The case study method in family business research: Guidelines for qualitative scholarship

The

case

study

method

in

family

business

research:

Guidelines

for

qualitative

scholarship

Alfredo

De

Massis

*

,

Josip

Kotlar

CentreforFamilyBusiness,InstituteforEntrepreneurshipandEnterpriseDevelopment(IEED),LancasterUniversityManagementSchool,UK

1. Introduction

As family business researchers who frequently work with

qualitativedata,weareoftenaskedtogiveseminarsonhowto

conduct qualitative research and review qualitative papers.

Through these experiences, we have deemed that there is

widespread misunderstanding about the range of roles that

qualitative data can play in research on family business and

howqualitativemethodsshouldbeusedtogeneratehigh-quality

andpublishableresearchonfamilybusiness.

Amongqualitative methods,case studiesplay a particularly

important role, as they represent one of the most adopted

qualitativemethodsinorganisationalstudies(Eisenhardt,1989)

andhavebeenacknowledgedasanapproachtogeneratingand

testing theory that has provided the mainstream management

fieldwithground-breakinginsights(Burgelman,1983;Chandler,

1962; Penrose, 1960; Pettigrew, 1973). Moreover, case studies

have been the most used qualitative methodology in family

businessresearchtodate(DeMassis,Sharma,Chua,Chrisman,&

Kotlar,2012).Tofamilybusinessstudentsandscholarsunfamiliar

withthecasestudymethodology,thereisoftenamisconception

aboutwhatacasestudyisandhowitcaninformtheorybuilding

and professional practiceas a formof qualitative research. For

example, in a doctoral-level introductory qualitative research

methodscourse,wehavelistenedtoattendantsdescribetheirview

ofcasestudiesasasuitablemethodologythatshouldbeprimarily

used for descriptive purposes, mostly used in areas of nascent

theory or toexclusively study individuals or specifichistorical

events.

In our view, family business is a heterogeneous field that

encompasses multiple theoretical approaches and levels of

analysis,andwebelievethatdiversitycanandshouldbereflected

inthewayinwhichcasestudiesareused.Itisoftenobservedthat

thereisnoaccepted‘‘boilerplate’’forwritingcasestudymethods

anddeterminingquality(Amis&Silk,2008;Bansal&Corley,2012;

Coffey&Atkinson,1996;Maxwell,2012;Pratt,2009).Thislackof

understanding as to what makes ‘‘quality’’ research (

Easterby-Smith,Golden-Biddle,&Locke,2008;Gioia,Corley,&Hamilton, 2012;Graebner,Martin,&Roundy,2012)isunfortunatebecause

papers that build theory from case studies are frequently

considered the ‘‘most interesting’’ (Bartunek, Rynes, & Ireland,

2006;Eisenhardt&Graebner,2007)andarealsoamongthemost

impactfulpapersintheacademiccommunity(Eisenhardt,1989).

Consistentwiththeideathatcasestudiesrepresentamethodthat

is ideallysuited togeneratingmanageriallyrelevantknowledge

(Amabileetal.,2001;Leonard-Barton,1990),weviewcasestudies

asapowerfulmethodologythatcanbeusedinarigorous,creative

* Corresponding author at: Institute for Entrepreneurship and Enterprise Development,LancasterUniversityManagementSchool,LancasterLA14YX,UK. Tel.:+441524594057.

E-mailaddresses:a.demassis@lancaster.ac.uk(A.DeMassis),

and wide-ranging variety of ways to advance family business

research.When theapproach isappliedas wesuggestandin a

robustway, it becomes a valuable method for familybusiness

scholarstodescribecomplexphenomena,developnewtheoryor

refineandextendexistingtheories.Ourintendedcontributionin

this article is to guide novice or experienced family business

scholarswhoareinterestedindeepeningtheirknowledgeoncase

study research methods in appreciating the potential value of

qualitativemethodstoenrichtheirresearchandinidentifyingthe

key elements of designing and implementing qualitative case

studyresearchprojects.Ourgoalistohumblyattempttoprovide

someguidelines that are useful to familybusiness researchers

whendecidingwhetherandhowtousecasestudiesandhopefully

toreviewers and editorswhen evaluating case studywork. An

overviewofthecasestudydesignisprovidedalongwithgeneral

suggestionsforchoosingthetypeofcasestudyapproachthatis

mostappropriatefortheresearchquestion,definingtheunitof

analysis, selecting cases (sampling), collecting information,

analysinginformation, presenting resultsand ensuring validity

andreliabilityinresearchfindings.Tofacilitatetheapplicationof

theseprinciples,wedrawonourownpublishedqualitativestudies

andthecomplementaryinsightsfromothercasestudyworkinthe

fieldtoproviderichandclearexamplesofthetypesofcasestudy

designs, research questions, sources of information, sampling

techniques, study propositions, and data display frameworks.

Specifically,weofferconcreteexamplesofarticlesandexemplar

quoteswithinthesestudiesthatareparticularlyillustrativeofthe

ideasexpressedinthisarticle.

In this way,wehopetohelpmove the familybusinessfield

beyonditscurrentstatebyestablishingtherelevanceandusefulness

ofthisparticularqualitativeresearchmethodforgeneratingaricher

anddeeperunderstandingoffamilybusinesses.Wealsoidentify

someareasinwhichweseespecialopportunitiesfortheuseofcase

studymethods.Arecentannotatedbibliographyofthe215most

citedfamilybusinessstudiesfrom1996to2010hasshownthatonly

a scantminority ofarticles (18) relies on case studies or other

qualitativemethodologies(DeMassisetal.,2012).Wehopethatthis

articlewillencourageanincreasingnumberofscholarstoengagein

thecasestudymethodinhigh-qualityfamilybusinessresearch.

2. Choosingthecasestudydesign

A casestudyis a particularstrategy forqualitativeempirical

researchthatallowsanin-depthinvestigationofacontemporary

phenomenon within its real-life context (Pettigrew, 1973; Stake,

1995; Yin, 2003). Case studies are particularly relevant to

organisation and management studies because they promote

‘‘understanding the dynamics present within single settings’’

(Eisenhardt,1989,p.533)byusingavarietyoflenses,whichallows

for multiple facets of the phenomenon to be revealed and

understood. This feature of case studies could be particularly

relevanttofamilybusinessresearchbecausefamilyfirmsexistatthe

intersectionoftwosystems–thefamilyandthebusiness(Tagiuri&

Davis,1992)–that interactinproducing idiosyncratic

organisa-tionaloutcomes.Tofullyunderstandtheorganisationalphenomena

associatedwithfamilyinvolvementinand/orinfluenceonbusiness

aswellastheirantecedentsandconsequences,researchersneedto

combine multiple perspectives and navigate multiple levels of

analysis.Inthisrespect,thecasestudydesignappearstobea

well-suitedmethodology,asit(i)copeswiththesituationinwhichthere

aremanifoldvariablesofinterestthatareembeddedinthecontext

ofinvestigationand(ii)reliesonmultiplesourcesofevidence,with

dataneedingtoconvergeinatriangulationfashion(Yin,2003).

Researchers who decide to usea case study approach must

choosewhattypeofcasestudywillbeconducted(Yin,2003).The

selectionofaspecifictypeofcasestudydesignwillbeguidedby

theoverallstudyobjective.Anexploratorycasestudyshouldbe

usedwhen theaimistounderstand how a phenomenontakes

place.Typicalexamplesofresearchquestionsthatareparticularly

suitable to be answered with this type of case study are, for

instance,‘‘How dothe individualgoalsof organisational members

influencetheorganisationalgoalspursuedbyfamilyfirms?’’or‘‘How

istheproductinnovationprocessmanagedandorganisedin

small-sizedfirms?’’Exploratorycase studiesaretypically employedto

gainanunderstandingofhoworganisationaldynamicsorsocial

processeswork:

In this study, we attemptto broaden and refine the extant

theoryintheareaoforganizationalgoalsandgoalformulation

processesinfamilyfirmsbyaddressingthefollowingresearch

question: How do the individual goals of organizational

membersinfluencetheorganizationalgoalspursuedbyfamily

firms?[...].Forthispurpose,weconductedatheory-building

qualitativestudytobetterunderstandtheunexplored

dynam-icsofgoalsettinginfamilyfirms.(Kotlar&DeMassis,2013)

More research is needed to uncover how the role of early

adopters changes depending on other innovation-specific

factors.Thisarticleextendspriorresearchbyexamininghow

thetypeofinfluenceplayedbyearlyadoptersisaffectedbythe

platform vs. non-platform nature of innovations. (Frattini,

Bianchi,DeMassis,&Sikimic,2013)

Anexplanatorycasestudyshouldbeusedwhentheaimisto

understandwhyaphenomenontakesplace.Forexample,thistype

ofcasestudyisparticularlysuitabletoanswerresearchquestions

suchas, ‘‘Whydo some successions leadto better outcomesthan

others?’’or‘‘Whyarefamilybusinessesmorelikelytoengagein

inter-firmcollaborations?’’Mostoften,theexplanatorynatureofacase

studyiscombinedwithitsexploratoryaim:

This paper contributes to opening up the ‘‘black box’’ on

innovationinfamilyfirmsbyinvestigatingwhatcharacterizes

theirproductinnovationprocess[...]Ourobjectiveistogain

theoretical clarifications as to how and why the product

innovationprocessinfamilyfirmsisdifferentfromnon-family

firms.(DeMassis,Frattini,Pizzurno,&Cassia,2013)

Finally,adescriptivecasestudyshouldbeadoptedwhentheaim

of the researchis toconvince someone that a phenomenonis

relevant. For example, this type of case study can be used to

provide rich evidence-supporting statements such as, ‘‘Family

businessesare verylikely tofail duringthe successionprocess’’or

‘‘Incumbentfamily firmsare subjecttoorganisationalinertiawhen

they develop radical innovations’’. For example, Dyck, Mauws,

Starke,andMischke(2002)offeradetailedportraitofsequence,

timing, technique and communication in executive succession

throughalongitudinalcasestudyofafailedexecutivesuccession

inasmall,family-ownedfirm.

