Revista
de
Administração
http://rausp.usp.br/ RevistadeAdministração52(2017)189–198
Human
Resources
and
Organizations
Informational
status
in
intra-organizational
networks:
The
role
of
knowledge
sharing
and
structural
holes
Status
informacional
em
redes
intraorganizacionais:
o
papel
do
compartilhamento
do
conhecimento
e
das
lacunas
estruturais
Estatus
informativo
en
redes
intraorganizacionales:
el
papel
del
intercambio
de
conocimiento
y
vacíos
estructurales
Cristiano
de
Oliveira
Maciel
∗,
Carlos
Eduardo
Liparotti
Chaves
PontifíciaUniversidadeCatólicadoParaná,Curitiba,PR,Brasil Received12July2015;accepted16August2016
Availableonline30December2016 ScientificEditor:MariaSylviaMacchioneSaes
Abstract
Theaimofthis studywastoevaluatetowhatextentthesimilarityofinformationalstatusofintra-organizationalactorsrelatestobehavioral (knowledgesharing)andstructuralantecedents(structuralholes)inaknowledge-intensiveorganization.Thestudywasoperationalizedthrough theanalysisof462dyads(22*(22−1))thatcomprisesocialrelationshipsinadevelopmentorganizationoftechnologyfortelemedicine.The
resultsindicatethatthesimilarityofindependentvariableswasassociatedtosimilarityininformationalstatus,butthereisnointeractionbetween them.Itisconcludedthattheequalstatuscanbeachievedevenwhentwoactorshavedifferentbasesforitsconstruction,whetherthroughknowledge sharingpracticesorthroughstructuralholes.ThisconclusionrelativizeswhatiscalledMattheweffectinstatusresearch.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP. PublishedbyElsevierEditoraLtda.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).
Keywords:Informationalstatus;Knowledge;Structuralholes;Intra-organizationalrelations;Socialnetworkanalysis
Resumo
Nesteestudo,oobjetivofoiinvestigarsetrabalhadoresemumaempresadeconhecimentointensivoquecompartilhamconhecimentonamesma proporc¸ãoequepossuemnúmerosimilardelacunasestruturaissãoavaliadoscomotendostatusequivalente.Oestudofoioperacionalizadopormeio daanálisede462díades(22*(22−1))quecompõemosrelacionamentossociaispresentesemumaorganizac¸ãodedesenvolvimentodetecnologia
paratelemedicina.Osresultadosapontamqueasimilaridadenasvariáveisindependentesseassociaasimilaridadedestatusinformacional,mas ainterac¸ãoentreelasnão.Conclui-sequeaigualdadedestatuspodeseralcanc¸adamesmoquandodoisatorespossuembasesdiferentesparasua construc¸ão,sejapormeiodepráticasdecompartilhamentodeconhecimentooupormeiodelacunasestruturais.Talconclusãorelativizaoqueé denominadoEfeitoMateusnaspesquisassobrestatus.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP. PublicadoporElsevierEditoraLtda.Este ´eumartigoOpenAccesssobumalicenc¸aCCBY(http://creativecommons.org/licenses/by/4.0/).
Palavras-chave: Statusinformacional;Conhecimento;Lacunasestruturais;Relac¸õesintraorganizacionais;Análisederedessociais
∗Correspondingauthorat:RuaXVdeNovembro,1299–CEP80.060-000Curitiba,PR,Brazil.
E-mail:crmaciel.adm@gmail.com(C.O.Maciel).
PeerReviewundertheresponsibilityofDepartamentodeAdministrac¸ão,FaculdadedeEconomia,Administrac¸ãoeContabilidadedaUniversidadedeSãoPaulo –FEA/USP.
http://dx.doi.org/10.1016/j.rausp.2016.12.008
Resumen
Elobjetivoenesteestudioesanalizarsiexistesimilituddeestatusinformativoentrelostrabajadoresdeunaorganizaciónintensivaenconocimiento quecompartenconocimientodemaneraproporcionalyquetienenlosmismosvacíosestructurales.Sehallevadoacaboelestudiopormediodel análisisde462pares(22*(22−1))quecomponenlasrelacionessocialesenunaorganizacióndedesarrollodetecnologíaparatelemedicina.Los
resultadosindicanquelasimilitudenlasvariablesindependientesserelacionaconlasimilituddeestatusinformativo,peronohayinteracciónentre ellas.Seconcluyequelasituacióndeigualdaddeestatuspuedelograrseinclusocuandodosactorestienendiferentesbasesparasuconstrucción, yaseapormediodeprácticasdeintercambiodeconocimientoodevacíosestructurales.DichaconclusiónrelativizaeldenominadoEfectoMateo enestudiossobreestatus.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP. PublicadoporElsevierEditoraLtda.Esteesunart´ıculoOpenAccessbajolalicenciaCCBY(http://creativecommons.org/licenses/by/4.0/).
Palabrasclave: Estatusinformativo;Conocimiento;Vacíosestructurales;Relacionesintraorganizacionales;Análisisderedessociales
Introduction
Thisstudydefinestheconceptofinformationalstatusin orga-nizationsasan occupationof ahierarchicalpositionresulting from the degreeof deference (Podolny, 2005) received by a socialactorfromco-workerstogainknowledgeandskills. Def-erenceisasymbolicmeansof appreciation,conveyed froma sendertoaparticularrecipientandwhichconsidersthe recip-ient’s attributes (e.g., knowledge at work) (Goffman, 1956). Sorokin (1927)originatedthis concept,but thepresent study focuses on the elements beyond formal education character-istics. Examininginformational status allows us to highlight the similarities anddissimilarity amongorganizational actors relatedtothe practicalandconceptualunderstandingof work activities. These features give substance to one of the main informal bases for stratification operating in parallel with bureaucratichierarchies,especiallyintheknowledge-intensive organizationsthataretypicaloftheNewEconomy(Makani& Marche,2012;Phelps,Heidl,&Wadhwa,2012).
