Porter's (1980) Generic Strategies as Determinants of Strategic Group Membership and Organizational Performance Author(s): Gregory G.

Dess and Peter S. Davis Source: The Academy of Management Journal, Vol. 27, No. 3 (Sep., 1984), pp. 467-488 Published by: Academy of Management Stable URL: http://www.jstor.org/stable/256040 . Accessed: 11/08/2011 02:43
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management Journal.

http://www.jstor.org

Journal ?Academyof Management 1984,Vol. 27, No. 3, 467-488.

Porter's

(1980)

Generic Strategies as Determinants of

and

Strategic Group Membership Organizational Performance1

GREGORYG. DESS FloridaState University PETER S. DAVIS Universityof Oregon

A multimethod,multivariateanalysis of "intended" providesempiricalsupportfor thepresenceof strategies strategic groupsbaseduponPorter's(1980)genericstratin and egies. Variations intraindustry profitability growth arefound to be relatedto strategicgroup membership. Firmsidentifiedwithat least one genericstrategyoutperformed firms identifiedas "stuck in the middle." The primarypurposeof this paper is to demonstratethe viability and usefulnessof categorizingfirms within an industryinto strategicgroups on the basis of theirintendedstrategies.Theseintendedstrategiesmay be identified on the basis of Porter's (1980) generic strategies-differentiation, overalllow cost, and focus. This study consists of threedistinctbut interrelated bephases. Phase 1, the field study, examinesthe relationship tween a firm's "intendedor espoused" (Mintzberg,1978)strategy-representedby the competitivemethods(e.g., competitivepricing)considered most importantby the firm's top management team-and the presenceof orientationswithin an industry.These strategicorientationsare strategic classifiedon the basis of which of the three alternativegenericstrategies they appearto representmost closely. Phase 2 consists of a panel of experts who assess the importance of each of the identified competitive methodsfor eachgeneric The strategy. use of thispanelservesto corroborate the researchers' inferencesdrawn from the field study. Phase 3 uses the perceptionsof the chief executiveofficers to cluster firms that exhibit a
'The authors wish to express their appreciation for the thoughtful comments made by Alan Bauerschmidt, Danny Miller, and Carolyn Woo on an earlier draft of this paper, as well as the helpful suggestions of three anonymous reviewers. An earlier version of this paper was presented at the 1982 Academy of Management Meetings in New York. This research was supported, in part, by grants from the Division of Research, College of Business Administration, University of South Carolina, and the Edna Benson Foundation at the University of Washington. 467

468

Journal Academyof Management

September

of into distinctgroups.Measures performance similarstrategicorientation to detect whether for firms comprisingeach strategicgroup are assessed significantdifferencesexist between strategicgroups within an industry. have importantimplicationsfor both practicDifferencesin performance interestedin the strategicgroup concept. ing managersand academicians

TheoreticalBackgroundand EmpiricalResearch
shiftawayfrom has The fieldof strategic management showna noticeable the atomistic view of strategy-in which each firm is consideredunique in all aspects-toward a new view that supportsthe recognitionof comhavebeen referred monalitiesthat existamongfirms. Theseconfigurations to as "gestalts"(Hambrick,1983b;Miller, 1981),said to represent"tightly integratedand mutuallysupportiveparts, the significanceof whichcan best be understoodby making referenceto the whole" (Miller, 1981, p. 3). Similarly,Hatten (1979), in his discussionof strategicgroups within an industry,recognizedthat subgroupsof firms employ different mixes variables. of whataresubstantially samestrategic the Thus, strategic groups a useful intermediateframe of referencebetweenviewingthe inprovide each firm separately (Porter, 1980).The dustryas a whole and considering of firms providesa frameworkfor emergingconcept of a strategicgroup recentcalls for empirical"evidencethat strategiesdiffer among answering results" firmsand that betterstrategiesmake a differencein performance (Schendel& Hofer, 1979, p. 517). Thepresence groupsof firmswithinan industry of followingsimilarstrategies has been identifiedin the home appliance(Hunt, 1972),the chemical process(Newman,1973),the consumer goods (Porter,1973),andthe brewmodels of the brewingindustry ing industry(Patton, 1976). Quantitative (Hatten& Schendel, 1977;Patton, 1976)recognizedthat firms within an industrydiffer along dimensionsother than size and market share. measurement strategyused of Unfortunately,most of the multivariate on has reliedalmostexclusively meato developthe strategic groupsconcept sures of implementedstrategy.One problemwith existingtypologieshas been that few of the propositionsregardingthe types of strategiesa firm may follow to become a leaderin its markethave been tested with data differentfrom that used to developthem (Schendel& Hofer, 1979). This methodologicallimitationmay lead to findings that lack generalizability and that are incapableof confirmationin otherresearchsettings.Another problemis that existingtypologiestend to put businessunit strategiesinto genericcategoriesbasedon the size or marketshareof the firmand its rate of returnon investment(Hatten, 1974;Porter, 1979). Such a narrowapof proachlimitsthe applicability many typologiesto a ratherselectgroup of firms within an industryand presentslittle in the way of prescriptions useful acrosssize categorieswithinan industry.It is believedthat Porter's frameworkof generic strategiesand competitivedimensionsprovides a

1984

Dess and Davis

469

potentiallyvaluableresearchtool for classifyingthe strategiesof all competitors within an industry. Many previousmodels of strategicgroups (Hambrick, 1983a;Porter, strategiescould best be inferredby 1974)have impliedthat organizational resourceallocations.However,the reanalyzingpatternsin organizational liance upon measuresof resourceallocationsmay inhibit recognitionof the centralthreador underlying logic of a firm's strategyby failingto considerthe roleof strategic choiceas exercised key organizational members. by Thenotionof strategic that choicerecognizes similar organizations operating withinthe sameenvironment difmay choose to addressthat environment based on the strategicorientationof their management(Ackoff, ferently 1970). Indeed, the elementof strategicchoice is inherentwithin the concept of strategy,which is viewed as implyingthat "the organizationpurin sues a purposive,directivecourse,whetherit is described termsof a pattern of decisions, or in terms of the goals, plans, or intentions of the organization"(White& Hamermesh,1981, p. 216, emphasisadded). Previous evidencefor the importanceof strategicchoice for marketbehavior, although compelling,has relied on indirectmeasuresto show that firms follow different strategieswithin an industry(Caves, 1980). Hambrick(1980)recognizedthat it shouldbe possibleto developmultivariatemeasuresof intendedstrategiesas well as implementedstrategies. Demonstration the abilityof a multivariate of measureof strategicchoice to classify firms into homogeneousgroups based on Porter's model of wouldprovidemuchneededempirical evidence,not only genericstrategies for the constructvalidityof Porter's (1980)typology but also for the notion of strategicgroups in general. Porterdevelopsthreepotentially for successful genericstrategies creating a defensibleposition and outperforming in a given industry. competitors The first, overallcost leadership,althoughnot neglectingquality, service, and other areas, emphasizeslow cost relativeto competitors.The second strategy,differentiation,requiresthat the firm createsomething,eithera as productor a service,that is recognizedindustrywide beingunique, thus the firm to commandhigherthan averageprices. The third is permitting a focus strategy,in which the firm concentrateson a particulargroup of customers, geographicmarkets, or product line segments. These three threebroadtypes of strategicgroups, and thus genericstrategies represent the choiceof strategy"can be viewedas the choiceof whichstrategic group to competein" (Porter,1980,p. 149).Firmsorientedtowardspecificstratfirmscharacterized Porteras "stuckin the midegiesshouldoutperform by dle." Portermaintainsthat this latterclass of firms, by failing to develop its strategyalong at least one of these threecategories,is "almost guaranteed low profitability"(1980, p. 41). Theunderlying of in proposition thispaperis thatvariations intraindustry profitabilityand growthmay be explainedon the basis of strategicgroup The strategicgroupsmodel also servesto explaindifferences membership. in performance sizes:evenwhenfirmsaredivided amongfirmsof equivalent