Insummary,casestudyresearchisparticularlyappropriateto

answerhowandwhyquestionsortodescribeaphenomenonandthe

real-life context in which it occurred. This approach openly

contrasts, but hopefully complements, statistical empirical

re-search approaches that are primarily used for confirmatory

objectiveswhen researchersalready know how a phenomenon

happens and have a robust idea of why it happens (i.e.,

hypothetical deduction is possible) and can quantitatively

measure all the variables of interest. In fact, such a research

strategyisusuallyappliedtoanswerwho,what,where,howmany,

andhowmuchquestions.

In addition tochoosing a specific type of case studydesign,

differentphilosophicaltraditionscanbeembracedinconductinga

discussionofthephilosophicalassumptionsofqualitativeresearch,

excepttoofferthenotethatYin(2003)primarilybaseshisapproach

tocasestudyonpositivism(Clegg,2008),withitsrootinempiricist

philosophyandadeductiveresearchapproach.Conversely,other

scholars base their approach to case study on interpretivism

(Bryman& Bell,2007; Denzin,2001; Miles& Huberman,1994; Schwandt,2000;Stake,2005),whichis groundedin ainductive

researchapproach.1_{Thisalternativecasestudytraditionassumes}

thatevidenceisformedandshapedinthemindoftheresearcher

(Stake, 1995, 2005). Thismeans that knowledge is constructed

ratherthanrevealed,andtheresearcherworkstowards

reconstruct-ingeventsandbelievesthathumansareactiveintheconstructionof

knowledgeratherthanbeingpassiverecipientsofknowledge.Thus,

knowledgeisconstructedthroughthecreationofconcepts,models,

andschemestomakesenseofhumanexperienceandarecontinually

interpretedand modifiedby the researcher. Thiscontrastswith

positivism,whichassumesthatthe researcher‘‘finds’’ orsimply

‘‘observes’’findingsandobtainsknowledgeinanobjectiveway.Put

differently, interpretative case studies seek to ‘‘understand the

human experience’’ (Stake, 1995, p. 38).The inductiveresearch

approachissometimesacknowledgedasbeingparticularlysuitable

tobuildgroundedtheoryfromcasestudydata(Eisenhardt,1989;

Glaser&Strauss,1967;Strauss&Corbin,1998).

BothFletcher(2000)andNordqvist,Hall,andMelin(2009)have

outlinedtheimportanceofinterpretivismtoadvancingthefieldof

family business research. However, although the philosophical

traditionisveryseldomexplicitlystated,thevastmajorityofcase

study articles in the family business field are based on the

positivistictradition,withonlyfewexceptions(e.g.,Hall,Melin,&

Nordqvist,2001;Hall&Nordqvist,2008;Nordqvist&Melin,2010; Parada,Nordqvist,&Gimeno,2010).

3. Definingtheunitofanalysis

Oncethecasestudymethodhasbeenchosen,averyimportant

stepinitsapplicationischoosingtheunitofanalysis(case).The

unitofanalysiscanbedefinedas‘‘a phenomenonof somesort

occurringinaboundedcontext’’(Miles&Huberman,1994,p.25).

Familybusinessscholarsshouldstartbyaskingthemselveswhat

theirunitofanalysisis;dotheywantto‘‘analyse’’theindividual?A

groupofindividuals?Aprocess?Aprogramme?Aprojectwithin

thefirm?Differencesbetweenfirms?Answeringthesequestions

canbeanappropriatestrategytooutlinetheunitofanalysis,anda

case study needs to be explicit about the underlying unit of

analysis.Forexample,theresearchobjectivemaybetounderstand

‘‘Howandwhytheanatomyoftheproductinnovationprocessdiffers

between[small]familyandnon-familyfirms’’(DeMassis,Frattini,

Pizzurno,etal.,2013).Inthisexample,theunitofanalysisisthe

productinnovationprocessofsmall-sizedfamilyandnon-family

firms.However,itmaybethatresearchersarelessinterestedinthe

process of productinnovation and more interested in focusing

specificallyontheexperiencesofR&Dandinnovationmanagersof

small firms. In this case, the research objective may be to

understand ‘‘How R&D and innovation managers of small-sized

firmsdescribetheirexperiencesinmanagingandorganisingproduct

innovation,andhowthesedescriptionsdifferinfamilyversus

non-familyfirms’’.Inthefirstexample,theunitofanalysiswouldbethe

managementandorganisationofproductinnovationinthisgroup

offirms,anditwouldbeaprocesstobeanalysed,butinthesecond

example, the unit of analysis would be the individuals or the

experiencesofR&Dandinnovationmanagersofsmall-sizedfirms.

In somecases,researchersmaybeinterestedin phenomena

thatoccuratmultiplelevels.Inthesecircumstances,multipleunits

of analysis cancoexist in thesame study,and theresearcher’s

abilitytonavigatethedifferentunitsofanalysisbecomesessential.

Forexample,inourstudyoforganisationalgoal-settingprocesses

of family firms, we initially noted that the formation of

organisationalgoalsinvolvesindividuals,groups,andthefirmas

awhole.Thus,westartedwithanindividuallevelanalysis,wethen

shifted totheanalysis ofsocial interactions between groups of

individuals (dyads), and finally, we shifted to the analysis of

organisational-levelphenomena,i.e.,organisationalgoaldiversity

andthecollectivecommitmenttofamily-centredgoals:

Goalsettinginvolvesindividuals,groupsandfirms;nevertheless,

therelativelyscarcestudiesavailablearelimitedtothefirmlevel,

typicallyrelyingonsingleinformants[...]andoverlookingthe

individualorganizationalmember’sroleintheprocess.[...]In

describingthisprocess,webeginwithananalysisofthegoals

pursuedbyorganizationalmembersinfamilyfirms,whichhelps

informon the individual-levelantecedents of goalsetting in

family firmsand, ultimately, on howgoaldiversity and goal

conflicts arise therein. We continue by analysing the social

interactionprocessesthroughwhichgoaldiversityismanagedas

wellastheirimplicationsfor[organizationalgoaldiversity]and

organizationalmembercommitmenttofamily-centeredgoals.

(Kotlar&DeMassis,2013)

4. Selectingthecases(sampling)

With the case study method, a very critical decision is the

selection ofcases.Contrarytothesamplingapproachestypically

adopted in quantitative research (e.g., random, stratified, or

statisticalsampling),casesareselectedbecausetheyareparticularly

suitableforilluminatingaphenomenonandforextending

relation-shipsandlogicamongvariables,i.e.,theoreticalsampling(Eisenhardt,

1989;Graebner&Eisenhardt,2004).

Afirst,criticalchoiceforresearchersiswhetheritisprudentto

conduct a single-casestudy orif a betterunderstanding ofthe

phenomenonwillbegainedthroughconductinga multiple-case

study.Thetheoreticalsamplingofsinglecasesisstraightforward.