Theargumentherefavorsananalysisofinformationalstatus basedonestablishedrelationshipsbetweenthevariousattributes of human capital and their effects on creativity and innova-tion. Morespecifically, the gesturesof devotionor deference somecolleaguestoanotheremployeebasedonthatemployee’s abilitytosolveproblemsandtoshowdivergent(creative) think-ingdistinguishesactorswithhighorlowinformationalstatus. The more a person is judged as creative and as a problem solver by his or her colleagues, the greater is that person’s informationalstatus.Therefore,the variationininformational status,i.e., theposition or rankof asocialactorinthe intra-organizationalnetwork,isamainfactorinfluencingtheresults of the work, especially in knowledge-intensiveorganizations (Campos-Castillo&Ewoodzie,2014).Thistypeofstatusisalso importantinemployeehiringdecisionsandinhowtheprocess of socializingnew membersproceeds (Guechtouli, Rouchier, &Orillard,2013).On theother hand,occupyingahighrank in knowledge is important for workcan hinder the retention ofemployeestreatedasexperts(Casimir, Lee,&Loon,2012; Crane,2012;Joe,Yoong,&Patel,2013).Thispositioncanbe evenmoreharmfulwhenactorswithhighinformationalstatus trytohidetheirknowledgefromco-workerstomaintaintheir appreciation(Peng,2013).Together,theseargumentsillustrate
theneedfor researchontheelementsrelatedtoinformational status,andespeciallyontheirantecedents.
Among the traditional variables that affect the degree of anykindofstatusareinherited(e.g.,race,gender)orachieved (e.g.,formaleducation,employment,andincome) characteris-tics(Piazza&Castellucci,2014).Morerecentstudiesintostatus inorganizationsenrichtheliteraturebyaccountingfor context-specificbehaviorsandthepropertiesofsocialnetworkstructures (Sauder,Lynn,&Podolny,2012).Thebehavioralelementsare actionsbythoseintheorganizationworthyofdeferencefrom peers. Inthissense,the literatureaboutknowledgein organi-zations, even without specifically mentioning the concept of informationalstatus,consistentlypointstoknowledgesharingas themainactivitythathelpsindividualsattaintheroleofexpertor creative(e.g.Crane,2012;Durst&Edvardsson,2012;Fullwood, Rowley,&Delbridge,2012;Joeetal.,2013).Ontheotherhand, theliteratureonsocialnetworksfocusesonlyontheeffectsof relationshipnetworks onknowledge flow,learning,idea gen-eration,creativity,andinnovation,neglectingtheimportanceof informationalstatus(e.g.Bell&Zaheer,2007;Burt,1987,2004, 2007; Chiu,Hsu,&Wang,2006).Whenresearchersconsider status,theconcept isrestrictedtotheideathatthegreater the numberofcontactswithmanyrelationships,thehigherthestatus ofthosewhocontrolsuchcontacts.Thisdefinitionisoften opera-tionalizedinsocialnetworkanalysisbyextractingBonacich’s centrality(Hanneman&Riddle,2011).
Therefore, consideringtheforemost importance of actions and structural features (i.e., features and positions in social networks)indeterminingthedeferencegiventoorganizational actorsinenvironmentswithtechnologiesbasedonexpert knowl-edge(Makani&Marche,2012),thisstudyaimstoevaluatethe extenttowhichsimilarknowledgesharingandnon-redundant intra-organizationalties(i.e.,structuralholes)explainthe sim-ilar informational statuses of intra-organizational actors in a knowledge-intensiveorganization.
of structural holes means that the central actor (ego) has non-redundantsourcesofinformation(contacts)toexploit.Ego (focalactor)canactinabrokerpositiontoincreaseinformation status in a given intra-organizational network by exploiting the control of information that only the broker can access. For example, this enables ego to have a greater probability of synthesizing his/her contact’s ideas. This study measures theproportion of structural holesbytheir absence (using the constraintindex).Becausethisstudyevaluatesthesimilarityof structuralholes,therearenolimitationstousingthisvariable.
Ingeneral,researchontherelationshipbetweentheelements discussedaboveprovidestheoreticalandmethodological con-tributionstostudieson knowledge inorganizations.The first theoreticalcontributionisaconceptualizationofinformational status in knowledge-intensiveenvironments as a hierarchical position attained from accumulateddeference, specifically in terms of (i) generating ideas and (ii) solving problems. The second contributionisthe establishmentof nomological rela-tionships between informational status (a measure of social assessmentbypeers)andbehavioralandstructuralvariables.
The relationships between these elements offer an expla-nation of how different kinds of social mechanisms operate, independently and in interaction, to contribute to the accu-mulation of a specific type of deference (i.e., creativity and problem-solvingability)thatconvergewiththemainmeanings andvalues of knowledge-intensive organizations (Verburg & Andriessen,2011;Whelan,Collings,&Donnellan,2010).
