksrediscovering known factors in its “new” theo-ry—the problem of reinventing the wheel. Pour-ing through qualitative data to find out whatdistinguishes, for example, two high-performingfrom two low-performing teams, a researcher islikely to identify such factors as goal clarity,group process, or the adequacy of team compo-sition or resources (e.g., Hackman, 1987). In short,systematically analyzed qualitative data willtend to uncover similar factors in response tosimilar questions. Given the time invested in theanalytic process and the other sunk costs in thestudy, a researcher may then feel pressure tooverstate the novelty or implications of his orher findings. An alternative in this situationwould be to analyze the data to investigate adifferent question (e.g., how team members copewith the pressures of self-management and peercontrol; Barker, 1993).

In a second type of poor fit, a study informedby a mature body of literature that integratesqualitative and quantitative research in a sin-gle paper faces the problem of the uneven statusof evidence. First, juxtaposing the interpretivenature of qualitative analysis with statisticaltests highlights the different functions of the twodata types. Specifically, qualitative data illus-trate and may reveal processes, but they do nottest or prove as well as quantitative data. Sec-ond, the combination will lengthen a researchpaper without increasing the strength of its con-clusions. If hypotheses are anchored in the lit-erature and are well argued, quantitative mea-sures and tests should provide powerful and


sufficient support for the ideas. In some cases,incorporating one or more stories may be usefulto familiarize readers with an unusual contextor to illustrate a finding, but when presented asformal evidence, they usually fall short.13 In

sum, long qualitative reports from the field areunlikely to strengthen research projects thatpresent and test hypotheses relating known con-structs.

Fortunately, this problem has a simple solu-tion; the study should rely on the quantitativedata as evidence and should use only as muchqualitative data as necessary to introduce or

13 To better understand why this is usually the case, recallthe three basic designs noted above for combining qualita-tive and quantitative data to develop and support a newtheory: (1) explore first, through interviews and observationsthat guide the development of subsequent quantitative sam-ples and measures; (2) collect follow-up qualitative data tobetter understand—usually surprising—quantitative find-ings; or (3) collect both types of data at the same time, totriangulate. When researchers can articulate good hypothe-ses from prior research and new logic, and can support thesewith quantitative analyses, all three hybrid approachespresent risks. In the first case, preliminary field interviewsor observations may help in the wording of survey items butgenerally would not be needed to discern or develop newconstructs and, thus, would not play a key role in suggesting

or supporting the theory. In the second case (follow-up qual-itative data), stories may illustrate how a theory works, butthey cannot provide evidence of a relationship between con-structs because the qualitative data are a biased sample,collected by a biased observer. In the third case (simultane-ity), the mix works well to triangulate across sources for newmeasures, but for known measures, triangulation is unnec-essary. All three cases thus share the problem that the qual-itative data are redundant and may undermine the clarity ofthe quantitative analyses if presented as results rather thanas background or illustrative material.

TABLE 6Problems Encountered When Methodological Fit Is Low

Prior Work on ResearchQuestion

Data Collectionand Analysis Problems Encountered Outcome

Mature: Extensive literature,complete with constructs andpreviously tested measures

Qualitative only Reinventing the wheel: Studyfindings risk being obvious orwell-known

Research fails to buildeffectively on prior work toadvance knowledge aboutthe topicHybrid Uneven status of evidence:

Paper is lengthened but notstrengthened by usingqualitative data as evidence

Intermediate: One or morestreams of relevant research,offering some but not allconstructs and measuresneeded

Quantitativeonly

Uneven status of empiricalmeasures: New constructs andmeasures lack reliability andexternal validity and suffer incomparison to existingmeasures

Results are less convincing,reducing potentialcontribution to the literatureand influence on others’understanding of the topic

Qualitative only Lost opportunity: Insufficientprovisional support for a newtheory lessens paper’scontribution

Nascent: Little or no prior workon the constructs andprocesses underinvestigation

Qualitative only Fishing expeditions: Resultsvulnerable to findingsignificant associationsamong novel constructs andmeasures by chance

Research falls too far outsideguidelines for statisticalinference to convince othersof its merits