470

Academy of Management Journal

September

of by sizeclasswithinan industry,eachdivisionmay containmembers more than one strategicgroup. Method Overview The field studywas comprisedof responsesfrom executivesof a sample of firmsthat wereusedto developthe dimensionsassociatedwiththe three genericstrategiesoffered by Porter. For the second phase, a panel of expertsfrom the academiccommunityprovidedrecommendations regarding content of each of Porter'sgenericstrategies.The use of the appropriate a panelof experts strategy in followsthe recommendations Harof research of and inferences rigan(1983)for the refinement corroboration investigator developedfrom field research.Thus, the first two phasesof the study enabledthe researchers combinethe descriptive to capabilityof field research with the normativerecommendations obtained from a panel of experts. Phase 3 of the study classifiedfirms with similarstrategicorientations into distinctgroups. These groupswere developedfrom the responsesof the subsample firmsin whichthe chief executiveofficerwas the primary of for respondent.Measuresof performance firms comprisingeach strategic groupwereassessedto detectthe presenceof significantdifferencesamong the strategicgroups.
Phase 1: Field Study

as Strategicorientationswithinan industrywereseen by the researchers teams about the competitive by represented the views of top management methodsfirmsuse in theirindustry.Whenthe research instrument conwas structedit was assumedthat all membersof the top management team had knowledgeof the strategyof their firm, and that the strategycould be inferredon the basisof the emphasisor importance givenvariouscompetitive methods availableto the firm. Porter recognizesthat the strategiesthat companiesuse to compete in an industrycan differin a widevarietyof ways, and he proposesa number of "strategic dimensions" shouldcapture possibledifferences that the among the strategicoptions of companiesin a given industry.These dimensions are comprisedof competitivemethods that include brand identification, channel selection, technological leadership, cost position, service, and methodsprovide leverage,amongothers(Porter, 1980).Thesecompetitive a meansfor characterizing strategies competitors of withinan industry. the A groupof firmswithinan industry that followsthe sameor a similarstratsimilarcompetitive egy (i.e., emphasizes methods)will comprisea strategic group (Porter, 1980). Research Instrument. Followinga reviewof Porter(1980),an instrument was inductivelyderivedto evaluatethe variouscompetitivemethodsthat

1984

Dess and Davis

471

a might be used to characterize particulargeneric strategy. The content was enhancedthrougha reviewof questionvalidityof the questionnaire naireitems used by previousstrategyresearchers (Bourgeois,1980;Child, 1975;Khandwalla,1976). In orderto pretestthe researchinstrumentin a field setting, the CEOs of four manufacturing firms, not includedin the to the and finalsample,wereinterviewed ascertain comprehensiveness phrasing of the questionnaireitems. Next, the relevanceof particularcompetitivemethods for firms within the industrywas ascertained interview throughthe use of a semi-structured in withthe chiefexecutive officer(or designated executive) eachof the sample firmswithinthe industrychosen for the study-the paintsand alliedproductsindustry.Duringthe interview executive askedto identifythose the was of teamwho aremost influentialin the makmembers the top management decisionsfor the firm. By incorporating richness anecthe of ing of strategic dotal informationabout the firm, the industrywithin which it competes, and methodsof competitionin that industry,the researchers were able to modify the instrumentin orderto enhanceits ability to capturecompetitive methods that identify the strategicorientationof a firm's decision makers.In orderto generalizeacrossthe sample,the sameinstrument was administered all respondents. to The interviewswith the executivesof the pretestand sample firms led to severalimprovements both the wordingand the compositionof the in list of competitivemethods.A modificationthat helpedrefinethe final instrumentillustratesthe benefit of on-site interviews.The presenceof intensepricecompetition mentioned severalrespondents suggested was and by the inclusionof the item "competitivepricing." The sentimentsof the exin ecutives thisregard largelyconsistent are witha U.S. Department Comof merce industryreport that stated that the profit-after-taxes a percent as of salesfor the paintindustry(the site of this study)was2.25 percent,comindustriesof 5.25 percent.The paredto an averagefor all manufacturing reportfurthersuggestedthat the depressedprofits were largelyattributed to a price-costsqueeze, reflectedby the 1978 ProducersPrice Index for to paintof 192.3(1967= 100)compared the averagefor paintrawmaterials of 212.7 (U.S. IndustrialOutlook, 1980). Sampleand Data Analysis. Severalcriteriawereused in the selectionof firmswithinthe paintsand alliedproductsindustry.First,the 4-digitStandardIndustrial Classification mea(SIC)codewas chosenas an appropriate sure of the industrialenvironment.This unit of analysisis supportedby Porter(1980). Second, the output of firms had to be concentratedin one line of businessto avoid confusion between methods used in competing in multiplebusinesses.The criterionfollowedRumelt's(1974)"singlebusiness" or "dominantbusiness" categorization: that is, at least 70 percent of the firm'stotal saleshad to be withina given4-digitSICindustry.Third, the organizationhad to be an autonomous, self-containedentity. Thus, the researchers ableto consider were and stratcorporate-level business-level as synonymous(Hofer, 1975). Fourth, the organizationshad to be egies

472

Journal Academyof Management

September

to relativelyhomogeneousin size. This criterionpermittedthe researchers controlfor any potentiallyconfoundingeffects imposedby widevariations in organizational resourcesand scope of operations.Fifth, the organizationshad to be similarin the technology theyemployed.Woodward's (1965) classificationof manufacturing processeswas used to classify the sample as organizations using a batch technology.Finally, the organizationshad to be locatedwithina limitedgeographicareain orderto facilitatethe performanceof the on-site interviews. Basedon an analysisof forcesthat drivecompetition,Porter(1980)clasof sifies industriesas representative one of five genericindustrialenvironor ments:(1) fragmented, emerging, mature,(4) declining, (5) global. (2) (3) The paints and allied productsindustryhas a 4-firm concentrationratio of less than 40 percent(i.e., 22 percent).There is no true marketleader, and thereare a largenumberof smalland medium-sized companies,many of them privately-held. Thesecharacteristics identify the paints and allied an that by productsas highlyfragmented, identification is confirmed Porter (1980). In the field study, 28 nondiversified firms in the paints manufacturing and alliedproductsindustry(SIC 2851) wereinitiallycontacted.Of these, 22 fully participated. The final industry-specific was questionnaire mailed to each CEO for distribution the previouslyidentifiedtop management to team members.These individualswere asked to indicatethe importance of the 21 competitivemethods (e.g., customerservice, brand identification) to their firm's overallstrategy.A 5-point scale was used with values rangingfrom "1 = Not at all important"to "5 = Extremelyimportant." A total of 78 of 99 possiblerespondents(79 percent)from the 22 sample firms completedthe questionnaire. Factor analysis of the questionnairedata on competitivemethods was usedto developthe competitive dimensions associatedwitheachof Porter's Factoranalysishas the abilityto producedescriptive sumgenericstrategies. mariesof datamatrices,whichaid in detectingthe presenceof meaningful patternsamong a set of variables. Resultsof Phase 1. The principalfactorsolutionobtainedafterthe varimax rotationfor the 21 competitive methodsis shownin Table1. Although five significantfactors(i.e., eigenvalues 1) emergedfrom the factoranal> the ysis, only the threefactorsexplaining greatestamountof total variance are shown in Table 1. The other two factorswere droppedin the interest of parsimony;additionally,a screetest (Cattell, 1976)indicatedthat they should be excluded. The factors displayedin Table 1 are rank orderedfrom left to rightaccordingto the proportionof total variancethey explained.The factorsare named to reflect the three genericstrategiesthey were interpreted the by writersas representing. methodsexhibited factorloadingsgreater Overall,15of the21 competitive than or equal to ? .50 on at least one factor. These loadingsmay be considered to be a conservativecriterion;Kim and Mueller (1978) suggest