Singlecasesarechosenbecausetheyareunusuallyrevelatoryor

extremeexemplarsorbecausetheyofferopportunitiesforunusual

researchaccess.Asanillustration,thisishowKarra,Tracey,and

Phillips(2006)reportthereasonsbehindthechoiceoftheirsingle case:

WechosetostudyNeroliforthreereasons.First,thecasehas

‘‘rareorunique’’qualitiesthatmakeit alogicalcandidatefor

‘‘theoreticalsampling’’[...].Preliminaryresearchrevealedthat

thefirmhadgrownrapidlyoverarelativelyshortperiodoftime

and relied upon a high level of family involvement. The

organization of the firm remained family-based, and most

employees were either relatives or shared the same ethnic

background as the owner. In addition, the entrepreneur’s

motivation forfoundingthefirm wasthebetterment ofthe

family,andthedynamicsofthefamilywerethereforecentralto

thefirm.Second,Karovprovidedaveryhighlevelofaccesstothe

firm. We were able to interview him and other important

membersofthefirmonmultipleoccasionsduringtheperiodof

the study, and he provided us with extensive archival data

relating to the history of the firm. In addition, one of the

coauthorswasabletotravelwithhimtoattendkeymeetings

withdistributorsandretailersinRussiaandEasternEuropeand

toattendtradeshowsandvisitimportantsuppliersinItaly.She

1_{Althoughitisnottheaimofthisarticletoprovideadetaileddiscussionofthe}

philosophicaltraditionsinqualitativeresearch,itisimportanttonotethatbeyond positivism and interpretivism, a range of philosophical traditions such as hermeneuticsand socialconstructionism (Schwandt,2000)orcritical realism (Welch,Piekkari,Plakoyiannaki,&Paavilainen-tymaki,2011)informqualitative research.

wasalsoabletointerviewallofthefamilymembersandother

keyindividuals,includingmanufacturingpartnersand

employ-ees.Third,thefirmwasonlyslightlymorethanadecadeoldat

thetimeofthestudy,andthefounderwasstilltheCEOofthe

company. This was significant because it increased the

likelihood that thedetailsof the founding of the firm andits

earlydevelopmentremainedfreshinthemindsofthefounderand

other interviewees. We thereforeconsider Neroli a ‘‘strategic

research site’’ [...] for studying altruism in familybusiness.

(Karraetal.,2006,p.865)

As explained in detail by the authors, the case was chosen

becausetheauthorshaddeepaccesstothefirmandbecausethecase

offeredadistinctiveandextraordinarysettinginwhichtoobserve

thephenomenonunderinvestigation(i.e.,theinfluencesoffamily,

kinship,andethnicityonaltruismandagencyinfamilyfirms).In

sum,asingleholisticcasestudytypicallyexploitsopportunitiesto

explorea significantphenomenon underrare,uniqueor extreme

circumstances.ThestudiesbyDeAngeloandDeAngelo(2000),Dyck

etal. (2002)and Salvato, Chirico,and Sharma (2010)are other

examplesofsingle-casestudiesinthefamilybusinessfield.

However,althoughsinglecasestudiescanrichlyand

persua-sivelydescribea phenomenon(Siggelkow, 2007),this approach

also suffers from serious limitations, especially related to the

externalvalidity of results.2 _{In this regard,multiple-case studies}

typicallyprovideastrongerbasefortheorybuildingorexplanation.

Multiple cases enable comparisons that clarify whether an

emergent finding is simply idiosyncratic to a single case or

consistently replicated by several cases. Thus, a multiple-case

studywillallowthefamilybusinessresearchertoanalysewithin

eachsettingandacrosssettings.Overall,thetheorycreatedfrom

multiplecasesisconsideredmorerobustbecausethearguments

aremoredeeplygroundedinvariedempiricalevidence.However,

thetheoreticalsamplingofamultiple-casestudyismoredifficult

and complex. Indeed, the sample cases must be selected for

theoretical reasons, i.e., because they allow the prediction of

similarresults(literalreplication) orcontrastingresultsbut for

predictablereasons(theoreticalreplication)orbecausetheyallow

the elimination of alternative explanations (Yin, 2003). A

particularly important theoretical sampling approach that is

widelyusedinthefamilybusinessfieldis‘‘polartypes’’sampling

(Eisenhardt & Graebner, 2007) in which a researcher samples

extremecases(e.g.,alongdimensionsofstructure,behaviour,or

performance)inordertomoreeasilyobservecontrastingpatterns

inthedata.Suchanapproachcanpositivelyimpressreviewersand

prospectivereadersbecausetheresultingtheoryislikelytobevery

consistentlysupportedbytheempiricalevidenceandthusallows

‘‘veryclearpatternrecognitionofthecentralconstructs,relationships,

andlogicofthefocalphenomenon’’(Eisenhardt&Graebner,2007,p.

27).Ourtwostudiesofproductinnovationinfamilyversus

non-familyfirmsand goal-setting processesin familyfirmsprovide

examplesofpolarsampling:

Ourstudyinvolved10Italianfirms,fiveofwhicharefamilyand

five non-family businesses, according to a ‘‘polar type’’

samplinglogic [...]We decidedtofocuson firmswhichare

well respected for their prowess and successin the area of

productinnovation.Havingselectedcompaniesthatconsider

productinnovationacriticaldeterminantoftheircompetitive

advantage, we could not misinterpret differences in the

anatomyoftheproductinnovationprocessdueto

heterogene-ityinthestrategicrelevanceassignedtothisactivity[...]We

adopted this convenience, theoretical sampling approach

becauseweneededtocreateanexperimentalempiricalbasis

thatallowedustostudythephenomenonunderparticularly

insightfulandilluminatingcircumstances.(DeMassis,Frattini,

Pizzurno,etal.,2013)

Werepresentthesegmentationofthefamilyfirm’s

organiza-tional members by examining Chief Executive Officers,

professional (non-family) top executives, young generation

familymembers,familyCEOspousesandoldgenerationfamily

members.Thissetoforganizationalmembersmaximizedthe

differences along four dimensions that were considered

particularlyrelevanttoourstudy.First,weselectedindividuals

thatownequitysharesofthefirmandothersthatdonot,since

ownership is likely toaffect theincentives and priorities of

organizational members [...]. Second, we selected family

membersas wellasnon-familymembersbecausethe latter

couldbeexpectedto followa self-serving attitude, whereas

familymembersarepossiblymoreorientedtowardsthefamily

firm’s collectivistic goals [...]. Third, we selected family

memberswhoareactivelyinvolvedinthebusinessandthose

whoarenot,sincetheymaydifferintermsoftheirpowerinthe

organizationand howtheyperceive thefamilyand business

priorities[...].Fourth,weselectedfamilymembersthatbelong

tothecurrentCEO’sgeneration,totheyoungergenerationand

totheoldergeneration,sinceresearchhaspointedtorelevant

differencesbetweenincumbentsanddescendants[...].Given

these differences among the types of family firms and

organizationalmembers,wefeltthatthiscombinationwould

provide enough distinct windows through which to view

organizationalmember individualgoals.Taken together,this

sample and context provide an excellent opportunity to

examinegoalsettinginfamilyfirms.(Kotlar&DeMassis,2013)

Asreportedinthequotestakenfromthetwoexemplarsabove,

theselectedcasesshouldofferenoughdistinct windowsthrough

which toobserve aninvestigatedphenomenonin auniqueand

extraordinary way. In a case study, researchers should not

underestimate the importance of explaining the sampling

ap-proachadoptedfortheselectionofcases.We encouragefamily

business scholarsto providea clearrationale for the case study

selection and ample details on the case study context (e.g.,

competitivedynamics,financialdata,thebusinesslifecycle,and

thegenerationoffamilycontrol)toallowthereadertoappreciate

the researchers’ sampling choices. The studies by Howorth,

Westhead,andWright(2004)andJohannissonandHuse(2000)

areotherexamplesofmultiple-casestudiesinthefamilybusiness

fieldwheretheauthorsexplicitlystateandexplainthetheoretical

samplingapproach.Inaddition,robustcasestudyarticlesnotonly

convey the rationale for case selection to the reader but also

actuallyseektoconvincethereaderofitsappropriatenessforagiven

research question. Continuing with the example of product

innovation in family versus non-family firms, this is how we

justifytothereaderwhythecharacteristicsoftheselectedfirmsin

termsofimportanceattachedtoproductinnovationandsmallsize

areparticularlysuitablefortheirresearchobjective:

Wedecidedtofocusonfirmswhicharewellrespectedfortheir

prowessandsuccessintheareaofproductinnovation.Having

selectedcompaniesthatconsiderproductinnovationacritical

determinant of their competitive advantage, we could not

misinterpret differences in the anatomy of the product

innovation process due to heterogeneity in the strategic

relevance assigned to this activity. Finally, we decided to

include in oursample onlysmall companies[...]. First, this

choicewassuggested bythefact that productinnovationis

2_{Itshouldbenotedtheexistenceofatrade-offbetweenresearchingasinglecase}

andcomparingmultiplecases.Althoughinsinglecasestheexternalvalidityof resultsisthreatenedanditmayalsobedifficulttoconveythespecificityofthecase tootherresearchers,singlecases,ifcomparedtomultiplecases,generallyallowfor ‘‘ticker’’,thatis,moredetaileddescriptions andinvolvemorein-depth(often longitudinal)understandingofthespecificsofthebroadercontextofthecase.

consideredoneofthemostcriticaldeterminantsofsustained

competitiveadvantageforthiscategoryoffirms[...].Second,

innovation in small firms has several peculiarities which

differentiate it from large companies [...]. What is more,

scholarlyresearchhasthusfarfocusedinparticularonlarge

firmsandonlytoalesserextentonsmallfirms[...].Byfocusing

on small companies only, we also reduced the risk of

unobserved heterogeneity due to differences in the size of

thefamilyand non-familyfirms in oursample. (De Massis,

Frattini,Pizzurnoet al.,2013)

Insteadofconductingandanalysingmultiplecasestudiesof

differentorganisations,researchersmight alsowanttoconduct

differentcasestudieswithinoneorganisation(anestedapproach,

e.g., Yin, 2003). This occurs, for example, in project-level case

studieswheretheunitofanalysisisaspecifictypeofprojectthatis

conductedwithinanexistingorganisation.