Hybrid Quantitative measures withuncertain relationship tophenomena: Emergentconstructs may suggest newmeasures for subsequentresearch, but statistical testsusing same data thatsuggested the constructs areproblematic


discuss the research context. To illustrate such ajourney, the authors of a mature theory paperarguing that tacit knowledge increases the het-erogeneity of learning curves across teams (Ed-mondson, Winslow, Bohmer, & Pisano, 2003)originally presented the paper at the Academyof Management annual meeting and then sub-mitted it to Decision Sciences using a blend ofquantitative and qualitative analyses as evi-dence. The paper tested theory-driven hypothe-ses—that knowledge type moderates the rela-tionship between experience and rate oflearning in surgical teams and that team stabil-ity promotes team efficiency. To supplementquantitative measures of established constructsin the team and knowledge literature, the au-thors also presented qualitative interview data.Reviewers, confused by the inclusion of thequalitative data, pushed back. Initially attribut-ing this response to closed-mindedness, the au-thors reluctantly removed the qualitative datafrom the findings, leaving only an anecdote ortwo in the discussion to convey the nature of theteams’ learning challenge. Despite the authors’reluctance, the clarity of the paper was greatlyimproved by the reviewers’ feedback, becauseprior work on tacit knowledge and learningcurves was sufficiently mature that stories toelucidate mechanisms were unnecessary. Theauthors’ initial unflattering attributions aboutthe reviewers’ judgment might have beenavoided had the reviewers used the language ofmethodological fit to convey why the qualitativedata did not strengthen support for the paper’sconclusions.

The challenge lies in realizing when qualita-tive data—while complex, interesting, and sub-tle— does not serve an evidentiary function,given the state of knowledge at the time. Inareas of mature theory, scholars thus encounterproblems when qualitative data are presentedas evidence, and these problems lessen the po-tential contribution of their work, as noted inTable 6. In short, mature theory is advancedwith compelling quantitative studies.

Two types of poor fit in areas of nascent the-ory. When little or no prior work related to aresearch question exists, researchers face prob-lems when they seek to collect purely quantita-tive data. First, it almost certainly will be thecase that the quantitative measures will havean ambiguous relationship to the phenomenaunder study. The measures may capture prelim-

inary ideas about emergent constructs, and theanalyses, which pertain to the measures ratherthan to the phenomena themselves, do not aidthe researcher in truly learning from the fieldsetting. Others (e.g., Nunnally & Bernstein, 1994)have illuminated the issues of measurement va-lidity and reliability thoroughly; here we simplynote that it is difficult to create measures ofacceptable external validity or reliability whenphenomena are poorly understood.

Another problem with using quantitativemeasures with nascent theory is that investiga-tors, even with the best of intentions, aretempted to go on fishing expeditions. Any statis-tically significant relationships among vari-ables that emerge by chance are likely to beoverinterpreted as evidence to support an emer-gent theory. Further, given the measurement is-sue outlined above, it is difficult to interpret thetrue meaning of observed statistical relation-ships or counts. Researchers need to go throughthe process of building new ideas iteratively,with extensive exposure to the phenomenon andan open mind, before becoming captivated bypotentially chance associations.

Similarly, a hybrid approach also suffers fromthe uncertain status of (quantitative) measuresthat are employed before sufficient explorationof a new area has pinned down factors to mea-sure. Quantitative measures indicate a prioritheoretical commitments that partially closedown options, inhibiting the process of explor-ing a new territory (Van Maanen, 1988). Yet evenif the qualitative and quantitative data are stag-gered in phases, the initial exploratory qualita-tive phase in a nascent area of research is un-likely to yield more than one new variable readyfor formal tests—even preliminary ones—in thesame study. Statistical tests, thus, are unlikelyto be as informative as they may seem to theresearchers. Quantitative tests take on a certainillusion of accuracy that may mislead in thiscontext. For example, some years ago, the firstauthor submitted a paper attempting to inte-grate qualitative and quantitative data to ex-plain how teams navigate the challenge oflearning a new technology. The quantitativemeasures were not only technically deficient(new and unvalidated) but also at odds with theespoused goal of exploring team processes.Eliminating the quantitative analyses strength-ened the paper considerably, allowing an ap-propriate focus on understanding the team


learning process (Edmondson, Bohmer, &Pisano, 2001).

When addressing a novel question, research-ers collect—as they should—qualitative dataopportunistically such that they are free tochase new insights that emerge in an interviewor observation. The sample is, by design, pathdependent. For instance, subsequent interviewquestions (or interviewees) are determined iter-atively as interesting ideas emerge in the pro-cess. Data analysis and data collection overlap,as noted above. This approach allows new in-sight and theory to take shape, but it precludesthe systematic sampling and consistent use ofmeasures required for meaningful statistical in-ference—even with the most lenient standards.Thus, in nascent areas the inclusion of qualita-tive data in a hybrid design does not overcomethe problems associated with the use of quanti-tative data.