Themainmethodologicalcontributionisademonstrationof howadyadicanalysisthroughtheMRQAPprocedureismore appropriatemeans tostudystatus thantraditional techniques, suchasOLSregressions.Sincestatusisahierarchicalposition thatcanonlyexistifsocialactorsdisplaydeferenceinequalities, itisessentialtoconsiderhowstatusdifferencesareexplainedin adyadiccomparison(betweenpairsofactorsinanetwork). Sta-tusinequalitiesareinitiallyperceivedbycomparingoneactorto theother;dyadicanalysisbecomesinterestingexactlybecause of this comparative logic. It is important to emphasize that the discussion throughoutthiswork, and the dataprocessing techniquesemployed,residesspecificallyatthedyadiclevelof analysis(Rivera,Soderstrom,&Uzzi,2010).Thismeansthat thestudy subjectisnotthe strengthofassociation, for exam-ple, betweenthe status of aparticular actor andhis/her own knowledge-sharinglevel.Thisdescribesatraditionalanalysis. Dyadanalysisusescomparisonsbetweenmultiplepairsformed usingplayersfromwithinagroup.IfJoao,Carlos,andAugusto formanetwork,therearenotthreeobservationunitstocompare butratherseveralpairs.Measuringthestatussimilaritybetween JoaoandCarlos(firstdyad)meansthatthiscanbecomparedto thedegreeofstatussimilaritybetweenJoaoandAugusto (sec-onddyad),andthenwiththedegreeofstatussimilaritybetween CarlosandAugusto(thirddyad),andsoon.
Moreover,analytical method overcomes the difficulties in researching knowledge in small organizations by applying a morerobusttechniquetoextractdatafromsmallgroups(Durst &Edvardsson,2012).TheMRQAPtechniquemakesitpossible togeneratematriceswithalargenumberofties(e.g.,N=462) touseinaregressionatthedyadlevel (Mizruchi&Marquis,
2006)evenforsmallorganizations,e.g.,n=22,asinthepresent study.
Intherestofthisarticle,“Theoretical-empiricalframework” section discussesthetheoretical andempirical frameworkfor thehypothesesregardingthesimilarityantecedentsofthe infor-mationalstatus.“Methodology”sectiondescribestheprocess, and “Dataanalysis and discussion” section presents the data analysisanddiscussestheresults.Finally,“Conclusion”section concludes.
Theoretical-empiricalframework
Similarity/dissimilarityininformationalstatus
The growth in the number of studieson status in organi-zationsisduetoitsimportanceforincreasedcontacts,greater autonomy,creativity, innovation, promotion, andsalary(e.g., Agneessens&Wittek,2012;Benderski&Shah,2013;Bitektine, 2011; Campos-Castillo&Ewoodzie, 2014;Lazega,Mounier, Snijders,&Tubaro,2012;Mura,Lettieri,Radaelli,&Spiller, 2013; Peng, 2013; Torelli, Leslie, Stoner, & Puente, 2014; Turner,1988).Theseearlierworksfoundthatthedegreeof sta-tusofasocialactorresultsfromsystemsofvaluesandmeanings thatnormativelyselectcharacteristics,behaviors,andaffiliations tobestowdistinction.Suchsystemscanbeanalyticallydivided intofourdomains:(i)social,(ii) economic,(iii) political,and (iv) informational.Socialstatus,the firstofthesedomains, is areflectionofcharacteristics(e.g.,race,age,formaleducation, behaviors)thatpromotehonor,respect,andinfluence.Economic status comes from the control and use of property. Political status derivesfrom occupying positions that confer authority (Riahi-Belkaoui,2009;Turner,1988).Informational status,in theorganizationalcontext,resultsfromactionsthatdemonstrate thatasocialactorhasknowledgewithvalueinthesesystemsof valuesandmeanings(Flynn,Reagans,Amanatullah,&Ames, 2006;Sorokin,1927).
Antecedentsofinformationalstatussimilarity
As in the argument above, similar informational statuses resultmainlyfromsimilarbehaviorsvaluedinthiscontext(i.e., knowledgesharing)andfromthestructuralcharacteristics(i.e., structural holes) of social actors in their intra-organizational networks. Knowledge-sharing occurs through social interac-tions, which highlights its necessarily relational character (Casimir etal.,2012; Swift&Virick, 2013).Tsoukas(1996) statesthatorganizationsaredistributedknowledgesystems,and thatthe knowledgeappliedinworkprocessesdoes notreside thehands ofasingleagent.That authoradds thatknowledge becomes tangible,at least inpart, duringthe performanceof socialpracticesinwhichthe implicitisrealized,revealingits valuetoothers.Thus,toshareknowledge,anagentmustbecome awareofitduringaninteractionthatinvolvessuchknowledge (Crane,2012).Moreover,asindividualsdisplay asimilar fre-quencyofinteractionsinwhichtheyshareknowledge,theywill accrueasimilardegreeofdeference(i.e.,informationalstatus). Therefore,thefollowinghypothesisisproposed:
H1. Dyadsmoresimilarinknowledgesharingarealsomore similarininformationalstatus.
Thepresence ofstructural holes(Burt,1992)isassociated withinformationalstatusduetotheeffectsofafocalactor’s(i.e., ego)non-redundanttiesonthegenerationofnewideas.Whena focalactorhastwoormorecontacts(i.e.,alters)withno connec-tionbetweenthem,thenthefocalactoractsasanintermediary intheinformationexchanges,whichtheactorcanhasan advan-tage.Thisposition is referred toas tertius gaudens or “third partywhobenefits”(Burt,1992,p.30).Burt(1992)developed theconceptofstructuralholesandclassifiedthebenefitsof infor-mationresultingfromthreetypes:(i)access,(ii)timing,and(iii) referents.Accessindicatesreceivingusefulinformation,timing meansreceivinginformationearlierthanothersdo,andreferents arecontactswithqualitynetworks,thatis,thosewhoprovide non-redundanttiesforego.AccordingtoBurt(2007),structural holesofferadvantagesfromthepotentialtoimmediatelyuseor synthesizeinformationaboutnewideas.Burt(2004)examined managersofalargeAmericanelectronicscompanyandfound thatthedegreeofstructural holesamongtheseorganizational memberswas relatedtoperformance,wages, andthe evalua-tionsoftheideastheyproposed.Therefore,structuralholesare relatedtothetypeof(non-redundant)informationrequiredto actcreatively,producenewideas,andsolvingproblems.From thisrelationship, itispossibletodeducethat thesimilarityof thistypeofpropertyinthefocalactor’snetworkrelatestothe similarityofinformationalstatus.