5. Collectinginformation

Veryoften,researchersassumethatcasestudiesshouldonly

relyonqualitativedata.Thisisclearlyamisconception,asthereare

no empirical research methods that only use qualitative or

quantitativedata. Datain thecase study methodare collected

bymultiplemeansthatmayconsistofpotentialqualitativedata

sources, such as interviews, direct observations, including

ethnographical and anthropological data collection techniques,

documentation,andhistoricalrecords,aswellasquantitativedata

sources(e.g.,surveys).Casestudydataoftenprovideadvantagesin

beingabletointegratebothobjectiveand perceptualdata.This

integrationisparticularlyimportantin familybusinessresearch

where family relationships and business issues are typically

inseparable in decision-making; thus, the use of a variety of

different data, including the combination of subjective or

interpretative and more objective factual information, can add

much to our understanding of organisational processes and

outcomes. The use of multiple data sources enhances data

credibility(Patton, 1990). Eachdata sourceis one piece of the

‘‘puzzle’’, with each piece contributing to the researcher’s

understandingofthewholephenomenon.Thisconvergenceadds

strengthtothefindings,asthevariouscomponentsofdataare

interweavedtogethertopromoteawide-rangingunderstandingof

thecase.Thetwoexamplesfromourstudiesalreadydiscussedin

theprevious sectionofferan illustrationof theuseof multiple

sourcesofdatainacasestudy:

Weundertookinterviewsandgatherednon-participant

obser-vationsandarchivaldocumentsfrom19familyfirms,thereafter

analysing these qualitative data to disclose the unexplored

dynamicsofgoalsettinginfamilyfirms.(Kotlar&DeMassis,

2013)

As regardsdata collection, we gathered information mainly

throughdirectinterviews,undertakenbetweenOctober2009

and December 2010. Specifically, the following steps were

taken:Attheoutsetofeachcase,arelationshipwasestablished

withaseniormanagerfromtheselectedfirm.Thispersonwas

briefed abouttheresearchprojectthrougha writtenproject

summaryand atelephone meeting.During thismeeting,we

asked the respondent to introduce us to the entire

top-managementteamandthestaffinchargeofproduct

innova-tion.Thenwepersonallyinterviewedatleasttwoinformants

foreachcompany.Weundertook twosemi-structured

inter-viewsforeachrespondent(eachlastingonaverageoneanda

half hours), for a total of more than 35hours of personal

interviews. Direct interviews followed a semi-structured

replicable guide that comprised a set of open questions for

each area ofthe productinnovationprocessincluded inthe

theoreticalframework.Secondaryinformationwascollectedin

theform ofcompanyreports and project documentation.In

particular,wegatheredandanalyzedalltheavailablecompany

documents,catalogs,familyinformationandreportsofproduct

innovation projects. This informed the researchers with

backgroundinformationabouttheselectedfirms,thetypeof

productinnovation theyundertake and theapproachesthey

usetoadministerproductinnovationactivities.Aboveall,these

secondaryinformationsourceswereintegrated,ina

triangula-tionprocess,withdatadrawn fromthedirect interviews, in

ordertoavoidposthocrationalizationandtoensureconstruct

validity.(DeMassis,Frattini,Pizzurno,etal.,2013)

Anotherremarkableexampleofuseofmultipledatasourcesin

a single-casestudyin thefamilybusinessfieldis thearticleby

Astrachan(1988),wheretheauthorintegratesdatacollectedfrom

a widerangeofsources(interviews,historical records,financial

data,productiondata,observation,andmediaaccounts)toprovide

a thoroughexplorationofhowthecultureofaparentcompany

affects the cultureand performance ofthe familyfirm whena

familyfirmisacquired.ThestudiesbySteier(2001),Steierand

Miller (2010), and Lambrecht (2005) are other illustrative

examples of family business case studies combining multiple

sourcesofdata.

Interviews areoften theprimary datasourcein case studies

(Eisenhardt&Graebner,2007).Theyareatargeted,insightfuland

highly efficientmeansby which tocollect rich,empirical data,

especiallywhenthephenomenonofinterestishighlyepisodicand

uncommon. However, interviews are often characterised by

several weaknesses, such as bias due to poorly articulated

questions, responseor personalinterpretation bias.In addition,

other sources of error may contaminate interviewee reports,

especiallyretrospectiveaccounts(Golden,1992;Huber&Power,

1985; Schwenk, 1985). Interviewee reports may suffer from

informants’ memory failure or inaccurate recall of past events

(Golden,1992)aswellasfrommemorydistortion(Nutt,1986).The

lattercanresultfromhindsight bias,attributionalbias,

subcon-sciousattemptstomaintainself-esteem,orimpression

manage-ment (Huber & Power, 1985; Salancik & Meindl, 1984). The

challengepresentedbyinterviewdataisbestmitigatedbydata

collectionapproachesthatlimitbias.Akeyapproachisusingmany

different and well-informed interviewees who view the focal

phenomenon from different perspectives. These interviewees

canincludefamilyandnon-familymembers,organisationalactors

from different hierarchical levels, generations, organisational

functions,andgeographies,aswellasactorsfromotherrelevant

organisations and outside observers suchas management

con-sultants.Itisunlikelythatthesevariedintervieweeswillengagein

convergent retrospective sensemaking and/or impression

man-agement.Table1illustratesthedemographyofintervieweesinour

two studieson product innovationin familyversusnon-family

firmsand on therole ofearlyadoptersin thediffusionofnew

productsinplatformversusnon-platforminnovations.

Anotherkeyapproachtomitigatingbiasistocombineinterviews

withdirectobservations.Forexample,inthestudyongoal-setting

processes in family firms (Kotlar & De Massis, 2013), direct

interviews were combined with non-participant observations.

Specifically,informantswerefollowedduringfamilyandbusiness

meetings(e.g.,meetingsoftheboardofdirectors,familymeetings,

andcasualmeetingssuchasfamilydinners),formulatinggeneral

observationsofhowgoalsformpartofeverydayfamilyandbusiness

life.Interviewsareappropriatewhenfamilybusinessscholarsseek

to gain a broad understanding of what interviewees consider

importantorwhentheresearchtopicfocusesondeeplyheldvalues

potentiallyunexpected, thoughtfulresponses frominterviewees.

For example, we relied extensively on interviews with family

businessCEOsinourmultiple-casestudyonsocialcapitalinfamily

firms(DeMassis,Kotlar,&Frattini,2013)becauseouraimwasto

understand CEO perceptions of competitive advantages and

disadvantagesderivingfromsocialcapitalresources.

Direct observations require that researchers spend time

observingand experiencingorganisationallifein a firm orin a

sampleoffirms.Thus,theseobservationsareverytime-consuming

andcostly,buttheyallowresearcherstoobtain‘‘richinsightsinto

thehuman,socialandorganisationalaspectsofbusiness

organiza-tions’’(Myers,2013,p.92).Thisdatasourceisparticularlysuitedto

studyingaspectsof organisationalculturebecauseit is onlyby

experiencingorganisational life that theunderlyingvalues and

philosophiescommonlyheldbyorganisationalmemberscaneven

begintobeunderstood.Researchersmustspendprotractedperiods

of time immersed in the organisation (e.g., attending business

meetings, family councils or boards of directors meetings or

observingR&Dworkorcorporate classrooms)andare typically

overwhelmedbyverylargevolumesofdata.Forexample,inour

studyongoal-settingprocessesinfamilyfirms(Kotlar&DeMassis,

2013),weobservedatleastthreemeetingsineachofthe19family

firms.These observationsprovided a large volumeof datathat

allowed the confirmation of informant descriptions of their

organisationalroles,an assessmentthe general familybusiness

environment, and observation of the processes through which

organisationalgoalsaresetascloselyaspossible.Bothparticipant

andnon-participantobservationscanbeusedinacasestudy.The

former occurs when the observers take a role and involve

themselves in the group or process being studied. The main

problemassociatedwithparticipantobservationsistheriskthat

theresearcherbecomestoocloselyinvolvedandlosesdetachment,

whichdeterioratestheimpartialreportingofcollectedevidence.