Both of these fit problems stem, in part, fromthe likely failure of quantitative measures andanalyses used in a nascent area to conform suf-ficiently to basic assumptions of statistical in-ference. Although organizational researcherstolerate deviations from ideal samples and ac-cept imperfect measures in their quantitativefield studies, designs that fall completely out-side the guidelines for normal science— be-cause they have sampled in a snowballing man-ner and/or deliberately used inconsistentquestions and techniques to collect data—distort the use of these powerful tools. As arguedabove, when little is known about a researchtopic or question, initial steps must be taken toexplore and uncover new possibilities beforeuseful quantitative measures can be informa-tive. Subsequent studies, building on an accu-mulation of early qualitative work, are betterable to conduct preliminary statistical tests ofemergent theoretical ideas.

Two types of poor fit in areas of intermediatetheory. Finally, when prior work related to aresearch question falls between nascent andmature, such as when a new construct appearslikely to explain an outcome of interest, researchdesigns that use either exclusively quantitativeor qualitative data both encounter problems. Inthe former case, new measures introduced tocapture new constructs lack credibility whenused without qualitative illustration and trian-gulation. For example, when the construct ofteam psychological safety was introduced (Ed-

mondson, 1999), qualitative evidence of differ-ences across teams in interpersonal climatewas important for establishing the external va-lidity of the construct, as well as for showing aclear relationship between the construct and thenew measure. Without such data, the new sur-vey measure would lack support for its implicitclaim that it captured a distinct new constructand would be more vulnerable to concernsabout common method bias. Therefore, researchin an area of intermediate theory that combinesnew and established measures without support-ing qualitative data is likely to suffer from theuneven status of empirical measures.

In contrast, a purely qualitative study in anintermediate theory area encounters the prob-lem of lost opportunity for preliminary statisticalsupport for its hypotheses. Although intermedi-ate theory hypotheses may suffer in comparisonto hypotheses relating well-established con-structs, when clearly argued, they merit initialtests. By not taking advantage of this opportu-nity, research products are likely to be less com-pelling than otherwise.

Complementing Prior Work on OrganizationalResearch Methods

For organizational researchers, a highly de-veloped body of work prescribes rules andguidelines for how to collect and analyze data(e.g., Cohen & Cohen, 1983; Miles & Huberman,1994; Pedhazur, 1982; Rosenthal & Rosnow,1975; Tabachnick & Fidell, 1989). In a growingbody of work, researchers expound the legiti-macy of qualitative research as a means ofexpanding organizational knowledge (e.g.,Eisenhardt, 1989b; Glaser & Strauss, 1967; Leeet al., 1999; Miles & Huberman, 1994), andmany advocates view qualitative methodol-ogy as particularly, if not exclusively, valu-able (Morgan & Smircich, 1980). Others debatethe appropriateness of combining qualitativeand quantitative methods in a single study(see both Sale et al., 2002, and Yauch & Steu-del, 2003, for reviews). While the importance ofmatching methods to questions has been rec-ognized (Bouchard, 1976; Campbell et al., 1982;Lee et al., 1999; McGrath, 1964), guidelineshave not been articulated to help researchersmake choices among the variety of potentialsources of data they face in the field— only


some of which provide a good fit with theirresearch questions.

To this prior work we add a framework forpromoting methodological fit in field research,with a particular emphasis on the conditionsunder which hybrid designs are most effective.Our goal is to help researchers think throughtheir options more systematically and explicitlyso as to produce high-quality field research thatadvances theory and practice. We also proposethat our framework may help reviewers and ed-itors assess manuscripts reporting on field re-search. We suspect that few are immune to theneed for this help.

For instance, while revising this manuscript,we encountered repeated real-life remindersof how challenging it can be to achieve meth-odological fit in field research. First, a gradu-ate student engaged in a qualitative disserta-tion in a nascent area gave a talk thatpresented a set of induced variables, com-pared via t-tests, to support new theory. Se-duced by the apparent certainty of quantita-tive data, the young researcher saw thestatistical tests as more powerful than thecareful thematic analyses that produced thenew variables. Perhaps this was inexperiencespeaking. Yet, shortly thereafter, a mid-careerresearcher, with quantitative expertise, re-quested feedback on a beautifully writtenqualitative paper—addressing a mature the-ory question. Enthusiastic about the richnessof verbatim data, the author failed to recog-nize that the gist of the paper’s findings repli-cated much that was already known in therelevant literature. Third, an even more ac-complished scholar reported plans to collectsurvey data—triggered by a field site’s desireto be surveyed—in a highly unusual contextabout which little was known.