H2. Dyadsmoresimilarinstructuralholesarealsomore sim-ilarininformationalstatus.
Prior works indicatedthat both informationsharing activ-ityandthepresenceofstructuralholesarecentralelementsto theroleofexpert(Burt,2004;Crane,2012;Guechtoulietal., 2013).Assumingaconnectionbetweenthesevariablesandthe informationalstatus,thentheremustbearelationshipbetween structuralholesandknowledgesharing.Knowledgesharingcan
varydependingontheproportionofnon-redundanttiesthata focalactorcontrols.Thus,thosewhocanprovidethebest qual-ity information interms of generating ideas, ultimatelyhave moretoshare.Therefore,ahigherproportionofstructuralholes enablemoreknowledgesharingbecauseitgivesthefocalactor thatcontrolsmorenon-redundanttiesthebenefitsof(i)access, (ii)timing,and(iii)referentquality(Burt,1992).Therefore,the similarityinstructuralholesmoderatestheeffectsofknowledge sharingsimilarityonthesimilarityofinformationalstatus.
H3. Dyads more similar in interactions between structural holes and knowledge sharing are also more similar in infor-mationalstatus.
Methodology
The organizational setting for this study is a knowledge-intensiveorganization.Itsoperatingcoreconsistsofengineering professionals developing telemedicine technologies who employmainlyknowledgeastheirmainproductioninput(Joe etal.,2013;Whelanetal.,2010).Theorganizationwasfounded in 2004by engineering studentson auniversity in the south of Brazil. Since then, the firmreceived numerousawards for innovation, such as the Prêmio Ozires Silva de Empreende-dorismo(OziresSilvaEntrepreneurshipAward),PrêmioFinep deInovac¸ãodaRegiãoSul(FINEPSouthernRegionInnovation Award),PrêmioCNI/FIEP(CNI/FIEPAward),Prêmio ANPRO-TEC (ANPROTECAward),andPrêmio IdeaBrasilMédicoe Científico(IdeaBrasilMedicalandScientificAward).During the course of thisstudy, the firm had 22 members,and was startingaprocessofinternationalization.
Thisstudycollectedrelationalandcompositiondatausingthe sociometricsurveymethod(Babbie,1998;Maciel& Machado-da-Silva,2009;Wasserman&Faust,2009).Thequestionnaire consisted of alist of names (relational data) and categorical and interval questions (composition data). As is common in sociometricstudies, thenetworkconsistedof asmallnumber of subjects(N=22),allowingfor acensus,i.e.,thecollection ofdatafromallmembersoftheorganization(Borgatti,Everett, &Johnson,2013).TheSPSS®version23andSmart-PLS soft-warewereusedtoanalyzecompositiondata;UCINET®6.485 wasusedtocalculatesociometricindicators,dyadiccorrelation, andregressionanalyses;andPAJEK®3wasusedtodevelopthe sociogram.
Measurements
(1) Production Department (N=5) (2) Administrative Department (N=5) (3) Marketing Department (N=5)
(4) Research and Development Department (N=7)
1 1
1
2 2
2
2 2
3
3 3
3
3
4 4
4
4
4
4 4
1
1
Fig.1.Interactionpatternswiththescaleofactors’informationalstatuses.( )(1)Productiondepartment(n=5);( )(2)Administrativedepartment(n=5);( ) (3)Marketingdepartment(n=5);( )(4)Researchanddevelopmentdepartment(n=7).
relationshipsweregenerated,whichwerethenanalyzedwitha regression.ThesizeofthenodesinFig.1showsthevariations ininformationalstatus.
Theindependentvariableknowledgesharingwasmeasured on a 10-pointLikert scale using five literature-based indica-torsreflectedinthefollowingstatements:(i)Isharemywork experienceswiththosecolleagueswhoneedthem;(ii)ingroup discussions, I do my utmost toshare my experiences; (iii) I always offerthe informationthat my colleagues mayrequire todotheirwork;(iv)ItellmycolleagueswheneverIthinkof somethingthatcanimprovetheirwork;(v)Iletcolleaguesknow howIdothingsatworksotheymaylearn.Anexploratoryfactor analysiswasperformedontheindicatorstochecktheir dimen-sionality.TheresultoftheKMOtesttoevaluatetheadequacy ofthe datamatrix forfactoranalysis was0.771. TheBartlett sphericitytestwassignificant:χ2[58,591]/degreesoffreedom[10], p-value<0.001.Thecumulativevariancewas65.085.All indi-cators showed a loading above 0.40. The consistencyof the measurewasassessedusingCronbach’salpha(0.85)andbya GuttmanSplit-HalfCoefficient(0.85)toaccountforthesample size.
Theresearcherstookadditionalcaretoassessthe knowledge-sharing construct and validated this measurement model by estimatingapartialleastsquare(PLS)regressionbecausethere werefewobservationscomparedtothenumberof indicators. The PLS technique is appropriate even when the number of indicatorsishigherthanthenumberofobservations.Toassess thevalidity of the constructvia the Smart-PLS software itis necessaryto haveat least two relatedconstructsin the same model. Thus, a perceived informational centrality measure,
which was notused inthis study, was collectedin the same questionnairethatcontainedtheknowledgesharingindicators. Thecentralitymeasureincludedthefollowingindicators,alsoon a10-pointLikertscale:(i)Iamconsideredsomeonewell con-nectedhereinthecompany, withmanycontacts;(ii)because of my contacts,Iam amongthe firsttoget information;(iii) I havemany contactswithpeople fromother departmentsof the company; (iv) informationreachesme very fast,because Iknowmanypeoplehere;(v) Ihaveso manycontactsinthe company that I end up knowing things about almost every-bodyhere.ThetwoconstructswereanalyzedintheSmart-PLS software.