Ontheotherhand,anon-participantobserverisdetachedfromthe

grouporprocessbeinginvestigated,sothepotentialforbiasis

mitigated.Inadditiontoobserving,itisimportantthatresearchers

makefieldnotestorecordtheirobservations.Thisdatasourcecan

beparticularlyusefultoanswerresearchquestionsaboutfamily

firmsthatrelatetotheirorganisationalculture,suchas‘‘Howdo

family valuesand the family tradition play arole in shaping the

innovationprocessoffamilyfirms?’’or‘‘Howdofamilyvaluesand

goalsaretransferredovergenerations?’’

Everyfirmrecordsasubstantialamountofinformationabout

itself,andthereisoftenalargeamountofinformationaboutan

Table1

Examplesofthedemographyofintervieweesincasestudies.

Firm Interviewees Agerange

A:

FirmA CEO(father) 61–80

ChiefCommercialOfficer(son) 25–40

ChiefTechnicalOfficer(son) 25–40

FirmB CEO(father) 61–80

HeadofR&D(brother-in-law) 41–60

ChiefCommercialOfficer(son) 25–40

FirmC CEO(son) 41–60

ChiefTechnicalOfficer(father) 61–80

HRExecutiveOfficer(non-familymanager) 25–40

FirmD CEO(father) 61–80

ChiefTechnicalOfficer(son) 41–60

FirmE CEO(father) 61–80

HeadofR&D(father’sbrother) 41–60

FirmF CEO 41–60

DirectorofTechnicalDepartment 41–60

FirmG CEO 61–80

HeadofR&D 41–60

ChiefCommercialOfficer 25–40

FirmH CEO 61–80

ChiefTechnicalOfficer 41–60

FirmI CEO 41–60 HeadofR&D 25–40 ChiefDesigner 25–40 FirmL CEO 61–80 DirectorofInnovation 41–60 HRExecutiveOfficer 41–60

Company Keyinformantsintheinnovatingfirms Keyinformantsinearlyadopters Keyinformantsinlateadopters B:

CompanyA  Marketingmanager  Salesagent

Facilitymanagerofalargebank  Facilitymanagerofanindustrialfirm CompanyB  Productmanager

 Generalmanager

 Scientificdirectorofalargepublic researchinstitution

Professorwithresponsibilityformanaging auniversitylaboratory

CompanyC  Chairmanandchiefmarketing manager

 Generalmanager

 Generalmanagerofoneofthemost well-respectedandhigh-qualityItalian woolproducers

 Directorofoperationsofatextilemanufacturer

CompanyD  Productdevelopmentmanager  Productmanager

 Headoftheengineeringdepartmentofa medium-sizedbiscuitproducer

DirectorofoperationsofanItalianpasta manufacturer

CompanyE  Marketingdeveloper  Salesagent

Generalmanagerofawell-respectedwine producerfromtheNorthofItaly

 Ownerofawinery CompanyF  Businessteammanager

 Marketingmanager

 Headofthepurchasingdepartmentofa largeItalianbank

Itwasnotpossibletointerviewanylateadopter CompanyG  Generalmanager

 Marketingmanager

 Itwasnotpossibletointerviewanyearly adopter

 Technicaldirectorofalargemanufacturerof plasticcomponents

CompanyH  Productmanager  Headofclinicengineeringofalargepublic Italianhospital

 Generalmanagerofaprivatehospital A:AdaptedfromDeMassis,Frattini,Pizzurno,etal.(2013);B:AdaptedfromFrattini,Bianchi,DeMassis,andSikimic(2013).

organisationthatisrecordedbyothers.Documentationisastable,

unobtrusive andexact source of data, butits weaknessstems

from low retrievability, biased selectivity and potentially

difficultaccessto suchdocuments.This difficultyin accessing

documentscanbeexacerbatedinfamilyfirmsthatareknownto

be protective of their privacy (Gersick, Davis, Hampton, &

Lansberg,1997;Tagiuri&Davis,1996).Forexample,inourstudy

ofgoalsettinginfamilybusiness,wemadeuseofdifferenttypes

ofdocumentation:

Wegatheredarchivaldocumentsfromvarious sourcesinall

firms,includingcontracts,historicalbooksaboutthe

organiza-tionandthefamily,corporatewebsites,newsarticlesaboutthe

firmandthefamily,andfirmpamphlets.(Kotlar&DeMassis,

2013)

In case study research, documents are typically used to

corroborate and augment evidence from additional sources, as

we didin the example mentioned previously: ‘‘Taken together,

these secondary sources of data provided a richer context to

understand goal setting in family firms [...] and corroborate the

collectedevidence’’(Kotlar&DeMassis,2013).

Historicalrecordsaredocumentsthathavebeenamassedover

time in the span of an individual’s or organisation’s existence.

Historicalanalysisistheprocessofassembling,criticallyexamining,

andsummarising these recordsof the past(e.g., articles,books,

business reports, and periodicals) that are typically gathered

throughpubliclyavailable,publishedsourcesofinformation.Itis

worthnotingthathistoricalanalysisdoesnotonlymeandigginginto

thepasttorecoverdatabutalsoframetheevidencecollectedinto

the propercontext, whichmay beextremely differentfromthe

present (Colli, 2012; Colli, Howorth, & Rose, 2013). This data

collectiontechniqueischaracterisedbyseveraladvantages.First,it

focusesoninformationcollectedatthetimeinwhichtheevents

occurred (notretrospective). It uses information gathered from

multiple sources (i.e., different reporters, scholars, and market

experts), allowing the collection of primarily factual data (not

interpretative). It is particularly well suited for studying the

chronologicaldimensionsofpastevents.Morespecifically,historical

recordsallowfamilybusinessscholarstoconsidertimehorizons

thatarelongerthanastandardresearchprojectandevenlongerthan

a particular individual’s lifespan. In addition, the analysis of

historicalrecordsisincreasinglyconsideredtoberelevantnotonly

becauseitprovideslongitudinalevidencebutalsobecauseithasthe

potentialtodevelopandextendexistingtheoryandtobuildnew

theorythankstothemasteryof‘‘twokeycomparators’’–timeand

space(Buckley,2009).Assuch,historicalrecordscanbeveryuseful

forfamily businessresearcherswhoareinterested inlong-term

trends(e.g.,generationaltransitionsoverlengthyperiodsoftime)or

multiple generations of family involvement and prove to be

particularlyusefulinsomecriticalareasoffamilybusinessresearch

that directly deal with the ‘‘long run’’, such as intra-family

succession,knowledgetransferand accumulation,andcorporate

values.Ibrahim,Soufani,andLam(2001),forexample,usehistorical

records to study how intra-family succession occurred through

multiplegenerations:

First,a studyofpublicdocumentswasconducted.Second,to

reduce the amount of data to a manageable form, critical

decisions,actions,andincidentswereidentifiedin

chronologi-calorder.Third,investigationofeachgenerationalperiodwas

carriedoutandacasehistorywasdeveloped.Finally,critical

actionsandincidentsweretriangulatedtoestablishvalidity.

Miller, Steier, and Le Breton-Miller (2003) offer another

exampleofexploratoryinductivestudybasedonhistoricalbook

accounts of thirteen major familyfirms as well as a series of

newspaperandjournalarticlesthattheauthorscompiledonwhat

happenedtothestrategyandorganisation ofthesefamilyfirms

during5–10yearsaftersuccession.

Insum,theuseofmultiplesourcesofdataisastrengthofacase

study.Byusingdifferentdatasources,researcherscantriangulate,

that is,adopt,differentanglesfromwhich toobserve thesame

phenomenon(Denzin&Lincoln,1994;Jick,1979;Pettigrew,1990;

Stake, 2013; Yin, 1984), thus making their findings more

convincing and accurate (Tracy, 2010). The triangulation of

evidence frommultiple data sources is especially important in

the familybusiness setting, where it is particularlydifficult to

separate familyrelationships and aspects related to thefamily

spherefrombusinessdecisionmaking.However,theprerequisites

forusingmultiplesourcesincludetheavailabilityofresourcesand

sufficient knowledge in different data collection techniques.

Moreover,theviewsofdifferenttypesoforganisationalmembers

inthefamilybusinessmaybeveryheterogeneous.Forexample,

theviewsoffamilymembersmaysystematicallyvaryfromthose

ofnon-familymembersbecausethefamilystatusororganisational

roles of both influence their interpretations. Two practical

problems that are typical of multiple-informant studies may

thereforeoccur(Kumar,Stern,&Anderson,1993):(i)theselection

problem, that is, the challenge of identifying two or more

informantscompetenttoreportonaparticulardyadicrelationship

and(ii)theperceptualagreementproblem,thatis,thefrequent

dissimilarityofthereportsofcompetentmultipleinformants.A

consensualapproachtoreconcilingmultipleinformantreportscan

thus be important in order toensure correspondencebetween

informantreportsandactualevents(Glick,Huber,Miller,Doty,&

Sutcliffe,1990)eventhoughtoourbestknowledge,thisapproach

hasneverbeenappliedinfamilybusinessresearch.

A useful way to organise and document the data collected

throughmultiplesourcesisthecreationofacasestudydatabase.