The Fitting Process

As others have noted (e.g., Fine & Elsbach,2000), iterating between inductive theory devel-opment and deductive theory testing advancesour understanding of organizational phenom-ena. This advance is rarely a sanitized linearprogression that starts with a literature review,moves on to the research question, data collec-tion, and analysis, and ends seamlessly withpublication, as illustrated in Figure 2. We con-ceptualize the process of field research as ajourney that may involve almost as many stepsbackward as forward. More specifically, we ar-gue that methodological fit is achieved througha learning process.

Although our first-hand knowledge is limitedto just a few of the studies described in thisarticle, from these we conclude that creating fitis an iterative process that centrally involvesfeedback and modification at many stages. Wemodel the fitting process as a funnel, drawing(not incidentally) from the product developmentliterature (Wheelwright & Clark, 1992). The fun-nel symbolizes the relatively greater latitudeand choice early in a project, which progres-sively narrow as time goes on. Before a singlepiece of data is collected, the options are almostunlimited. At a certain point, feedback contrib-utes only to minor refinements in output—theslate is no longer blank. Figure 3 depicts thismodel.

The fitting process necessarily starts— or insome cases restarts—with some level ofawareness of the state of prior work in an areaof interest. Ideally, a researcher develops areasonably good understanding of majorstreams of work in one or more bodies of re-search literature and then begins to shape aresearch question. This question substantiallynarrows down the possibilities for the re-search design. In this way a study design (set-

FIGURE 2Traditional Implicit View of the Field Research Process


ting, type of data, sample, analyses) followslogically from and addresses issues of interestto the researcher as well as to others in thefield who care about the same topics. In short,a researcher is agreeing to engage in a dia-logue—albeit a slow and stilted one—withpeers who care about related issues and ques-tions. This dialogue transpires primarilythrough papers, the written products of ourresearch.

As Figure 3 conveys, options decrease asdecisions are made. Because data collectionnarrows the scope of subsequent decisions, itis important to spend sufficient time iteratingwithin the first three stages in the process, asindicated by the wider cyclical arrows in themodel. As a research question becomes morefocused, initial research design ideas emergeand are refined and elaborated. Designchoices broadly involve the type of data to becollected and the methods used to collect thedata (e.g., observation, interviews, surveys).As a researcher strives to resolve the tensionbetween the ideal version of his or her projectand one that is feasible and viable, the designevolves. Considering how to operationalize,explore, or test different research questionsoften leads to the realization that those ques-tions or hypotheses need to be sharpened, re-vised, or scrapped.

Just as consideration of design choices mayresult in reformulation of research questions,

experiences during data collection may suggestthat the research design be modified. For in-stance, work that focuses on validating a newconstruct and understanding how it functions (aprocess orientation) is likely to be conductedwith qualitative methods. During the course ofthe investigation (e.g., through interviews or ob-servations), information may arise that suggeststhat a new construct is related to other, moreestablished variables of interest (e.g., perfor-mance) in ways that appear predictable. Thismay lead a researcher to create a measure of thenew construct and to use established measuresof other relevant constructs to collect quantita-tive data in order to tentatively investigate vari-ance-based hypotheses.

In the messy reality of field research, datacollection opportunities may emerge before theresearcher has a clear idea about how the datawill be used. At other times original researchdesigns may be disrupted by layoffs or otherenvironmental changes beyond the investiga-tor’s control. In such situations the researchermust iterate back up the funnel in Figure 3,returning perhaps to the literature for direction,or deciding to collect new data of a differentnature to deepen understanding of a differentphenomenon (e.g., see Meyer, 1982, for a superbexample of researcher flexibility).

Once data are collected, an effective re-searcher employs analytic techniques that

FIGURE 3Field Research As an Iterative, Cyclic Learning Journey


match the nature and amount of data.14 Theprocess of writing up the results of the analy-ses may trigger additional questions for theresearcher, or suggest investigating alterna-tive explanations during data analysis. Fi-nally, as anyone who has submitted a manu-script for publication knows, the researchercan expect additional cycles of learning priorto publication. Even rejected manuscripts ar-rive with comments and suggestions from re-viewers that can inform revision of the manu-script in preparation for submission to anotherjournal. It is not uncommon for reviewers tosuggest that authors return to the literature tofurther develop their hypotheses or researchquestions from prior theory or to better informthe discussion of their results. By framingthese suggestions and recommendations asinputs in a learning process, rather than asdevastating criticism, researchers improve thequality of their research.