Table 1 CompetitiveMethods: Factor Structureand Commun

Factor One Differentiation Squared Factor Factor Loadings Loadings
a a1) ( (a2)

Competitive Methods

Factor Two Low Cost Squared Factor Factor Loadings Loadings

(a2 (a22)

V1. New productdevelopment V2. Customer service V3. Operating efficiency V4. Productqualitycontrol V5. Experienced/trained personnel V6. Maintainhigh inventorylevels V7. Competitive pricing V8. Broadrangeof products V9. Developing/refining existingproducts V10. Brandidentification in Innovation marketing V11. techniquesand methods V12. Controlof channelsof distribution V13. Procurement raw materials of V14. Minimizing of outsidefinancing use V15. Servingspecialgeographic markets to V16. Capability manufacture specialtyproducts V17. Productsin high pricemarketsegments V18. Advertising V19. Reputation withinindustry V20. Forecasting marketgrowth V21. Innovation manufacturing in processes Eigenvalue Percentof commonvariance Percentof total variance

.19858 .03943 -.26645 .07100 .29412 .08650 .16526 .02731 .05293 .00280 .24855 .06118 .04730 .00223 .02949 .00087 .19764 .03906 .82943 .68795 .85953 .73879 .70853 .50201 .50326 .25327 .23042 .05309 .17321 .03000 -.08241 .00679 .22651 .05131 .83112 .69076 .04930 .00243 .55085 .34034 .44429 .19739 6.7871 62.60 32.3

.15352 .02357 .48492 .23515 .51166 .26180 .80309 .64495 .58847 .34630 .07925 .00628 .00040 -.01997 -.11203 .01255 .61536 .37867 .12707 .01615 .20290 .04117 .29166 .08507 .61069 .37294 .30128 .09077 .10626 .01129 .16097 .02591 .00842 .00070 .01627 .00026 .78639 .61841 .51302 .26319 .61579 .37920 2.2416 20.74 10.7

-. -.

474

Journal Academyof Management

September

factorloadingsof .30 as a cutoff for significance.Similarly,Nunnallysuggests that "it is doubtfulthat loadingsof a smallersize be taken seriously, becausethey represent than ten percentof the varianceof the factor" less methods(i.e., V13, and V20)loaded (1978,p. 423). Two of the competitive highly on more than one factor, indicatingthat they may be germaneto more than one generic strategy. A commonmethodologicalweaknessthat mightthreatenthe reliability and validityof the factor analyticresultsis the possibleinstabilityof the factor loadings. Instabilityof the factor loadingsbecauseof samplingerror may resultfrom the use of a relativelysmall ratio of subjects(n = 78) to measures(n = 21). This ratio of 3.7 approachesbut does not exceedthe desirablebut conservative ratio of four or five to one advocatedby some authors(Hair, Anderson,Tatham,& Grablowsky,1979).The samplesize does exceed that suggestedby Lawleyand Maxwell(1971) for the maximumlikelihoodsolutionmethodof confirmatory factoranalysis.Theysugif gest thatthis test is appropriate the samplecontainsat least 51 morecases than the numberof variablesunderconsideration.Furthermore, giventhe exploratorynatureof the researchquestionas well as constraints-time, of resources,availability firms-inherent in field research,the samplesize of this study is not considereda significantlimitationin interpreting the results.
Phase 2: Panel Study

A panel of expertswas used to developnormativerecommendations regardingthe appropriatecontent of each of Porter's (1980) three generic selectedon the basis strategies.The panelconsistedof sevenacademicians of theirexperience expertisein the field of strategicmanagement.The and panelwas used to evaluatePorter's(1980)conceptsregarding genericstrategies becauseof the common core of knowledgethey possessedrelevant to the area of inquiry.Each panel memberwas asked to reviewPorter's and chapteron "Generic Competitive Strategies" thencompletethreequestionnaires,one for each of Porter'sthreegenericstrategies.Eachquestionnaireconsistedof the same21 competitivemethodsand associated5-point scalesused in the field research.The panelmemberswereaskedto indicate the importanceof each competitivemethodfor each genericstrategy.The data wereanalyzedto determine whetherexpertswould uniquelyorganize the competitivemethodsfor each strategy.Furthermore, because"expert judgeswill not show well-knownbiasesas leniency, halo, etc." (Einhorn, 1974, p. 562), it is reasonableto expectthat the variablesbeing measured actually represent explanatory concepts when combined into global
constructs.

ResultsofPhase2. Thedescriptive statistics(means,standard deviations) for the panel's responsesare shown in Table 2. The statistics developed were used to evaluatewhich competitivemethods the expertsconsidered most importantfor each of Porter's genericstrategies.Table 2 indicates

1984

Dess and Davis

475

Table 2 Panel Technique:DescriptiveStatistics

Methods Competitive V1. New productdevelopment V2. Customerservice V3. Operating efficiency V4. Productqualitycontrol V5. Experienced/trained personnel V6. Maintainhigh inventorylevels V7. Competitive pricing V8. Broadrangeof products V9. Developing/refining existingproducts V10. Brandidentification V11. Innovationin marketing techniques and methods V12. Controlof channelsof distribution of V13. Procurement raw materials V14. Minimizing of outsidefinancing use V15. Servingspecialgeographic markets to V16. Capability manufacture specialty products V17. Productsin high price marketsegments V18. Advertising withinindustry V19. Reputation V20. Forecasting marketgrowth V21. Innovationin manufacturing processes Meanitem value Meanstandarddeviation Differentiation Strategy Mean S.D. 4.71 4.29 2.57 4.29 4.57 2.57 1.71 1.14 4.00 5.00 4.86 4.14 2.43 2.29 2.71 3.86 4.57 4.71 4.29 3.29 2.57 3.60 .49 .49 .53 .49 .53 .53 .49 .69 1.00 .00 .38 .69 .53 .49 1.25 1.07 .53 .49 .49 .49 1.13 1.07 OverallLow Cost Strategy Mean S.D. 2.29 1.71 5.00 3.00 3.57 3.14 4.86 2.14 3.86 1.86 1.71 3.00 4.86 3.29 1.57 1.14 1.29 2.43 2.57 4.00 4.14 2.93 1.25 .49 .00 1.00 .98 1.21 .38 1.07 .69 .69 .76 1.41 .38 .95 .79 .38 .49 1.72 .98 .82 1.07 1.21 Focus Strategy Mean S.D. 3.71 4.29 3.00 3.57 3.57 2.57 3.57 2.43 3.29 4.57 4.14 3.29 2.71 2.43 4.86 5.00 3.71 3.86 3.50 2.86 3.71 3.55 .49 .76 1.00 .79 .53 1.27 .79 .53 1.11 .53 .90 .49 .76 .79 .38 .00 .76 1.07 1.04 .69 1.38 .74