Theindependentvariable,structuralholes,wasobtainedfrom thenetworkofinteractionpatternsdisplayedinthesociogram (Fig.1).The respondentsalso hadtoindicate 3–5colleagues withwhom theyinteract moreoften atwork. From this net-work, the structural holesfor each ego were quantified.This was measured through the constraintindex, a common mea-surethatrepresentstheabsenceofstructuralholes(e.g.,Burt, 1992, 2004, 2007; Jensen, 2008). Because this study evalu-atesthe similarityof structural holes,thereareno limitations totheuseofthisvariable.Themeansof theknowledge shar-ingindicatorsofthe22observations,aswellastheirstructural holemeasurements,weretransformedintosquarematriceswith thedifferencesbetweennodespairsinthenetwork.The vari-ableinteractioneffectbetweenstructuralholesandknowledge sharingwasalsoobtainedthroughthisproceduretotransform egomeasuresintodyadicmeasures.Toavoidmulticollinearity, theinteractionwasgeneratedwiththevariablecenteredonthe mean.
Someof thecontrol variableswereextractedfromSantos, Rossoni,andMachado-da-Silva(2011)(department, hierarchi-cal level, education level, age, and tenure), and others were employedspecifically due tothe contextand research objec-tives: structural equivalence, physical proximity, sex, marital status, and skill in English language. Structural equivalence was generated through a square matrix of dyads that reveals how similar the position of a focal actor (ego) is compared toothersinterms ofthe patternof connections.Iftwo social actorshavethesamepatternof connections,theirrolesinthe social structure are equivalent, and therefore they may have similarinformationalstatuses(Borgattietal.,2013;Mizruchi, 1993). Physicalproximity consistedof a square matrix with thedistanceinmetersbetweeneachpairoforganization mem-berworkstationsordesks.Theremainingvariablesalsoserved to build newmatrices to control for the similarities between other characteristics of the network actors. Age and tenure were subjected todyaddissimilarities analysisandgenerated twomorematrices.Forgender,maritalstatus,education,skill inEnglishlanguage,department,andhierarchicallevel, pairs withthe samevalues wereassigned a valueof 1, and0 oth-erwise.Therefore,14matricesweregenerated,eachwith462 observations.
Dataanalysisanddiscussion
Thedyadicdatamatricesweresubjectedtocorrelationand regression analysis. Table 1 shows the means and standard deviations for each continuous variable and their correla-tionsthrough thequadraticassignmentprocedure(QAP).The results indicate an association between informational status similarity and othervariables. Thus, theserelationships were examinedusingMultipleRegressionQuadraticAssignment Pro-cedure–Double-DekkerSemi-Partialling(MRQAP)(Dekker, Krackhardt, & Snijders, 2007). This procedure is better at assessing relational data than the traditional ordinary least squares(OLS)estimationmethod,becauseitisanon-parametric techniquethat islesssensitivetocollinearityissuesand auto-correlation(Dekkeretal.,2007;Kirschbaum,2012;Reinert& Maciel,2012;Santosetal.,2011).
The QAP multiple regression model is built from dyadic datainmatrices.Anetworkwith22actors,for example,will havematricesgeneratedforeachvariable.Therefore,each com-positecolumnof data(i.e.,non-dyad),such asage, structural holes, or informational status,must betransformed into rela-tionaldatabasesbyassemblingasquarematrixwiththenames oftheactorsintheverticalcolumnandthecalculated dissimi-laritiesforagivenvariablebetweeneachpairofactors,minus thediagonal(n=22).Fornetworkdatathatarealreadydyadic, transformationisunnecessarybecausethearrayisalready com-posed of dyadicrelationships, such as structural equivalence. Inthisstudy,thereare13variablesthatgenerated13matrices with462dyadseach,foratotalof 6006dyads.Thenextstep istoinsertthe13matricesratherthanthevariablecolumns,as intraditionalanalysis.Thus,theMRQAPmodelisaregression ofmatriceswithmatrices.TheQAPproceduregeneratesa dis-tribution of coefficientsthrough the permutation of rowsand columnsofmatrices.Consequently,evenwithoutchangingthe network structure,the orderof nodeschangesrandomly. The generated distribution allowsfor acalculation of the statisti-calsignificanceofthecoefficients(Dekkeretal.,2007;Maciel, Taffarel,&Camargo,2014).Amorecommonexample ofthe applicationofthistechniqueisintheanalysisoftradebetween countries.Becausetheunitofanalysisisdyadic,itisnecessary touseestimationproceduresthatconsidertherelationallevel.
Table1
Results:QAPcorrelations.
Mean Standarddeviation 1 2 3 4 5 6 7
1.Age(years) 27.81 10.62 1.00
2.Tenure(years) 2.10 2.56 0.04 1.00
3.Physicalproximity(m) 6.52 4.30 0.00 0.00 1.00
4.Structuralequivalence 0.10 0.23 0.00 0.00 −0.18** 1.00
5.Structuralholes(SH) 0.488 0.135 0.08 −0.68*** 0.00 0.00 1.00
6.Knowledgesharing(KS) 8.12 1.07 0.22 −0.36** 0.00 0.00 0.09 1.00
7.Interactionvariable(SH)×(KS) 0.21 2.02 0.00 0.00 0.08* 0.13*** 0.00 0.00 1.00
8.Informationalstatus 3.95 5.15 0.07 0.60*** 0.00 0.00
−0.67*** 0.18 0.00
Table2
Results:MRQAPmodels.