The journey varies in certain predictableways across the continuum. In more mature ar-eas, intensive conceptualization occurs early inthe process while the literature is being di-gested, and compelling hypotheses and modelsare developed. Before collecting extensivequantitative data, the researcher wants to beconfident that the key hypotheses are sensibleand likely to be supported. This requires exten-sive conceptual work to develop the ideas care-fully, obtaining considerable feedback from oth-ers, and refining the predictions before datacollection. Once hundreds of surveys are sentout, for instance, the stakes are quite high andthe data irreversible. In contrast, in the nascentstage, the intensive conceptualization work oc-curs later in a project, during and after datacollection, through an inductive process of seek-ing patterns to explain the data. More data canbe collected to dig into anomalies encountered.Thus, at both extremes, effective researchprojects require learning cycles, but the timingof intense theoretical development varies. Forall field research endeavors, however, a learn-ing-oriented mindset that values and welcomes

critical feedback is an essential asset of thefield researcher seeking methodological fit.

Educating the New Field Researcher

One implication of an emphasis on method-ological fit in field research is that researcherswho wish to explore different types of questionsin their careers must be methodologically ver-satile. New field researchers need exposure toboth quantitative and qualitative techniques,and they need to develop specific skills as wellas general awareness of when each is most ap-propriate. In this way the researcher will gain alarger toolbox with which to work, expandingthe types of research questions he or she cananswer effectively, and thereby also benefitingthe field. Although not every researcher will be-come a renaissance methodologist with deepexpertise and skill in all research techniques, arealistic goal is to provide students with enoughawareness of multiple methods to become effec-tive collaborators with others whose deep skillsin particular methodologies complement theirown skills and preferences.

A second implication of these ideas for meth-odological education is the need to explicitlyteach the notion of methodological fit. We offerseveral suggestions for how to do this. First, newfield researchers can be taught methodologicalfit by deconstructing exemplars, similar to ourapproach in this paper. A set of exemplars canbe put together based on topic (e.g., teams) ormethod (e.g., hybrid approach) to help studentsidentify a range of issues and trade-offs thatunfold in real research projects. Students canidentify strengths and weaknesses of decisionsor trade-offs they discern, suggest changes thatmight have improved the work, and appreciatethe ways that the researchers’ choices were ef-fective and mutually reinforcing.

Second, students can be invited to create re-search proposals that include preliminary ideasabout each of the elements of field researchshown in Table 1, to be evaluated by professorsand peers. In this way students can obtain feed-back on the degree of logical consistency amongthe proposed elements. Because it is easier todetect others’ fit gaps than one’s own, this feed-back is invaluable. Such research proposals in-volve students in their areas of interest, whileallowing them to view and shape the process ofdeveloping a research project. This learning

14 As previously mentioned, many sources for data anal-ysis techniques can be recommended; for example, we referthe interested reader to Miles and Huberman (1994) for infor-mation on the analysis of qualitative data and to Tabach-nick and Fidell (1989) for information on conducting multi-variate analyses of quantitative data.


process is likely to require a climate of psycho-logical safety to help students take the interper-sonal risks of sharing early efforts at designingresearch. Group dialogue is particularly usefulwhen it allows students to think through ideastogether, raise questions, help each other eval-uate their own decisions in terms of method-ological fit, and identify potential pitfalls. At-tention to methodological fit complementsstrategies others have suggested for devisingsignificant and satisfying research questions(Campbell et al., 1982).

Third, involving students in the planning andexecution phases of professors’ research is an-other way to enhance their access to the reason-ing, decisions, and trade-offs early on andthroughout the research process. This close in-volvement demands time and patience on thepart of professors, which pays off in effectiveexperiential learning through apprenticeship.

Fourth, explicit thought experiments can helpstudents think through issues of methodologicalfit in field research. For instance, a professormight begin by posing a research question andasking his or her students to determine how thequestion should be refined to become action-able. From there, students can be asked to iden-tify how the piece of research might be different,depending on where along the continuum fromnascent to mature the existing literature is po-sitioned.

In sum, implications of our framework for ed-ucating new field researchers include the needfor skill set versatility, the use of exemplars (ormodels of fit in published work) as case studiesfrom which to induce implicit principles, directexperience of the research process, and genera-tive conversation about possibilities to supple-ment training in specific methodological toolsand techniques. Thus, we advocate experientialeducation rather than lecture in communicatingthese ideas. To the extent possible, methodolog-ical fit should be “discovered” by studentsrather than merely described to them. The sat-isfaction of discovering them firsthand maymake the lessons more powerful and lasting.