Usingadatabaseincreasesthereliabilityoftheresearchbecauseit

enablestheresearchertotrackandorganisedatasourcesincluding

notes (e.g., results of interviews, observations), key case study

documents,tabular materials(e.g., quantitativedata), narratives

(e.g.,open-endedanswerstothequestions),photographsandaudio

filessothattheycanbeeasilyretrievedatalaterdate.Doingso

makestheprocessthroughwhichfindingsareobtainedextremely

explicitandreplicable,therebypermittingtheconstantreplication

ofthestudyacrosstime,researchers,andanalysistechniques.For

example,whilecollectingthedataaboutgoalsandgoalformulation

processesinfamilyfirms(seeKotlar&DeMassis,2013),wetracked

theemergingresearchdesignandkeptan‘‘audittrail’’,thatis,an

exhaustivechronologyofresearchactivitiesandprocessesaswellas

theemergingthemes,categories,models,andanalyticmemos.In

thefinalarticle,wethenfollowedthegraphicapproachproposedby

GioiaandThomas(1996)toshowkeyaspectsoftheprocessthrough

which the main constructsand relationships inourstudy were

derived.Asanotherexample,Karraetal.(2006)organisedcasestudy

data ‘‘into an ‘event history database’ [...] by chronologically

orderingdescriptionsofeventstakenfromtherawdata—interview

transcripts,interviewandfieldnotes,andsecondarysourcessuchas

journalists’accountsofthepoliticalandeconomiccontext—andby

juxtaposingmultipleaccountsagainsteachothertoascertainthe

degreeofconvergence’’(pp.865-866).

A typicalpitfall that qualitativeresearchers fall into isthat they fail

toestablisha‘‘chainofevidence’’intheirstudies.Thisisunfortunate

because we strongly believe that case study research should

unambiguouslyallowanexternalobservertofollowthederivation

ofanyevidencefromtheinitialresearchquestionstotheultimate

conclusion.Wethereforeencouragefamilybusinessresearchersto

pay particular attention to this aspect, clearly explaining and showing

how the empirical evidence is obtained, linking such empirical

evidence to the data sources from which it is derived, and

6. Analysinginformation

Inadditiontofollowingorderedandtransparentdatacollection

procedures, it is also important to analyse qualitative data

systematicallyandexplainthedataanalysisprocess.Asforother

qualitative methods, data collection and analysis often occur

concurrently. An illustration is provided by our study of

goal-settingprocessesinfamilyfirms:

We iterativelyanalysedthequalitative databymoving back

and forth between the data and an emerging structure of

[empiricalevidenceand]theoreticalargumentsthatresponded

tothetheoryquestionspresentedabove,accordingtothreekey

steps.(Kotlar&DeMassis,2013)

Whiletheoverlapbetweendatacollectionandanalysisallows

researchersto make on-going changes to some aspects of the

researchdesignandtoidentifynewissuesthatarerelevantforthe

researchobjective,italsoinvolvespotentialrisksrelatedtodata

manipulation.

There is a vast arrayof techniques foranalysing qualitative

information:explanationbuildingandwithin-caseanalysis,

cross-caseanalysis,andpatternmatching.3_{Veryoften,qualitativefamily}

businessresearchersdevotesubstantialspacetoreportthatthey

collectedmultiplesourcesofdata,butforgettoexplainhowthey

analysedanythingotherthantheinterviewdata.Thisis

unfortu-natebecauseaclearanddetailedexplanationofhowthecollected

data have systematically been analysed is very important for

reviewerstobetterevaluatethestrengthsofacasestudymethod.

Onefurtherriskassociated withthedataanalysisphaseis that

eachdatasourcewouldbetreatedindependentlyandthefindings

reportedseparately.Thisisnotthepurposeofacasestudy.Rather,

theresearchermustensurethatthedataareconvergedinanattempt

tounderstandtheoverallcase,notthevariouspartsofthecase,orthe

contributingfactorsthatinfluencethecase.Tothisaim,apotentially

useful strategy may include involving other research team

membersinthedataanalysisphaseandaskingthemforfeedback

ontheabilitytointegratedifferentdatasourcesandlinkthemto

the research findings. A hallmark of high-quality case study

research is a clear and unambiguous explanation of how the

researcherevolvedfromthedatatothefindingsoftheirstudy.We

stronglybelievethatthemoresystematicallyqualitative

research-ersareabletoperformthedataanalysisphase,themoresuccessful

theywillbeindevelopinghigh-qualitycasestudypapers.

Before beinganalysed, information collected througha case

studymethodmust be‘‘prepared’’ byrelyingon datareduction,

datadisplay,datacategorisationanddatacontextualisation

techni-ques.4_{Datareductioninvolvesselecting,focusing,condensing,and}

simplifyingthecollectedmaterialinordertoeasetheanalysisof

thecasestudyevidence.Theprocessshouldbeguidedbythinking

aboutwhichdatabestanswertheresearchquestions.Datadisplay

involvescreatinganorganised,compressedwayofarrangingdata

(suchasthroughdiagrams,charts,matrixes,imagesortexts).The

aimistomaketheinformationasaccessibleaspossibleinorderto

facilitatetheidentificationofthemesand conclusions.Thisstep

usuallyinvolvesdatacoding,wheretheresearchersmarkpassages

oftext(orpartsofimagesorsectionsofavideo,etc.)thathavethe

samemessageorareconnectedinsomewayandthenwritean

accompanyingexplanationofwhattheselectedpassageshavein

common.Datacategorisationinvolvesdistinguishingandgrouping

different categories of information. The aim is to decompose

information,aggregatingthemintocategoriesthatallow

compar-isons and distinctions. Finally, data contextualisation involves

assemblingthecollectedinformationandtheexternal

contingen-ciesandidentifyinglinksandconnections.Theaimistoenlighten

the likely relationships with events and contextual conditions.

Here follows a brief illustration of the data manipulation and

analysisproceduresthatwefollowedinourstudyontheproduct

innovationprocessoffamilyversusnon-familyfirms:

Beforebeinganalyzed,informationgatheredthroughthecase

studieswasmanipulatedbyapplyingdatacategorizationand

contextualizationtechniques[...].We thenfolloweda

struc-tured process for data analysis, made up of a preliminary

within-case study, an explanation-building investigation,

followed up by a cross-case comparison. These structured

proceduresfordatacollectionandanalysis,aswellastheuseof

the semi-structured interview guide, helped enhance the

reliabilityoftheresearch[...].(DeMassis, Frattini,Pizzurno,

etal.,2013)

Thefollowingishowweanalysedinformationinanotherstudy

ontheroleofearlyadoptersinthediffusionofnewproductsin

platformversusnon-platforminnovations:

The collected information was manipulated before being

analysedbyapplyingdatacategorisationandcontextualization

techniques[...].Wethenfollowedastructureddataanalysis

processconsistingofa preliminarywithin-casestudyandan

explanation building investigation, followed by a cross-case

comparison. We used differentcategorizations to searchfor

similarities and differences between the cases by creating

severalpartiallyorderedmatrices(Miles&Huberman,1999).

Wetheninterpretedtheobserveddifferencesregardingtherole

of early adopters and launched decisions in light of the

distinctivecharacteristicsofplatformandnon-platform

inno-vations. We decided to consider as reliable a piece of

informationonlyifitwasreportedinatleasttwointerviews

with key informants from different classes of firms (i.e.

innovatingfirm, earlyadopter and late adopter)or it found

confirmationinthedatacollectedthroughsecondarysources,

suchascompanyreportsandproductliterature.Theintegration

ofthesedifferentdatasourcesinatriangulationprocess[...]

was done to reduce post hoc rationalization and personal

interpretation biases from the interviewed people. These

structuredproceduresfordatacollectionandanalysis,aswell

as the use of the semi-structured interview guide, helped

enhancetheconstructvalidity,internalvalidityandreliability

ofourresearchapproach.(Frattinietal.,2013)

Theuseofqualitativedataanalysissoftware(NVIVO,

Hyper-RESEARCH and ATLAS.ti are only someexamples) can be very

usefultobringrigour tothedataanalysisphase, asit supports

researchersinsystematicallycodingand organisingvoluminous

amountsofdataandinmanagingtheanalysisworkofdeveloping

categories,tracinglinkagesbetweenconcepts,andunderstanding

relationshipsamongcategories,whichiswhatwedid,forexample,

inourstudyofgoalsettinginfamilyfirms:

Inordertounderstandthegoalformulationprocessesinfamily

firms, we independently read interviews, observations and

archivaldata,applyingopeninvivocodingusingthequalitative

dataanalysis programNVIVO(QSR International,version9),

whichalsoenabledexchangingmemostocapturethemesand

broadobservations.(Kotlar&DeMassis,2013)

ThestudiesbyGravesandThomas(2008)andIravaandMoores

(2010)arefurtherexamplesoffamilybusinesscasestudiesusing

NVIVOsoftwareforcodingandforexploringpatternsacrosscases.