thatthe panelof expertsidentifiedsomecompetitive methodsas eithermost or least importantfor each of the genericstrategies.The analysisclarified the relationships methodsand the genericstrategies betweenthe competitive of and aided in the interpretation the factor analysisdevelopedfrom the field study. of of Comparison the Results Phase1 andPhase2 An important in the dataanalysisfor Phase 1 involveda comparison step between:(1) the principalfactor solutionwith varimaxrotationfor the 21 methodsusedin the fieldstudyand (2) the opinionsof the panel competitive as to the appropriate contentof each of the genericstrategies.Thesefindin are summarized Table 3. ings Table3 is presented the basisof Porter'sthreegeneric on Under strategies. each genericstrategy,the competitivemethodsidentifiedby the managers and the panel of expertsare arrayedas most and least important.For the fieldstudy,the mostimportant methodswerethoseitemswhose competitive factor loadings were greaterthan .50. The least importantcompetitive methods were those items whose factor loadingswere between +.30 and -.30 (Kim& Mueller, 1978;Nunnally, 1978). For interpretive purposes, competitivemethodsthat exhibitedeithertheir highestor lowest loadings

476

Academy of Management Journal

September

on the two factorsthat wereexcludedfrom analysiswerenot incorporated into Table3. For the panel, the most importantand least importantcompetitivemethods consisted of those items whose mean value was greater than or less than, respectively,one standarddeviationfrom the aggregate mean.

Table 3 Content of GenericStrategies:Summaryof Findings

Differentiation Experts Managers (most important) V10. Brand identification V1. New product development V 1. Innovation in marketing techniques and V10. Brand identification methods V11. Innovation in marketing techniques and methods V12. Control of channels of distribution V18. Advertising V13. Procurement of raw materials V18. Advertising V20. Forecasting market growth (least important) V2. Customer service V7. Competitive pricing V8. Broad range of products V13. Procurement of raw materials V14. Minimizing use of outside financing Overall Low Cost Experts V3. V7. V13. V21. Operating efficiency Competitive pricing Procurement of raw materials Innovation in manufacturing processes Managers important) V3. Operating efficiency V4. Product quality control VS. Experienced/trained personnel V9. Developing/refining existing products V13. Procurement of raw materials V19. Reputation within industry V20. Forecasting market growth V21. Innovation in manufacturing processes (least important) Customer service V8. Broad range of products Innovation in marketing techniques and V15. Serving special geographic markets methods V18. Advertising Serving special geographic markets Capability to manufacture specialty products Products in high price market segments Focus Experts V2. V10. V15. V16. V6. V8. V13. V14. Managers (most important) Customer service VI. New product development Brand identification V16. Capability to manufacture specialty products Serving special geographic markets V17. Products in high price market segments Capability to manufacture specialty products (least important) Maintain high inventory levels V3. Operating efficiency Broad range of products V6. Maintain high inventory levels Procurement of raw materials V7. Competitive pricing V12. Control of channels of distribution Minimizing use of outside financing V13. Procurement of raw materials V14. Minimizing use of outside financing V21. Innovation in manufacturing processes

V2. VI1. V15. V16. V17.

1984

Dess and Davis

477

methodsidenThe resultssuggestthat factorone consistsof competitive tified by the expertsas most importantto a differentiation strategy.Three of the four competitivemethods identifiedby the panel for this generic in led are strategy contained factorone. Thiscloseassociation the researchers of to interpretfactor one as representative a differentiation strategy.This is interpretation supportedfurther:only one (i.e., V13) of the competitive methodsidentifiedby the panel as least importantappearedwith a high loading on factor one. Many of the competitivemethods loading highly on factorone appear be indicative a marketing of to orientation (e.g., brand Porter(1980)notesthatone of the principal economiccauses identification). industriesis high productdifferenfor the existenceof highlyfragmented if tiation, particularly it is based on image. factortwo containsthreeof the fourcompetitive methodsidenSimilarly, tified by the panel as most importantto an overalllow cost strategy.Factor two suggestsa predominantly productionorientation(e.g., operating efficiency).Such an orientationlends supportto the identificationof facof tor two as representative Porter'soveralllow cost strategy.Portercona tendsthat "achieving low overallcost positionoftenrequires highrelative a suchas favorable marketshareor otheradvantages, accessto rawmaterials" (1980, p. 36). Although therehas been some empiricalresearchto refute the requirement a high relativemarketsharein orderto obtain a low for cost position (Hall, 1980), it should be noted that one of the competitive methodswith a high loading on this factor was V13 (i.e., procurement of rawmaterials).As with factor one, interpretive supportis indicated;none of the competitive methodsregarded the panelas leastimportant this to by with high loadings on factor two. strategyappeared Moresimilarity on betweenthe paneland managers whatdoes appeared not ratherthan what does constitutea focus strategy.Three of the four methodsseenby the panelas leastimportant a focus strategy in competitive in factor threewith theirlowest factorloadings. However,those appeared methods(e.g., V16-specialty products-and V17-high priced competitive marketsegments)exhibitinghigh loadings on factor three do suggestan emphasison specificmarketsegmentsand imply concentrationon a particularniche.
Phase 3: StrategicGroup Membership and Organizational Performance

Phase 3 uses the responsesfrom the 19 firms whose CEO, ratherthan a "designated The chief executive executive,"was the primary respondent. is most importantin strategyformulation(Chandler,1962). Chief executives' perceptions theirorganizations' of are strategies morecloselyaligned to externalmeasuresof strategythan are the perceptionsof other executives(Hambrick,1981b).Therefore, CEOs'responses the wereusedto catefirmsinto clusters similarintendedstrategies. Performance gorize reflecting data wereprovidedfor 15 of these firms and wereused to analyzedifferencesin performance amongthe clusters.Thisreductionin the samplesize

478

Journal Academyof Management

September

firms(Porter, is regrettable not uncommon the studyof privately-held but in 1979). calculated Theresearchers individual factorscoresfor eachCEOon each to of the threefactorscorresponding Porter's(1980)threegenericstrategies in accordancewith the following formula:
A= alizl + a2;i2 + ? ?.+ ajizj,