Model1 Model2 Model3 Model4 Model5
Sex 0.000(0.000) 0.000(0.000) −0.000(0.000) −0.000(0.000) 0.000(0.000)
Age 0.050(0.092) 0.106(0.085) 0.062(0.088) −0.001(0.087) −0.001(0.086)
Maritalstatus −0.000(0.000) 0.000(0.000) −0.000(0.000) 0.000(0.000) −0.000(0.000)
Instruction 0.000(0.000) 0.000(0.000) 0.000(0.000) −0.000(0.000) 0.000(0.000)
English 0.000(0.000) −0.000(0.000) −0.000(0.000) −0.000(0.000) −0.000(0.000)
Tenure 0.597(0.479)** 0.514(0.517) 0.774(0.538)*** 0.489(0.578)** 0.489(0.575)**
Department 0.000(0.000) −0.000(0.000) −0.000(0.000) 0.000(0.000) −0.000(0.000)
Hierarchicallevel 0.000(0.000) −0.000(0.000) 0.000(0.000) 0.000(0.000) −0.000(0.000)
Physicalproximity 0.000(0.000) 0.000(0.000) 0.000(0.000) 0.000(0.000) 0.000(0.000)
Structuralequivalence 0.000(0.000) 0.000(0.000) 0.000(0.000) −0.000(0.000) −0.000(0.000)
Structuralholes(SH) 0.506(1.347)** 0.369(1.294)** 0.369(1.300)*
Knowledgesharing(KS) 0.473(1.069)*** 0.390(1.064)** 0.390(1.039)**
Interaction(SH)×(KS) 0.000(0.000)
R2 0.362** 0.495*** 0.543*** 0.609*** 0.609***
AdjustedR2 0.348** 0.483*** 0.532*** 0.599*** 0.598***
Dyadicobservations 462 462 462 462 462
Permutations 2000 2000 2000 2000 2000
Standarderrorinparentheses.
* p<0.10. ** p<0.05. ***p<0.01.
Totestthehypotheses,fivemodelsweredeveloped(Table1). Model 1 considers only the control variables and explained about35%ofthevariationintheinformationalstatussimilarity (AdjustedR2=0.348,p-value<0.05).Model2addsthe struc-turalholesvariableandincreasedtheexplanatorypowerto48% (AdjustedR2=0.483, p-value<0.01). Model3 adds onlythe knowledgesharingvariabletothecontrolvariables.Thismodel hadanadjustedR2=0.532,p-value<0.01.InModel4,boththe structuralholesandknowledgesharingvariablesareaddedtothe controlvariables.Theresultofthismodelindicatedthatthetwo maineffectvariablesexplainapproximately60%ofthevariation inthesimilarityofinformationalstatus.Model5,inturn, consid-ersallvariablesinthepreviousmodelandaddstheinteraction betweenstructuralholesandknowledgesharing.Theinteraction variableisformedbymultiplyingthematricesofbothvariables withvaluescanteredonthemeantoavoidmulticolinearity.In thislastmodel,theexplanatorypowerofthevariationin similar-ityofinformationalstatusdidnotincrease(AdjustedR2=0.598, p-value<0.01).Overall,acomparisonbetweenthefirst4models showedthataddingthemaineffectvariables,whetherseparately ortogether,increasedtheexplanationpowerofthevariationin similarityoftheinformationalstatus.
Theresultsofthehypothesesaremorespecificallyinterpreted herefromthelastregressionmodelinTable1.Thefirst hypoth-esis(H1)wasconfirmed(β=0.390,p-value<0.05),indicating thatknowledge-sharingsimilaritycontributestothesimilarityin informationalstatus.Thesecondhypothesis(H2)wasalso con-firmed(β=0.369),butwithp-value<0.10,asindicatedinModel 5.Theconfirmationofthishypothesisshowsthatthesimilarityin structuralholesbetweentwoactorsinthenetwork,i.e.,the pro-portionofnon-redundantties,givesthemsimilarinformational status.Thethird hypothesis(H3),whichpredictedthatdyads withmoresimilarityininteractionsbetweenstructuralholesand knowledgesharingwouldalsohavemoresimilarinformational
statuses,wasnotconfirmed.Rejectingthishypothesisimplies that theratioofnon-redundanttieshasnoknowledgesharing moderatingeffects oninformational status.Therefore, regard-less of the proportion of structural holes, knowledge sharing canwork,eveninisolation,asasignificant sourceof similar-ityininformationalstatusesamongmembersofanorganization (Table2).
Ingeneral,theresultsforthehypothesesemphasizethe sig-nificantinfluenceofthesimilarityinbehavioralandstructural antecedentsonthevariationininformationalstatussimilarity. Bothknowledgesharingbehaviorandtheproportionof struc-turalholesaresignificantpredictorsofsimilarityinakeytypeof statusinknowledge-intensiveorganizations(i.e.,informational status),whichisoftenmerely referredtoas expertise(Crane, 2012).However, itis necessarytocarefully considerthe fact that the hypothesized relationshipbetween the interaction of structuralholesandknowledgesharingwiththedependent vari-ablewasnotconfirmed. Evenif statisticalgeneralizations are not possible, an important analytical (theoretical) generaliza-tionisthatstructuralholesandknowledgesharingbehaviorare alternativepathstoconstructinformationalstatusin knowledge-intensiveorganizations.Thisdoesnotmeanthatitisimpossible toaddthe isolatedeffects ofthesevariables;the resultsshow thatthesimilarityininformationalstatusbetweentwo organi-zational actorsmay resultfrom knowledge sharingsimilarity and/or fromthe proportionof structural holes,butnot neces-sarilyfromthemoderationofonevariableontheeffectsofthe other.