3

Explainingeachofthesetechniquesisbeyondthescopeofthisarticle,butthe mostusedreferencesareEisenhardt(1989),Yin(2003),andStake(1995).Allof themprovideusefulsupport,andweencouragequalitativefamilybusinessscholars toreferencethesesourcesinordertojustifythedataanalysisproceduresthathave beenfollowed.

4

Commonly usedreferencesforqualitative datapreparationare Milesand Huberman(1994)andTesch(1990).

Nevertheless,therearealsosomepotentialconcernsassociated

withtheuseofqualitativedataanalysissoftware(Barry,1998).

These include progressively deterministic and rigid processes,

privilegingofcodingandretrievaltechniques,increasedpressure

onresearchers tofocusonvolumeandbreadthratherthanon

depthandmeaning,timeandenergyspentonlearning touse

computer packages, and distraction from the real work of

analysis.Inourexperience,qualitativedataanalysissoftwareis

particularly useful for coding when dealing with massive

amountsofdata.Ausefulfeatureofmanysoftwareprogrammes

isalsothepossibilitytoexportdataafterthecodingprocedure

in ordertotakeadvantage ofother software fordataanalysis

andpresentation.

7. Presentingresults

Acriticalaspectofcasestudyresearchispresentingtherich

body of empirical evidence and results of the study. The

presentation ofresultsis a key challengefor qualitative family

businessscholars(Chenail,2009),whotoooftenproduce

manu-scriptsthatareexcessivelylongandresultinmassive,unreadable

documents.

Itisimportanttodistinguishbetweentwotypesofdatatobe

displayed: the raw material and the synthetic evidence. The

challengeofpresentingtherichbodyofqualitativerawmaterial

canbeaddressedbypresentingarelativelycompleterenderingof

thestorywithinthetext.Thestorytypicallyconsistsofanarrative

descriptionof thecasestudiesthatis interspersedwithquotations

fromkeyinformantsandothersupportingevidence.Thefollowing

quote,extractedfromourstudyongoalsettingprocessesoffamily

firms,isillustrativeofhowweaccountedforapieceofthestory

abouthow bargaining emerged fromsocial interactions among

organisationalmembers:

[Our data allowed us to] observe two major stages that

characterize the practical patterns through which

organiza-tional members interacted: the bargaining mode [...], and

stabilizationoftheirgoals[...].Bargainingwasobservedwhen

organizationalmembersconveyedinterpersonalinteractionsin

the form of either settlements or disputes. Settlements

consisted in negotiations of goals among two or more

organizational members with symmetrical influences. For

instance,asettlementepisodewasobservedduringameeting

between a CEO and a minority ownerwho also servesas a

professional manager. The parties were discussing an issue

related to next year’s budget and, when the professional

manager expressed some concerns relating to the dividend

policy, the two started negotiating until the percentage of

profitsto bereinvestedin thefirm wasincreased.After the

meeting,theprofessionalmanagercommented:‘‘Thisisnotthe

firsttimeourviewshavediffered,butafterall,werespecteach

otherandbothbelievewecanalwaysfindacompromiseaslong

as wewant it’’ (ProfessionalManager).(Kotlar& DeMassis,

2013)

Sometimes,extensivetablesthatreporttherawmaterialare

alsoincludedinthetext.Evenifthesetablesarenotshowninthe

finalsubmission,theuseoftablesisaneffectivewaytoshowthata

rich body of evidencesupports thefindings. Qualitative family

businessscholarsshouldpayparticularattentionthentointerlace

theirstory withthe emergingtheorytoprovea closeconnection

betweenempiricalevidenceandemergenttheory.Thisinterlacing

isextremelyimportantinordertocomeupwithahigh-quality

casestudypaper.

The challenge of presenting the synthetic evidence can be

addressed by using synthetic tables or figures to organise the

findings.Aseparatetablethatprovidesasynopticrepresentation

of thecollectedevidenceis usuallyan effectivewaytopresent

the case study evidence.For example, Table2 summarises the

evidencethatwecollectedinourstudyonthedifferencesinthe

productinnovationprocessbetweenfamilyandnon-familyfirms.

Atableindicatingthescopeandtypesofthefocaltopicunder

investigationandhowitis‘‘measured’’cansometimesbereported.

Continuingwiththesameexample,wereportedTable3asfollows

in order to display an excerpt of comprehensive information

regarding the scope and type of product innovation activities

carriedoutinthe10sampledfirms.

To provide additional illustration, Table 4a–c displays a

synthetic viewof some ofthe findings that emerged fromour

studyongoal-settingprocessesinfamilyfirms.

Often,thefindingsofacasestudytaketheformoftheoretical

propositions.Forexample,inourstudyongoal-settingprocessesin

familyfirms,weinductivelydevelopedseventheoretical

proposi-tions that arereported inTable 5.Theuseof propositionsthat

inspirefutureconfirmatoryresearchincreasesthe‘‘testability’’of

thefindingsandallowsthecreationofaparticularlystrongbridge

fromthequalitativeevidencetotheory-testingresearch.

Finally,westronglyrecommendusingfigurestohelpinterpret

thevariousconceptsandtheirrelationshipsinthequalitativedata.

Figures can be a very effective way of capturing the chain of

evidenceordepictinghowaprocessunfolded.Forexample,inour

studyon goal-setting processesinfamilyfirms,weconstructed

Fig.1tohelpmakesenseofthevariousconcepts thatemerged

from ourstudy and their relationships in ourdata. Thisfigure

summarises and generalises the main findings of our study,

graphically presenting the observations and propositions that

emergedfromouranalysis.

Figuresareveryhelpfultovisuallyrepresentthestudyfindings.

Continuing with the same example, Fig. 2 illustrates how we

visuallyshowedtheempiricalevidencesupportingourproposition

Table2

Synopticrepresentationofthecasestudyevidence. Company Family

business

Strategy Organisation Climate

Radicalvs.incremental Closedvs. openapproach

Cross-functionalteam vs.functionalorganisation

Highvs.lowdecisional autonomy

Risk-taking vs.risk-averse

Highvs.low formalisation

FirmA Yes Incremental Open Functional High Risk-averse Low

FirmB Yes Incremental Open Cross-functional Low Risk-taking Low

FirmC Yes Incremental Open Functional High Risk-averse Low

FirmD Yes Incremental Open Functionalandcross-functional High Risk-averse Low

FirmE Yes Incremental Closed Functional High Risk-averse Low

FirmF No Radicalandincremental Closed Cross-functional High Risk-taking High

FirmG No Radicalandincremental Closed Cross-functional Low Risk-taking High

FirmH No Radicalandincremental Closed Cross-functional Low Risk-taking High

FirmI No Radicalandincremental Closed Cross-functional Low Risk-averse High

FirmL No Radical Closed Cross-functional Low Risk-taking High

thatgoaldiversityisexpressedmorestronglywhenanintra-family

successionisimminent.

Asanadditionalillustration,Fig.3visualisesthefindingsofour

study on the management and organisation of the product

innovationprocessinsmallfamilyandnon-familyfirms.

Murray (2003) offers another interesting example of use of

figuresinacasestudy.Sheplotslongitudinalcasestudydataalong

atimelineandillustratesthesequenceofphasesthroughwhich

thefamilyfirmsprogressovertime,providingspecificfiguresto

visualise the three different types of succession journeys that

emergedfromherstudy.

Insum,reportingtheresultsofacasestudycanbeadifficult

taskforanyfamilybusinessresearcherduetothecomplexnature

ofthismethod.Itischallengingtoreportthefindingsinaconcise

mannerthatis appropriatefor publicationin management and

organisationjournals,andtheresearcher’sabilitytocondenseand

convert a complex phenomenon into a format that is readily

understood bythereader isvery important.A typicalpitfallin

presentingtheresultsofacasestudyisbeingoverwhelmedand

distracted by the huge amounts of interesting data that are

superfluoustotheresearchquestions.Asuggestedwaytoavoid

thispitfallandreportacasestudyisbytellingthereaderaconsistent

storyinwhichtheresearchernotonlydescribesthethemes,but

howthose themes fittogether. Providinga clearand convincing

descriptionofthecontextwithinwhichthephenomenonoccursas

wellasthephenomenonitselfisacommontraitofhigh-quality

casestudyarticles.

8. Ensuringvalidityandreliabilityinresearchfindings

Therearefourmaincriteriathathavebeenadoptedtoassess

the rigour and trustworthiness of case study research (e.g.,

Campbell, 1975; Eisenhardt, 1989; Gibbert, Ruigrok, & Wicki, 2008;Gibbert &Ruigrok, 2010;Yin,1981). Table6provides an

Table3

Productinnovationinthestudiedcases.