where:aj;is the factor score coefficient for competitivemethod j (= 1, value ..., 21) on factor i (i= 1, 2, 3), and zj is the CEO's standardized on competitivemethodj. The factor scores generatedfor each of the CEOs were used as input to a K meansclustering algorithm(Dixon, 1975).The CEOswereswitched from one clusterto anotheruntil the optimalclusterconfiguration,which maximizesbetweenand minimizeswithinclustervariances,was obtained. The clustersconsistedof subsetsof CEOsthat weremore similaror closer to each other in orientationthan they were to CEOs outside the cluster. Initiallya three clustersolution was chosen in order to facilitatecomparison with Porter's typology of three generic strategies.However, the three clustersolution did not adequatelydistinguishamong the clusters. with the exploratory Consistent natureof this research,a four clustersolution also was examined.The clustersthat resultedconsistedof CEOswho were most similar in respect to the factors (genericstrategies)that had emergedfrom Phase 1 of this study. The overallsignificanceof the clustersolutions obtainedwas tested by means of a one-way analysisof variance-based on the assumptionthat the scoresin eachof the variousgroupshave approximately same varithe ance. However,becausethe variousgroupsdo not containthe same numberof subjects,a Bartlett'sBoxF-test on the within-cell variances used was to test for homogeneityamong variances. Using Parsons'(1956)classificationof organizations type of goal or by function, the organizationsin this study would be classified as oriented toward economic production.Therefore,consistentwith the "economic primacy"of the subjectorganizations,the derivedgoals of the financial community(e.g., profitability,growth)should be consideredof primary in of importance the assessment organizational performance. Consequently, the researchers obtained"total firm sales" figuresfor the period 1976 to 1980to determine"sales growth" and "averageafter tax returnon total assets" from 1976to 1980to determine profitability.Annual salesgrowth ratesfor the five-yearperiodrangedfrom .8 percentto 37.7 percent,and averageaftertax returnon total assetsrangedfrom 1 percentto 42 percent for the sample. An F-valuewas calculated usinga one-wayanalysisof varianceto decide whether thereweresignificant differences amongthe strategic groups(clusters)on the basis of theirmean valuesfor the two performance measures. To determinewhere the significant differenceslie, Scheffe's posteriori

1984

Dess and Davis

479

means. contrasttest was usedto compareall possiblepairsof performance Among the variousmultiplecomparisontests, Scheffe's is consideredto test be the most conservative (Huck, Cormier,& Bounds, 1974).This test to also offers the advantagesof applicability groupsof unequalsizes and is relativelyinsensitiveto departures from normalityand homogeneityof variances(Hays, 1963). Resultsof Phase3: The ThreeClusterSolution.The profilesof the three fromthe clusteranalysisandthe performance strategic groupsthatemerged resultsare presentedin Table 4.

Table 4 of Cluster AnalysisResultsandPerformance Summary Relationships: The Three ClusterSolution

A. Mean Scores Cluster 1 (n= 12) 2 (n=4) 3 (n=3) Grand Means B. Mean Squares Differentiation Between groups 2.9503 Within groups .3083 2,16 d.f. F-ratio 9.568 .001 P-value = .086, p = .917 Bartlett's Box F-test Coordinate Centroids Overall Low Cost Focus Differentiation .5057 (.4332)c -.5063 (.5977) -1.0351 (.6462) .1441 .2437 (.8034) -.3723 (.3266) -.1808 (.7499) .0470 Overall Low Cost .6616 .6162 2,16 1.074 .388 .4000 (.5247) -1.2356 (.3484) .5225 (.4516) .0750 Performance Annual Return on Sales Growtha Total Assetsb .197 .084 .308 .118 .055 .044

Focus 4.3692 .2751 2,16 15.883 .001

aAnalysis of variance results for the three clusters: F= 5.531, p .022 bAnalysis of variance results for the three clusters: F=.724, p=.507 CStandarddeviations are in parentheses.

The one-wayanalysisof varianceprocedure indicatedthat the threeclusters of firms were significantlydifferent from each other on the basis of theiremphasison the differentiation (p strategy < .001)andthe focus strategy (p < .001)but not on the basisof the overalllow cost strategy = .388). (p The researchers anticipated eachof Porter'sthreegenericstrategies had that would be requiredto distinguishclearlythe different strategicemphases that existedamongthe clustersof firms. Thus, the inabilityof the low cost dimensionto accountfor significantdifferencesamong the three clusters wasunanticipated provided rationale developing fourcluster and for the the solution. The firstclusterhad its highestcentroidscoreon the differentiation strategy, and it was the only clusterthat displayeda positivescore for the differentiationstrategy.Clusternumbertwo displayednegativescoresfor all

480

Journal Academyof Management

September

three of the genericstrategies.This lack of a positive score on any of the this that firmscomprising grouphavefailed indicates threegenericstrategies to developtheirstrategyin at least one of Porter's(1980)threedirections. Clusternumberthreeexhibitedthe highestpositivescoreof any clusterfor and the focus dimension negativescoresfor the othertwo genericstrategies. in combination,the patternof centroidscoresthat emergedfrom Taken to the clusteranalysisled the researchers identify clusterone as representative of a strategicgroup of firms orientedprimarilytoward a differentiation strategy.However,the patternof scores was not conclusive(i.e., the relativelyhigh scoreson the alternative genericstrategiesindicatedthe withinthis cluster of emphasison morethanone genericstrategy possibility of firms).Clusternumberthreeappearsto consist of a groupof firmsemphasizinga focus strategy.Clusternumbertwo evidencesan apparentlack of commitment any of Porter'sgenericstrategies.Therefore,this cluster to be comprisedof firms that are "stuck in the middle" (Porter, 1980, may p. 41). This does not imply that firms that are stuck in the middle do not emphasizecertaincompetitivemethods that are key componentsof one or more genericstrategies;however, the composite strategythat emerges may lack internalconsistency. The next step in the analysiswas to compareperformanceamong the threeclustersof strategicgroups. Table4 indicatesthat the threestrategic groups were not significantlydifferent from one anotherwith regardto profitabilityas measuredby returnof assets (p = .507). However, for the other performancemeasure,annual sales growth, Table 4 indicatesthat was strategicgroupmembership significant(p = .002). This findingis consistentwithPorter's contention firmsthatadopta generic that strategy (1980) shouldoutperformthose stuckin the middle. Clusterone, the differentiation strategy 19.7percent annualsalesgrowth;clusterthree, group,averaged the focus group, averaged percentannualsalesgrowth;and clustertwo, 31 stuck in the middle, averagedonly 8 percent annual sales growth. The Scheffe test for significantdifferencesamong the groups on the performance measure"sales growth" shows that the mean value for the focus strategygroup is significantlygreaterthan the mean value for the stuck in the middle group. None of the other differencesis significant. Resultsof Phase3: Four ClusterSolution. The profilesof the four strategic groupsthat emergedfrom the clusteranalysisand theirlinks with organizationperformanceare presentedin Table 5. The four cluster solution resultedin the splittingof clusternumberone in Table4 into two separateand uniqueclusterslabeled one and four in Table 5. The composition of clustersnumberedtwo (i.e., stuck in the middle) and three (i.e., focus) remainedintact from the three clustersolution to the four cluster solution. Theone-wayanalysis variance of indicated the fourclusters that procedure of firms weresignificantlydifferentfrom each other on the basis of their dimension(p < .003), the focus dimension emphasison the differentiation (p<.001), and the overall low cost dimension(p<.001). Table 5 shows

1984

Dess and Davis

481

Table 5 of Cluster Resultsand Performance Analysis Relationships: Summary The Four Cluster Solution
A. Mean Scores Cluster 1 (n = 4) 2 (n=4) 3 (n= 3) 4 (n=8) GrandMeans B. Mean Squares Betweengroups Withingroups d.f. F-ratio P-value CoordinateCentroids Overall Focus Differentiation Low Cost .3268 -.5063 (.6902) -1.0351 (.7914) .5951 (.4195) .1441
(.5254)C

Performance Annual Returnon Sales Growtha TotalAssetsb .201 .084 .308 .170 .255 .055 .044 .089

1.2685 -.3723 (.3771) -.1808 (.9185) -.2687 (.4445) .0470

(.1444)

.6226 -1.2356 (.4023) .5225 (.5531) .2887 (.3176) .0750

(.8757)