Limitations and Boundaries

First and foremost, the ideas in this paper areintended for field research, thereby excludingmany important areas of management scholar-ship. It is also important to point out that our

data sources in developing the framework weredrawn from research in micro- and meso-orga-nizational behavior focused on teams, becausethis is the area we know best—the literature weread most often, the papers we review, and theresearch we conduct. Although this helped nar-row the scope of inquiry into a manageablebody of data, it is possible that the frameworkpresented here would need to be modified forother areas of management research.

Second, a key limitation on the production ofmethodological fit is the versatility of the re-searcher. The aim of this paper is to explicitlydiscuss methodological fit to help researchersmake more informed decisions as they workthrough the iterative, nonlinear process of con-ducting field research. Yet we recognize thatmany, if not most, management scholars havestrong preferences for methods they feel com-fortable with. As such, a contingency approachmay not always seem desirable or feasible.When research questions call for the flexibilityof a contingency approach, scholars may needto collaborate with those whose skills and pref-erences are different from and complementaryto their own.

Third, our framework does not address morenarrow and precise methodological fit choices,such as which type of interviewing style to usein a given research site or which statistical testsprovide the best fit for a given data set. We donot describe techniques for quantitative andqualitative data collection and analysis or forsampling, topics that have been well coveredelsewhere (e.g., Cohen & Cohen, 1983; Cook &Campbell, 1979; Eisenhardt, 1989b; Glaser &Strauss, 1967; Keppel, 1991; Miles & Huberman,1994; Nunnally & Bernstein, 1994; Tabachnick &Fidell, 1989).

CONCLUSION

This article pulls together key elements frommanagement field research into a single frame-work that provides language and advice for dis-cussing and promoting methodological fit. Wedrew on contemporary research on teams toshow a spectrum of theoretical developmentand its methodological implications. We do notadvocate one method over others but, rather,clarify how methodological choices can en-hance or diminish the ability to address partic-ular research questions. In developing a new


framework, we revisited old territory with amodern lens that acknowledges the growing im-portance of field research for developing theoryat all stages.

We showed that fit is achieved by logical pair-ings between methods and the state of theorydevelopment when a study is conducted. As anarea of theory becomes more mature withgreater consensus among researchers, most im-portant contributions take the form of carefullyspecified theoretical models and quantitativetests. Conversely, the less that is known about aphenomenon in the organizational literature,the more likely exploratory qualitative researchwill be a fruitful strategy. In the middle, a mix ofqualitative and quantitative data leveragesboth approaches to develop new constructs andpowerfully demonstrate the plausibility of newrelationships.

This article emphasizes fit as a critical, butnot exclusive, input to high-quality field re-search. Notably, the quality of the individualelements of research, including the review ofrelated literature and effective techniques fordata collection and analysis, matters greatly.Our argument is simply that fit is important andpotentially overlooked by busy or inexperiencedresearchers who may fail to see larger patternsthat give rise to inconsistencies between theiraims and their methods. The exemplars dis-cussed in this article illustrate how fit amongthe elements of field research can be achievedacross different field research contexts. In thisway we hope to encourage other researchers toconsider methodological fit in their efforts tocontribute to our collective understanding of or-ganizational phenomena and managementpractice.

REFERENCES

Adler, P. A., & Adler, P. 1987. Membership roles in fieldresearch. Beverly Hills, CA: Sage.

Alderfer, C. P., & Brown, D. L. 1972. Designing an “empathicquestionnaire” for organizational research. Journal ofApplied Psychology, 56: 456–468.

Allmendinger, J., & Hackman, J. R. 1996. Organizations inchanging environments: The case of East German sym-phony orchestras. Administrative Science Quarterly, 41:337–369.

Barker, J. R. 1993. Tightening the iron cage: Concertive con-trol in self-managing teams. Administrative ScienceQuarterly, 38: 408–437.

Barley, S. R. 1990. Images of imaging: Notes on doinglongitudinal field work. Organization Science, 1: 220 –247.

Baron, R. M., & Kenny, D. A. 1986. The moderator-mediatorvariable distinction in social psychological research:Conceptual, strategic and statistical considerations.Journal of Personality and Social Psychology, 51: 1173–1183.

Bouchard, T. J., Jr. 1976. Field research methods: Interview-ing, questionnaires, participant observation, systematicobservation, unobtrusive measures. In M. D. Dunnette(Ed.), Handbook of industrial and organizational psy-chology: 363–413. Chicago: Rand McNally.

Campbell, D. T., & Fiske, D. W. 1959. Convergent and dis-criminant validation by the multitrait-multimethod ma-trix. Psychological Bulletin, 56: 81–105.

Campbell, D. T., & Stanley, J. C. 1963. Experimental andquasi-experimental designs for research. Boston: Hough-ton Mifflin.