Company Typeofproductinnovationprojects No.ofresourcesin R&D(full-time equivalent) AnnualR&D budget(Euro) Productinnovation performance

FirmA Ninety-fivepercentofNPDprojectsthefirmhasinitiatedin thelastfiveyears(approximately20)hasbeenmotivatedby theattempttosatisfyalatentneedinthemarket.Veryoften, customersgetintouchwiththefirmtoaskforimprovements ormodificationstotheircurrentproducts.Theseimproved productsarethenalsotransferredtootherclientsofthefirm.It happenslessfrequently(in5%ofproductdevelopment projects)thatthefirmdevelopsnewsolutionsortechnologies forwhichamarketapplicationissought.Eightypercentof NPDprojectsconsistofon-demandmanufacturingnewplastic mouldsbyapplyingthefirm’sknowledgeinthefieldof thermosettingtonewmaterialsandcutting-edgetechnologies thatarepurchasedfromcorporatepartners.Theaverage budgetperNPDprojectis160kEuro,andtheaverageproject durationis8months.

4 800,000

(80%forproduct innovation)

40%ofsalesfromnew productssoldinthelast3 years;productinnovation isthemainissueinthe firm’sindustrialplan 2010-2012

FirmB Classicproductinnovationprojects(70%ofthetotal)entail developingnewproductsthatbettersatisfycustomerneedsby improvingthesuperficialmaterialofwoodenproducts.Five yearsago,thefirmdecidedtoleveragetheincreasingattention ofitsclientstowardstheenvironmenttoinnovateitsproducts andmanufacturingprocessesbyreducingtheirenvironmental impact.Sincethen,thefirm’sinnovationeffortshavebeen directedtowardsimprovingthesustainabilityofitsoffering,to takeadvantageofthegrowingdemandfor‘‘green’’products andservices.Environmentalinnovations(30%ofthetotal)are carriedoutbyco-designingecologicalandsociallysustainable newproductswithuniversities,researchcentres,suppliersand customers.Fournewprojectsarelaunchedeveryyear,the averagebudgetperNPDprojectis40kEuro,andtheaverage project’sdurationis6months.

6 160,000

(70%forproduct innovation)

Thewholerangeof productshasbeen completelyrenewedin thelast3years;winnerof severalawardsfor productinnovation

... ... ... ... ...

FirmL Productinnovationprojectsareusuallytriggeredbythe identificationofnewtechnologies,whicharethendeveloped sothattheycanbeappliedtothefirm’sproducts.The functionalitiesenabledbythenewtechnologiesaretested withselectedclientstounderstandhowtheycanbebetter integratedinthefirm’soffering.Ninety-fivepercentof projectsconsistofthedevelopmentofnewcutting-edge technologies,materialsandproductarchitectures characterisedbyveryinnovativefunctionalitiesand exceptionalperformance.Theremaining5%ofprojectsconsist ofmakinggradualimprovementsoverexistingproducts.Two newproductinnovationprojectswerestartedin2010.The averagebudgetperinnovationprojectis100kEuro,andthe averageprojectdurationis30months(6monthsforthe minorityofprojectsimplyingsimpleimprovementsover existingproducts). 4 750,000 (70%forproduct innovation) 50%ofprofitsinvestedin activitiesrelatedto productinnovationevery year

Table4

Summaryofempiricalevidenceemergedfromthestudyongoal-settingprocessesinfamilyfirms.

a. Organisationalmembergoalsinthefamilyfirmbygoalcontentandgoalrecipient

Goalcontent Goalrecipient

Family Non-family

Economic Familycontroloverthecompany

Familywealth

Firmgrowth Firmsurvival

Firmeconomicperformance

Non-economic Familyharmony

Familysocialstatus Familyidentitylinkage

Firminternalserenity Externalrelations

b. Differencesbetweenprofessionalandfamilialsocialinteractionprocesses

Professionalsocialinteractions Familialsocialinteractions

Setting Businessenvironmentonly Businessandfamilyenvironment

Norms Schedulesanddefinedroles Irregularityandambiguousroles

Meansofbargaining Promisesofrewards,threatsofsanctions Valueabstraction,expressionsofaffect

Meansofstabilisation Formalagreement Socialcontrol

c. Stabilisationsandactsofcommitmenttofamily-centredgoalsbyorganisationalmembershipofinformants

Actsofconformationtofamilyvalues Resourcededication Total Familymembers Shareholders Activelyinvolved Familymembers Shareholders Activelyinvolved Professionalsocialinteractionsa

18% 27% 21% 26% 22% 29% 32%

Familialsocialinteractionsb

66% 50% 48% 78% 53% 58% 74%

FromKotlarandDeMassis(2013).

a

Percentagesrefertothepercentageofprofessionalinteractionsconcludedbystabilisation;92totalepisodes.

b

Percentagesrefertothepercentageoffamilialinteractionsconcludedbystabilisation;73totalepisodes.

Table5

TheoreticalpropositionsdevelopedbyKotlarandDeMassis(2013).

Proposition1 Goaldiversityismorestronglyexpressedwhenanintra-familysuccessionisimminent.

Proposition 2 Thegreaterthegoaldiversity,thehighertheoccurrenceofgoal-centredsocialinteractionprocesses.

Proposition 3 Professionalsocialinteractionsinvolveadministrativebargaining,whereasfamilialsocialinteractionsinvolveaffectivebargaining. Proposition 4 Inprofessionalsocialinteractions,stabilisationisachievedthroughformalcontrols,whereasinfamilialsocialinteractionsstabilisation

followssocialcontrolmechanisms.

Proposition5 Thestabilisationoffamily-centredgoalsismorelikelytooccurthroughfamilialthanprofessionalsocialinteractions. Proposition 6 Thehighertherelianceonprofessionalsocialinteractions,thelowerthecollectivecommitmenttofamily-centredgoals. Proposition 7 Thehighertherelianceonfamilialsocialinteractions,thegreaterthecollectivecommitmenttofamily-centredgoals.

Fig.1.Aprocessviewofgoal-settinginfamilyfirms. FromKotlarandDeMassis(2013).

overview of the four validity and reliability criteria and

sum-marisesthechallenges,issues,andsuggestedresearchstrategies

thatqualitativeresearchersmaytakeforeachcriterion.

Construct validity refers to the extent to which a study

investigateswhatit claimstoinvestigate,that is,thequalityof

theconceptualisationoroperationalisationoftherelevantconcept

(Denzin&Lincoln,1994). Oneofthemainissuesforcasestudy

researchers is that these researchers tend to use ‘‘subjective’’

judgmentsratherthanusingawell-consideredsetofmeasures.

Threeresearchstrategiescanbeconsideredtoenhanceconstruct

validity. First, researchers shouldseek totriangulate data from

multiplesources (Yin,2003). The collectionand comparison of

thesedata ensuresthat thetopicunderstudyis analysedfrom

multipleperspectivesandenhancesconstructvaliditybasedonthe

conceptofideaconvergenceandtheconfirmationofresults(Knafl

&Breitmayer,1989).Second,researchersshouldsharetranscripts

and drafts with participants (i.e., members of the researched

organisation) for consistency and accuracy. Researchers may

integrateaprocessofmembercheckinginwhichtheycancheck

or clarify events that the informant participated in, and the

informantcancontributeadditionalperspectivesontherelevant

topic.Third,researchersshouldsharethetranscriptanddraftsof

theircasestudywithotherinvestigatorsandhavethemreviewed

bypeers(i.e.,academicsotherthantheauthorsofthecasestudy).

Internalvalidityreferstothedataanalysisphase(Yin,2003)and

refers to the establishment of causal relationships between

variables and results.5 _{Here, the issue deals with whether the}

researcher is able to establish a plausible causal relationship,

logicalreasoningthatisrigorousandconvincingenoughtodefend

the research conclusions. Three research strategies can be

consideredtoenhanceinternalvalidity.First,researchersshould

analysetheircasestudybybuildingasoundexplanationaboutthe

case(explanationbuilding).Morespecifically,theyshouldstipulate

apresumedsetofcausallinks,whichdemonstrate,forexample,

that variable x leads tooutcome y and that y wasnot caused

spuriouslybyathirdvariablez.Second,throughpatternmatching,

researchers should empirically compare the observed patterns

witheitherthepredictedpatternsorthepatternsestablishedin

previous studies and in different contexts. Third, cross-case

Fig.2.Goaldiversityinthestudiedfirmsbystageofintra-familysuccession.Goaldiversityisproxiedbythenumberofdifferentgoalsthatemergedfrominterviewsatthe firmlevel.

FromKotlarandDeMassis(2013).

Fig.3.Differencesintheproductinnovationprocessbetweenfamilyandnon-familyfirms. FromDeMassis,Frattini,Pizzurno,et al.(2013).

5

Itshouldbenoted,assuggestedbyYin(2003),thatinternalvalidityasa criteriontoassesstherigourandtrustworthinessofcasestudyresearchcanbe appliedonlytoexplanatorystudies(nottodescriptiveorexploratorystudies).