Overall Differentiation Low Cost 2.0309 .3161 3,15 6.425 .003 2.5417 .2372 3,15 10.715 .001

Focus 3.0119 .2736 3,15 11.009 .001

Bartlett's Box F-test= .567, p= .638

of aAnalysis varianceresultsfor the four clusters:F= 3.89, p= .041 of bAnalysis varianceresultsfor the four clusters:F= 3.24, p= .069 cStandard deviationsare in parentheses.

that althoughclusterone displayspositivecentroidscores for each of the threegenericstrategies,the emphasison overalllow cost strategyis clearly dominant.Also, the primary orientation the fourthclusteris a differenfor tiation strategy.Thus, though the three cluster solution was anticipated to parallelPorter'sthreegenericstrategies, four clustersolutionaffords the an interpretation that is more consistentwith Porter's framework. It is importantto note that clusteranalysis,unlikemost parametric statisticaltechniques,does not explicitlyprovidea clearlyacceptableor unacso ceptablesolution. It merelyprovidesa structure that relationships may it is importantto make explicitthe criteriathat guidedthe emerge;thus, selectionof an appropriatesolution. In the presentcase, both subjective and objectivecriterialed to the selectionof the four clustersolutionas the moreappropriate. subjective The criterion of involvedan identification how manyclusterstheorywouldlead one to expect.Initially,threeclusterswere one The anticipated, for eachof Porter'sthreegeneric strategies. appearance of the stuckin the middlegroupas a clusterin the initialthreeclustersolution suggested that the thirdgenericstrategy-overall low cost-which had not appeared a significant as determinant groupstructure of may havebeen pooled into one or more of the other groups. The most importantobjective criterionwas the appearanceof a significantF-value for the overall low cost factor when the numberof clusterswas increasedfrom three to four.

482

Journal Academyof Management

September

is Thevalidityof thesefindings interpretations enhanced the strong and by betweenthe compositionof the clustersand anecdotalinforrelationship A the mationobtained fromthe CEOsduring on-siteinterviews. briefdiscussion of a typicalfirm from each of the four clustersin Table 5 will convey this convergence. CEOof a firmin the overalllow cost clusterstressed The in the primacyof high productivity manufacturing operations.Production in incentivesand extensivereinvestment modernfacilitiesand equipment werementioned veryimportant.Uncertainties as regarding product-market the scope seemedto characterize responsesof one of the CEOswhose firm was in the stuck in the middle cluster. Perhapsthe relativelylow performance of his firm may be attributedto the lack of a well articulatedand consistentstrategybecauseof a lack of continuity(mentioned the CEO) by in the compositionof the top management team. The CEO of one of the firmsin the focus clusteremphasized importanceof successfullycomthe niche. Favorablerelationswith peting withina particular product-market channels(i.e., paintingcontractors) wereconsidered be to key distribution highly important.Lastly, the CEO of one of the firms in the differentiation clusterassertedthat his firm's successlay more in differentiating the firmthanthe product.He contended his firm'sreputation "superior that for serviceand quick reactionto customerneeds" enabledit to enjoy a high profit marginby charginga premiumfor his product. These remarksare clearlyconsistentwith Porter's generic strategies. The next step in the analysiswas to compareperformanceamong the four clustersor strategicgroups. Table5 indicatesthat differencesamong the four strategicgroupsregarding returnon total assets approachedstatistical significance(p = .069). This representsa noticeableimprovement over the resultsof the three clustersolution. The highest averageannual returnon total assets (25.5 percent)was achievedby clusterone with its emphasison overalllow cost; followed by clusterfour-differentiationwith 8.9 percent;clustertwo-stuck in the middle-with 5.4 percent;and clusterthree-focus-with 4.4 percent. The Scheffe procedureindicated that the differencesbetweenclusterone and clusterfour were significant For the other performance measure-sales growth-the overallF-ratio indicatedthat the groups were significantlydifferent from one another three-focus-was the leaderwith31 percent annualsales (p = .041).Cluster cluster growth;followedby clusterone-overall low cost-with 20 percent; four-differentiation-with 17 percent; and cluster two-stuck in the middle-with 8 percent.The Scheffeprocedure indicated thatthe difference betweenclustertwo and clusterthreewith regardto sales growthwas significant (p <.05). (p <.05).

Discussion Phase 1 of this study analyzedthe competitivemethods consideredto be most importantby the top managementof firms competingwithin a

1984

Dess and Davis

483

consistentcomsinglehighlyfragmented industry.Threesets of internally petitivemethods were identifiedthat conformedto Porter's (1980) three genericstrategies.The ability of questionnairedata to identify different "intended"strategiesin industrialsettingshas been supportedby others (Bourgeois,1980;Hambrick,1980). However,such an approach,with its emphasison "intended"strategies,has rarelybeen used to examinethe to presenceof strategicgroupsand their relationships organizational performance. The patternof competitivemethodsthat emergedfrom the field study closely resembledthat obtainedin Phase 2 from the panel of expertsand In providesadditionalsupportfor the authors'interpretations. addition to identifyingcompetitivemethodsof high importance,the managers and the panelalso identified itemsof lesserimportance eachgeneric for strategy. This convergence orientationsare indicative impliesthat these managerial of the two genericstrategies that Porter(1980)identifiedas differentiation and overalllow cost. However,the normative recommendations the panel of for the third of Porter's genericstrategies-focus-did not clearly coincide with the third orientationof the managers. The inabilityto identifyclearlythe thirdmanagerial orientation reflecas tive of a focus strategymay be due, at least in part, to what some regard as a limitationof the focus strategyas currently outlinedby Porter. Hofer (1982), for example, proposedthat when a firm deals in a marketcomposedof differentsegments,it may, for anygivenproduct,chooseto utilize eithera common appealacross all the varioussegmentswithin which the productcompetesor the firm may tailordifferentappealsfor specificcustomer groups. Under such a broaderview of a focus strategyit may not be possibleto prescribe singleset of competitivemethodsapplicablefor a all the potentialstrategycombinationswithinthis single genericstrategy. An importantimplicationis that equifinality(von Bertalanffy,1955)may characterize focus strategies,that is, thereis a broadrangeof differentbut consistentsets of competitive methodsavailableto firmsemployinternally ing a focus strategy. The clusteranalysisused in Phase 3 was based on the assumptionthat clustersof firms could be identifiedthat would correspondto a model of strategic groupsbasedon Porter'sthreegenericstrategies.Two of the clusin tersthatemerged thethreecluster solutiondidappear represent to Porter's differentationand focus strategies.However,the overalllow cost dimension did not appearas significantin determining compositionof orgathe nizationalstrategies.Further,the apparentemphasison more than one genericstrategy,reflectedin the clustercentroidscoresof clusterone, led the researchers investigatethe possibilitythat aggregationof dissimilar to firmsmay have occurredwithinthe threeclustersolution. The four cluster solutionservedto disaggregate largercluster(n = 12)previouslyidentithe fied as differentiationinto two separateclusters.Each clusterreflecteda unique strategicorientation-differentiation (n = 8) and overalllow cost (n = 4)-as well as differentlevels of organizationalperformance.