Campbell, J. P., Daft, R. L., & Hulin, C. L. 1982. What to study:Generating and developing research questions. BeverlyHills, CA: Sage.

Campion, M. A., Medsker, G. J., & Higgs, A. C. 1993. Relationsbetween work group characteristics and effectiveness:Implications for designing effective work groups. Per-sonnel Psychology, 46: 823–850.

Chen, G., & Klimoski, R. J. 2003. The impact of expectationson newcomer performance in teams as mediated bywork characteristics, social exchanges, and empower-ment. Academy of Management Journal, 46: 591–607.

Cialdini, R. B. 1980. Full-cycle social psychology. AppliedSocial Psychology Annual, 1: 21–47.

Cohen, J., & Cohen, P. 1983. Applied multiple regression/correlation analysis for the behavioral sciences. Hills-dale, NJ: Lawrence Erlbaum Associates.

Cohen, S. G., & Ledford, G. E. 1994. The effectiveness ofself-managing teams: A quasi-experiment. Human Re-lations, 47: 13–43.

Cook, T. D., & Campbell, D. T. 1979. Quasi-experimentation:Design and analysis issues for field settings. Boston:Houghton Mifflin.

Denzin, N. K., & Lincoln, Y. S. (Eds.). 2000. Handbook of qual-itative research. Thousand Oaks, CA: Sage.

Edmondson, A. C. 1996. Learning from mistakes is easiersaid than done: Group and organizational influences onthe detection and correction of human error. Journal ofApplied Behavioral Science, 32: 5–28.

Edmondson, A. 1999. Psychological safety and learning be-havior in work teams. Administrative Science Quarterly,44: 350–383.

Edmondson, A. C., Bohmer, R. M., & Pisano, G. P. 2001. Dis-rupted routines: Team learning and new technology im-plementation in hospitals. Administrative Science Quar-terly, 46: 685–716.

Edmondson, A. C., Winslow, A., Bohmer, R., & Pisano, G. 2003.Learning how and learning what: Effects of tacit andcodified knowledge on performance improvement fol-


lowing technology adoption. Decision Sciences, 34: 197–223.

Eisenhardt, K. M. 1989a. Building theories from case studyresearch. Academy of Management Review, 14: 532–550.

Eisenhardt, K. M. 1989b. Making fast strategic decisions inhigh velocity environments. Academy of ManagementJournal, 32: 543–576.

Fine, G. A., & Elsbach, K. D. 2000. Ethnography and experi-ment in social psychological theory building: Tactics forintegrating qualitative field data with quantitative labdata. Journal of Experimental and Social Psychology, 36:51–76.

Georgopolous, G. S. 1986. Organizational structure, problem-solving, and effectiveness. San Francisco: Jossey-Bass.

Gersick, C. J. G. 1988. Time and transition in work teams:Toward a new model of group development. Academy ofManagement Journal, 31: 9–41.

Giddens, A. 1984. The constitution of society: Outline of thetheory of structuration. Berkeley: University of Califor-nia Press.

Glaser, B. G., & Strauss, A. L. 1967. The discovery of groundedtheory: Strategies for qualitative research. New York:Aldine.

Goodman, P., Devadas, S., & Hughson, T. L. 1988. Groups andproductivity: Analyzing the effectiveness of self-managing teams. In J. P. Campbell & R. J. Campbell(Eds.), Productivity in organizations: 295–327. San Fran-cisco: Jossey-Bass.

Greene, J. C., Caracelli, V. J., & Graham, W. F. 1989. Towarda conceptual framework for mixed-method evaluationdesign. Educational Evaluation and Policy Analysis, 11:255–274.

Hackman, J. R. 1987. The design of work teams. In J. W. Lorsch(Ed.), Handbook of organizational behavior: 315–342.Englewood Cliffs, NJ: Prentice-Hall.

James, L. R. 1982. Aggregation bias in estimates of percep-tual agreement. Journal of Applied Psychology, 67: 219–229.

James, L. R., Demaree, R. G., & Wolf, G. 1984. Estimatingwithin-group interrater reliability with and withoutresponse bias. Journal of Applied Psychology, 69: 85–98.

Jick, T. D. 1979. Mixing qualitative and quantitative methods:Triangulation in action. Administrative Science Quar-terly, 24: 602–611.

Jorgensen, D. L. 1989. Participant observation: A methodologyfor human studies. Newbury Park, CA: Sage.

Kenny, D. A., & LaVoie, L. 1985. Separating individual andgroup effects. Journal of Personality and Social Psychol-ogy, 48: 339–348.