present insofar as there are significant inequalities between members of an organization, can be created and maintained through different elements (i.e., knowledge sharing practices andstructuralholes).Theresultsshowthatbothpracticesand the local structure of the focal actor’s (ego) social networks aresourcesofsocialinequalityinorganizations.Thatis,they appearasadvantagesfortheaccumulationofdeferencerelated toknowledge.However,eveniftheseinequalitiesand dissim-ilaritiesinpositionintheinformalhierarchyaredeemedunfair (Sauder etal., 2012),the practices andelements of the focal actor’ssocialnetworkingarecharacteristicsthatarewonrather thaninherited. Thismeans that changingandevenbalancing suchinequalitiesis,tosomeextent,withinthescopeofthe orga-nizationalactorswhoareatthebaseoftheinformationalstatus pyramid.Knowledgemaybedevelopedandsharedbyanyone, andstructuralholescanbedeliberatelybuilt(Burt,1992).
Anyhow,itisworthmentioningthateffortstoachieve mobil-ity or to balancethe informational status dissimilarities tend todemandmoreeffortthanmaintaininganalmostunchanged socialhierarchy. Thisis duetothe so-called Mattheweffect, namedforthebiblicalpassageMatthew25:29,‘forwhosoever hath,willbegiven,andheshallhaveabundance,butfromhim whohasnot,evenwhathehaswillbe takenaway’,whichis quotedinstudiesofstatusandreferstothefactthatthosewho alreadyhavehighstatuscanmoreeasilyincreasethebasisof theirsocialdeference(Sauderetal.,2012,p.270).Thismeans thathighinformationalstatuspossiblyallowsanorganizational actortoincreasestructural holesdeliberately, andtobemore sought for knowledge. Therefore,the dynamics of the strati-ficationprocessinknowledge-intensiveorganizations,atleast related to informational status, can be better understood by accountingfortheeffectsofstructuralholes,knowledge shar-ing,andthefact thatthesevariablesdonotinteracttoforma moderatingeffect.
Conclusions
Thisstudy evaluatedthe extent towhich the similarity in informational status of intra-organizational actors relates to behavioral(i.e.,knowledgesharing)andstructural(i.e., struc-turalholes)antecedentsinknowledge-intensiveorganizations. Additionally, it tested whether the interaction between these antecedents could result in an additional effect from the (i) knowledge sharing and (ii) structural holes variables, when treated separately. This study used the dyad as the level of analysis(Mizruchi&Marquis,2006),notonlybecausethe orga-nizationusedas thecontextof thisstudy hadasmallnumber of membersin the organization,but also toreflect that insti-tutions,standards,actionlogics,orsystemstratificationresults fromtheinteractionsthatoccuratthemicrolevel.According toGranovetter (1973),investigating socialnetworks allowsa researchertoexaminehowinteractionsandrelationshipsatthe micro level canstructure the behavioral patterns of agreater magnitude. This means that groups or even dyads (pairs of actors)operate asvectors of socialorderandstratificationof highersocialsystems.
The results of this study provide conclusions that can contribute toresearch effortson knowledge inorganizations, especiallywhenconsideringtheembeddednessoforganizations andtheiractors, aswellasthe hurdlestoquantitative investi-gation in contextswithlittle structural complexity andfewer members.Thefirst conclusionisthatthe informationalstatus definition presentedhere, whichconsidersmultiplex relations of(i)ideagenerationand(ii)problemsolving,hasnomological validity (i.e., the variable relates to its antecedents as theo-retically predicted). The second conclusion is that there is a statistically significant association between the informational and status variablesof behavioral andstructural nature.This findingunderscorestheneedtorecognizetheinfluenceof prac-tices(e.g.knowledgesharing)aswellastheembeddednessof organizationalphenomena.Thethirdconclusionisthat knowl-edgesharingandstructuralholesareindependent.Bothvariables relatetostatus,buttherewasnostatisticallysignificant associ-ationwhenexaminingtheirinteraction.Anetworkarchitecture withmorestructural holesisan alternativeor perhaps evena replacementforknowledgesharingbehavior.Thismayindicate thattwoorganizationalsubjectswithhighinformationalstatus, forexample,mayhaveacquiredtheirstatusesthroughdifferent routes:onethroughpracticeandtheotherviarelational archi-tecture. Thisdoesnotexclude thepotential toaddthe effects oftheindependentvariablesconsideredhere(seeresultsofthe regressionmodels).
The fourthand finalconclusion concerns the main impli-cationfor researchonstatusinorganizations.Conceptualized asapositionwithinahierarchicalcategorization,informational status mustalsobeconsideredinany discussionof stratifica-tionasadynamicprocess.Thistypeofstatus,asconceptually definedhere,doesnotoriginate(directly)ininherited character-istics,becauseanyorganizationalactorcancultivatestructural holesandknowledgesharing,evenmakingitpossibleto con-vertinformationalstatusinothertypesofstatus.Therefore,that thisstratificationimperativerests onlessunequalopportunity bases comparedtosomecharacteristicsofsocialstatus. How-ever,theMattheweffectcanalsoaffectinformationalstatus,and thatthosewhoattainedahighstatushavealowercosttoincrease andmaintainit.