1984

Dess and Davis

484

An importantfindingof the study is the apparentlack of singularityin that the orientation characterizes highestperformance strategic group-cluster numberone. This grouphad the highestperformance the returnon on assets criterionand the second highestperformanceon sales growth. On the basis of its centroid scores, this cluster was identified as primarily orientedtoward an overall low cost strategy. However, this group also withthe highestcentroid GivenPorter's scoreon the focus strategy. emerged (1980)caution againstthe commitmentto multiplegenericstrategies,the exhibitedby the membersof this clustermay appearinhighperformance consistent.However,as Cyertand March(1963)suggest,highperformance may generateslack resourcesthat can be used to enable firms to expand theirpresentscopeof operations.Also, constraints posed by budgetlimitationsmayrequire thata firmlimitits emphasis only one marketsegment. to The presentcross-sectional research designinhibitsthe abilityto do more thaninfercausalrelationships. a largeextentthe viabilityof inferences To drawnfroma cross-sectional of measurement organizational relies strategies on the conceptof strategicmomentum(Miller& Friesen, 1980). The persistenceof strategiesover time may resultfrom decisionmakingprocesses as well as industry characteristics. viewof organizations conservative The as andresistant changeis promulgated the Carnegie to by group(Carter,1971; Cyert& March, 1963;March& Simon, 1958).Also, the presenceof intrabarriers mobility(Caves& Porter, 1977;Harrigan,1982;Oster, to industry 1982)may requiresuch an investmentof scarceresourcesas to make strategic change or exit costly if not prohibitive.Oster asserts: In the case of small organizations,the scarcityof resourcesand the high riskattendantwith strategic changemay help to explainwhy organizations continueto pursuemarginally profitablestrategies.The problemsmay be firms because of the lack compoundedin the presenceof privately-held of publiclyavailableinformationby whichto judge the strategiesand performanceof other firms. In summary,the research findingsare generallyconsistentwith Porter's contentionthat commitmentto at least one of the threegenericstrategies will resultin higherperformance than if the firm fails to developa generic strategy(i.e., becomesstuckin the middle).Additionally,the findingsthat the overalllow cost clusterhad the highestaveragereturnon total assets Andersupportpriorresearch Woo and Cooper(1981)and Hamermesh, by son, and Harris(1978)that rejectsthe notion that high marketshareis a for of Porter requirement the successful implementation a low cost strategy. also comments that "low cost maybe achievable withouthighshare"(1980, p. 44). However, it may be importantfor competitorsto identify with a genericstrategythat does not placethemin directcompetitionwith a large numberof firms evidencinga similarstrategicorientation.For example, a largenumberof firmsin the samplewereidentifiedas pursuinga differentiationstrategy,and this may have inhibitedthe ability of firms in this
To havesignificance the allocationof resources, strategy for a mustnecessarily involve somecommitment it is irreversable, leastfor a time. It is difficultyof movement that at that makesgroupstructures potentiallyimportant(1981, p. 377).

1984

Dess and Davis

485

as strategicgroup to realizeas high a level of performance those in other less populatedgroups. Lastly, the group of firms identifiedwith a focus strategymay illustratea potentialfor trade-offsbetweengrowthand profitability.The focus groupwas the highestperforming groupon salesgrowth but had the lowest level of returnon assets. The findings in this respect are consistentwith the recent empiricalfindings of Trostel and Nichols firmsmay choose to emphasize (1982),whichindicatedthat privately-held salesgrowthas a decisioncriterionat the possibleexpenseof profitability.
Limitationsof the Study

Limitationsof this study should be noted. First, the generalizability of the studyis limitedbecausethe firmsused in the study representonly one of Porter's (1980) five genericindustrialenvironments.The relativeimportanceof the competitivemethodsmay varyacrossas well as withininFor dustryenvironments. example,Porterposits that the emphasison cost controland serviceorientationmay be of greaterstrategicimportancefor firmscompetingwithina fragmented industrythan an emergentindustry. Second, the relativelysmall sampleof firms and executivesincludedin the field studymay lead to some instabilityin the factorloadingsobtained from the factoranalysis(Kim& Mueller,1978;Nunnally,1978).However, natureof the studyand resourceconstraints inherent giventhe exploratory in field research,this limitationis not considereda majorbarrierin interpretingthe results. administered the executiveswas used to deterto Third,the instrument mine "intended"firmstrategies may have servedto enactthe environand ment (Weick, 1979)to whichthe subjectresponded.Thus, executivesmay haveoverintellectualized whattheyactuallydid or attempted do (Dunon to can, 1979)in the formulationof their firm's strategy.Alternatively,the parsimonyof the instrumentmay have excludedimportantcompetitive methodsused by a firm. Thus, the instrumentmay not fully capturethe richnessor complexityof a firm's intendedstrategy. Fourth, the strategyrealizedby an organizationmay be differentfrom that intendedby the decisionmakers(Hambrick,1981a;Mintzberg,1978). An observeddiscrepancy betweenintentions realizedstrategy and mayarise from the inabilityof the firm to translateits intendedstrategiesinto actions becauseof unpredictable environmental change,a lack of appropriate or implementational capabilities, unrealistic expectations (Mintzberg, 1978). Singly,or in combination,thesefactorsmay resultin an emergentstrategy that is observedto be different from the intendedstrategy. Fifth, because all of the sample firms were privately-held,secondary sourcesfor confirmationof reporteddata wereunavailable.However,the authorsbelievethat the time, effort, and cooperation extended the manby of the samplefirms as well as the researchers' assurancesregarding agers confidentialityand sharingof the resultsof the study enhancedthe reliability of the informationprovided.

486

Journal Academyof Management

September

Implicationsfor FutureResearch

that the research couldestablish similarities existamongthevarious Future that have been proposedfor classifyingfirms. For example,to typologies of whatextentis Porter'sclassification firmsstuckin the middlecompatible with Milesand Snow's (1978)reactorsor Millerand Friesen's(1978)stagtypologycapable natingfirms?Thegoal wouldbe to developa parsimonious of classifyingthe arrayof strategicconfigurationsavailablewithin an industry.Suchsynthesiswould serveto answera recentcall for "developing and 'configurations"'(Miller& Mintzberg,1983, p. 57). The study of publicly-heldfirms would provide access to secondary sourcesof data, whichthen could be used to determineif firms classified on the basis of strategicchoice would be similarlyclassifiedon the basis of structural configurationsexhibitedby the firm. The use of maximally differentmethodscould providestrongevidencefor the validityof a particulartypology. Some evidencehas emergedthat strategiesappearto be enduringover time(Miller,1981).Systematic on research a longitudinal basiswouldreveal whetherperformance within and among strategicgroupsclassifiedon the basis of strategicorientationvariesover time. One may posit, for example, that as the industrylife cycleprogresses,the firmsstuck in the middle more may actuallybe ableto adaptto changesin the industryenvironment readilythan firms committedto a specific strategy. Finally,the role of strategicchoice as a method of classifyingstrategic groupsmay varyacrossas well as withinindustries,as suggestedby Hambrick (1983a), Snow and Hrebiniak(1980), and others.
or isolating composites which take the form of. . . 'gestalts,' 'archetypes,'

References
Ackoff, R. L. A conceptof corporate planning. New York: Wiley, 1970. L. in Bertalanffy, von. General systemstheory.Main Currents ModernThought,1955, 11(4),75-83. and Bourgeois,L. J. Performance consensus.Strategic Journal, 1980, 1, 227-248. Management E. E. Behavioral of the firmand top levelcorporate decisions.Administrative Carter, Science theory Quarterly,1971, 16, 413-429. An to Cattell,R. B. Factoranalysis: introduction essentials.(1) the purposeand underlying models, (2) the role of factor analysisin research.Biometrics,1965, 21, 190-215. and Journal Economic Caves,R. E. Industrial organization, corporate strategy structure. of Literature, 1980, 18, 64-92. to Caves, R. E., & Porter, M. E. Fromentrybarriers mobilitybarriers: decisionsand Conjectural contriveddeterrence new competition.Quarterly to Journalof Economics, 1977, 91, 421-434. A. and in Chandler, D. Strategy structure: industrial Chapters the historyof theAmerican enterprise. Mass.: The M.I.T. Press, 1962. Cambridge, and factorsassociated withcompanyperformance-Part2. JourChild,J. Managerial organizational nal of Management Studies, 1975, 12, 12-27. Cyert,R. M., &March,J. G. A behavioral theoryof thefirm. Englewood Cliffs, N.J.: Prentice-Hall, 1963.