Keppel, G. 1991. Design and analysis: A researcher’s hand-book. Englewood Cliffs, NJ: Prentice-Hall.

Klein, K. J., & Kozlowski, S. W. J. 2000. Multilevel theory,research, and methods in organizations: Foundations,

extensions, and new directions. San Francisco: Jossey-Bass.

Lee, T. W., Mitchell, T. R., & Sablynski, C. J. 1999. Qualita-tive research in organizational and vocational psy-chology, 1979 –1999. Journal of Vocational Behavior, 55:161–187.

Maznevski, M. L., & Chudoba, K. M. 2000. Bridging space overtime: Global virtual team dynamics and effectiveness.Organization Science, 11: 473–492.

McGrath, J. E. 1964. Toward a “theory of method” for researchon organizations. In W. W. Cooper, H. J. Leavitt, & M. W.Shelly, II (Eds.), New perspectives in organization re-search: 533–547. New York: Wiley.

McGrath, J. E. 1984. Group interaction and performance.Englewood Cliffs, NJ: Prentice-Hall.

Meyer, A. D. 1982. Adapting to environmental jolts. Adminis-trative Science Quarterly, 27: 513–537.

Miles, M. B., & Huberman, A. M. 1994. Qualitative data anal-ysis. Thousand Oaks, CA: Sage.

Mohr, L. B. 1982. Explaining organizational behavior. SanFrancisco: Jossey-Bass.

Morgan, G., & Smircich, L. 1980. The case for qualitativeresearch. Academy of Management Review, 5: 491–500.

Nunnally, J. C., & Bernstein, I. H. 1994. Psychometric theory.New York: McGraw-Hill.

Pedhazur, E. J. 1982. Multiple regression in behavioral re-search: Explanation and prediction (2nd ed.). New York:Holt, Rinehart & Winston.

Perlow, L. A. 1999. The time famine: Toward a sociology ofwork time. Administrative Science Quarterly, 44: 57–81.

Rosenthal, R., & Rosnow, R. L. 1975. Primer of methods for thebehavioral sciences. New York: Wiley.

Sale, J. E. M., Lohfeld, L. H., & Brazil, K. 2002. Revisiting thequantitative-qualitative debate: Implications for mixed-methods research. Quality and Quantity, 36: 45–53.

Stewart, G. L., & Barrick, M. R. 2000. Team structure andperformance: Assessing the mediating role of intrateamprocess and the moderating role of task type. Academyof Management Journal. 43: 135–148.

Tabachnick, B. G., & Fidell, L. S. 1989. Using multivariatestatistics. New York: Harper & Row.

Tompkins, P. K., & Cheney, G. 1985. Communication andunobtrusive control in contemporary organizations. InR. D. McPhee & P. K. Tompkins (Eds.), Organizationalcommunication: Traditional themes and new directions:179–210. Newbury Park, CA: Sage.

Van Maanen, J. 1988. Tales of the field: On writing ethnogra-phy. Chicago: University of Chicago Press.

Wageman, R. 2001. How leaders foster self-managing teameffectiveness: Design choices versus hands-on coach-ing. Organization Science, 12: 559–577.

Weber, M. 1958. The protestant ethic and the spirit of capi-talism. New York: Scribner.

Weick, K. E. 1979. The social psychology of organizing. NewYork: McGraw-Hill.


Wheelwright, S. C., & Clark, K B. 1992. Revolutionizing prod-uct development: Quantum leaps in speed, efficiency,and quality. New York: Free Press.

Yauch, C. A., & Steudel, H. J. 2003. Complementary use ofqualitative and quantitative cultural assessment meth-ods. Organizational Research Methods, 6: 465–481.

Amy C. Edmondson ([email protected]) is Novartis Professor of Leadership andManagement and chair of the doctoral programs at Harvard Business School. Shereceived her Ph.D. in organizational behavior from Harvard University. Her researchexamines leadership and interpersonal interactions that enable organizational learn-ing in hospitals and other operational contexts.

Stacy E. McManus ([email protected]) is a management consultant atMonitor Executive Development in Cambridge, Massachussetts. She earned her Ph.D.in industrial and organizational psychology from the University of Tennessee. Herresearch interests include organizational mentoring relationships and career devel-opment.


METHODOLOGICAL FIT IN MANAGEMENT FIELD … · METHODOLOGICAL FIT IN MANAGEMENT FIELD RESEARCH AMY...

Documents

Transcript of METHODOLOGICAL FIT IN MANAGEMENT FIELD … · METHODOLOGICAL FIT IN MANAGEMENT FIELD RESEARCH AMY...