1984

Dess and Davis

487

Dixon, W. J. (Ed.). BMDP: Biomedical computerprograms. Berkeley, Cal.: University of California Press, 1975. Duncan, R. B. Qualitative researchmethods in strategic management. In D. Schendel & C. Hofer (Eds.), Strategic management. Boston: Little, Brown, 1979, 424-447. Einhorn, H. J. Expert judgement: Some necessary conditions and an example. Journal of Applied Psychology, 1974, 59, 562-571. Hair, J. F., Anderson, R. E., Tatham, R. L., & Grablowsky, B. J. Multivariate data analysis. Tulsa, Okla.: Petroleum Publishing Co., 1979. Hall, W. K. Survival strategies in a hostile environment. Harvard Business Review, 1980, 58(5), 75-85. Hambrick, D. C. Operationalizing the concept of business-level strategy in research. Academy of Management Review, 1980, 5, 567-576. Hambrick, D. C. Environment, strategy and power within top management teams. Administrative Science Quarterly, 1981a, 26, 253-276. Hambrick, D. C. Strategic awareness within top management teams. Strategic Management Journal, 1981b, 2, 263-279. Hambrick, D. C. Some tests of the effectiveness and functional attributes of Miles and Snow's strategic types. Academy of Management Journal, 1983a, 26, 5-26. Hambrick, D. C. An empirical typology of mature industrial-productenvironments. Academy of Management Journal, 1983b, 26, 213-230. Hamermesh, R. G., Anderson, M. J., & Harris, J. E. Strategies for low market share businesses. Harvard Business Review, 1978, 56(3), 95-102. Harrigan,K. R. Exit decisions in matureindustries.Academy of ManagementJournal, 1982, 25,707-732. Harrigan, K. R. Research methodologies for contingency approaches to business strategy. Academy of Management Review, 1983, 8, 398-405. Hatten, K. J. Strategic models in the brewing industry. Unpublished Ph.D. dissertation, Purdue University, 1974. Hatten, K. J. Quantitative research methods in strategic management. In D. Schendel & C. Hofer (Eds.), Strategic management. Boston: Little, Brown, 1979, 448-467. Hatten, K. J., & Schendel, D. E. Heterogeneity within an industry: Firm conduct in the U.S. brewing industry, 1952-1971. Journal of Industrial Economics, 1977, 26, 97-112. Hays, W. L. Statistics. New York: Holt, Rinehart and Winston, 1963. Hofer, C. W. Toward a contingency theory of business strategy. Academy of Management Journal, 1975, 18, 784-810. Hofer, C. W. Personal correspondence, May 13, 1982. Huck, S., Cormier, W., & Bounds, W. Reading statistics and research. New York: Harper & Row, 1974. Hunt, M. S. Competition in the major home appliance industry, 1960-1970. Unpublished Ph.D. dissertation, Harvard University, 1972. Khandwalla, P. N. The techno-economic ecology of corporate strategy. Journal of Management Studies, 1976, 13, 62-75. Kim, J., & Mueller, C. W. Factor analysis: Statistical methods andpractical issues. Beverly Hills, Cal.: Sage University Press, 1978. Lawley, D. N., & Maxwell, A. E. Factor analysis as a statistical method. London: Butterworth, 1971. March, J. G., & Simon, H. A. Organizations. New York: Wiley, 1958. Miles, R. E., & Snow, C. C. Organizational strategy, structure andprocess. New York: McGraw-Hill, 1978. Miller, D. Toward a new contingency approach: The search for organization gestalts. Journal of Management Studies, 1981, 18, 1-26. Miller, D., & Friesen, P. H. Archetypes of strategyformulation. Management Science, 1978, 24, 921-933.

488

Journal Academyof Management

September

and revolutionin organizational Miller,D., & Friesen,P. H. Momentum adaptation.Academyof

Management Journal, 1980, 23, 591-614.

H. In Miller,D., & Mintzberg, The case for configuration. G. Morgan(Ed.), Beyondmethod. Beverly Hills, Cal.: Sage, 1983, 57-73. H. Mintzberg, Patternsin strategyformation.Management Science, 1978, 24, 934-948. A Newman,H. H. Strategic groupsand the structure-performance relationship: study with respect to the chemical Ph.D. dissertation,HarvardUniversity,1973. process industries.Unpublished 1978. Nunnally,J. C. Psychometric theory.2nd ed. New York: McGraw-Hill, S. Intraindustry structure the easeof strategic and Oster, change.Reviewof EconomicsandStatistics, 1982, 64, 376-383. for Administrative Parsons,T. Suggestions a sociologicalapproachto the theoryof organizations. ScienceQuarterly,1956, 1, 218-242. Patton, G. R. A simultaneous equationmodel of corporate strategy:Thecase of the U.S. brewing Ph.D. dissertation,PurdueUniversity,1976. industry.Unpublished in Porter,M. E. Retailer power, manufacturer strategyandperformance consumer goods industries. Ph.D. dissertation,HarvardUniversity,1973. Unpublished in Porter,M. E. Consumer behavior,retailer power,andmarketperformance consumer goods industries. Reviewof Economicsand Statistics,1974, 56, 419-436. withinindustries companies' and Reviewof Economicsand Porter,M. E. The structure performance. Statistics,1979, 61, 214-227. Porter, M. E. Competitive strategy.New York: Free Press, 1980. R. and Boston:Harvard Rumelt, P. Strategy, structure, economic performance. Press,1974. University Boston: Little, Brown, 1979. Schendel,D. E., & Hofer, C. W. Strategicmanagement. L. and Snow, C. C., & Hrebiniak, G. Strategy,distinctive competence, organizational performance. ScienceQuarterly,1980, 25, 317-336. Administrative and A Trostel,A. O., &Nichols,M. P. Privately-held publicly-held of companies: comparison strategic choices and management processes.Academyof Management Journal, 1982, 25, 47-62. U.S. industrial outlook, Washington,D.C.: U.S. Government PrintingOffice, 1980. Weick,K. Thesocial psychologyof organizing.2nd ed. Reading,Mass.: Addison-Wesley, 1979. R. E., & Hamermesh, G. Towarda model of businessunit performance: integrative R. White, An approach.Academyof Management Review, 1981, 6, 213-223. of low Woo, C. Y., &Cooper,A. C. Strategies effective sharebusinesses. Strategic Journal, Management 1981, 2, 301-318. Woodward,J. Industrialorganization.London:OxfordUniversityPress, 1965. GregoryG. Dess is Associate Professor of Managementat FloridaState University. PeterS. DavisisAssistant at Professorof Management the Universityof Oregon.