Limitationsandsuggestionsforfutureresearch
Afocusononetypeofrelationshiptothedetrimentofother typesalways requirescaution whenanalyzing theresultsand conclusions of a study. This study investigated only interac-tion ties;othertypesof ties, suchas friendship,alsonegative valence relationships,andmayplayakeyroleinthe associa-tionsbetweenthevariablesinvestigatedherein.Thislimitation resultedfromtheuseofthenamegeneratormethodtocompose thenetworks,whichrequiresalotofeffortfromtherespondents andoftencompromisesthequalityofthedata.
betweenthevariablesstructuralholes,knowledgesharing,and tenure.Theresultsshowedastronginfluencefromtenureinall modelstested.
Conflictsofinterest
Theauthorsdeclarenoconflictsofinterest.
References
Agneessens,F.,&Wittek,R.(2012).Wheredointra-organizationaladvice rela-tionscomefrom?Theroleofinformalstatusandsocialcapitalinsocial exchange.SocialNetworks,34(3),333–345.
Babbie,E.R.(1998).Thepracticeofsocialresearch.California:Wadsworth. Benderski,C.,&Shah,N.P.(2013).Thedownfallofextravertsandriseof
neurotics:Thedynamicprocessofstatusallocationintaskgroups.Academy ofManagementJournal,56(2),387–406.
Bell,G.G.,&Zaheer,A.(2007).Geography,networks,andknowledgeflow. OrganizationScience,18(6),955–972.
Bitektine,A.(2011).Towardatheoryofsocialjudgmentsoforganizations:The caseoflegitimacy,reputation,andstatus.AcademyofManagementReview, 36(1),151–179.
Borgatti,S.P., Everett, M.G., &Johnson, J.C. (2013).Analyzing social networks.ThousandOaks:SagePublications.
Burt,R.S.(1987).Socialcontagionandinnovation:Cohesionversusstructural equivalence.AmericanJournalofSociology,92(6),1287–1335.
Burt, R. S. (1992). Structural holes: The social structureof competition. Cambridge:HarvardUniversityPress.
Burt,R.S.(2004).Structuralholesandgoodideas.AmericanJournalof Soci-ology,110(2),349–399.
Burt,R.S.(2007).Secondhandbrokerage:Evidenceontheimportanceoflocal structureformanagers,bankers,andanalysts.AcademyofManagement Journal,50(1),119–148.
Campos-Castillo,C.,&Ewoodzie,K.(2014).Relationaltrustworthiness:How statusaffectsintra-organizationalinequalityinjobautonomy.SocialScience Research,44(1),60–74.
Casimir,G.,Lee,K.,&Loon,M.(2012).Knowledgesharing:Influencesoftrust, commitmentandcost.JournalofKnowledgeManagement,16(5),740–753. Chiu,C.M.,Hsu,M.H.,&Wang,E.T.G.(2006).Understandingknowledge sharinginvirtualcommunities:Anintegrationofsocialcapitalandsocial cognitivetheories.DecisionSupportSystems,42(3),1872–1888. Crane,L.(2012).Trustme,I’manexpert:Identityconstructionandknowledge
sharing.JournalofKnowledgeManagement,16(3),448–460.
Dekker,D.,Krackhardt,D.,&Snijders,T.A.B.(2007).SensitivityofMRQAP teststocollinearityandautocorrelationconditions.Psychometrika,72(4), 563–581.
Durst,S.,&Edvardsson,I.R.(2012).KnowledgemanagementinSMEs:A literaturereview.JournalofKnowledgeManagement,16(6),879–903. Flynn,F.J.,Reagans,R.E.,Amanatullah,E.T.,&Ames,D.R.(2006).Helping
one’swaytothetop:Self-monitorsachievestatusbyhelpingothersand knowingwhohelpswhom.JournalofPersonality&SocialPsychology, 91(6),1123–1137.
Fullwood,R.,Rowley,J.,&Delbridge,R.(2012).Knowledgesharingamongst academicsinUKuniversities.JournalofKnowledgeManagement,17(1), 123–136.
Goffman,E.(1956).Thenatureofdeferenceanddemeanor.American Anthro-pologist,58(3),475–499.
Granovetter,M.(1973).Thestrengthofweakties.AmericanJournalof Sociol-ogy,78(6),1360–1380.
Guechtouli,W.,Rouchier, J.,&Orillard,M.(2013).Structuringknowledge transferfromexpertstonewcomers.JournalofKnowledgeManagement, 17(1),47–68.
Hair,J.F.,Hult,G.T.M.,Ringle,C.M.,&Sarstedt,M.(2014).Aprimeron partialleastsquaresstructuralequationmodeling(PLS-SEM).Thousand Oaks:Sage.
Hanneman,R.A.,&Riddle,M.(2011).Conceptsandmeasuresforbasic net-workanalysis.InJ.Scott,&P.J.Carrington(Eds.),TheSagehandbookof socialnetworkanalysis(pp.340–369).ThousandOaks:SagePublications. Jensen,M.(2008).Theuseofrelationaldiscriminationtomanagemarketentry: Whendosocialstatusandstructuralholesworkagainstyou?Academyof ManagementJournal,51(4),723–743.
Joe,C.,Yoong,P.,&Patel,K.(2013).Knowledgelosswhenolderexperts leaveknowledge-intensiveorganisations.JournalofKnowledge Manage-ment,17(6),913–927.
Kirschbaum,C.(2012).RappersemSãoPaulo:conexões,desconexõese trans-gressões.Redes–RevistaHispanaparaelAnálisisdeRedesSociales,22(2), 5–22.
Langeheine,R.,&Andresen,N.(1982).Asociometricstatusindexwithout subgroupmembershipbias.SocialNetworks,4(3),233–242.
Lazega,E.,Mounier,L.,Snijders,T.,&Tubaro,P.(2012).Norms,statusand thedynamicsofadvicenetworks:Acasestudy.SocialNetworks,34(3), 323–332.
Maciel,C.O.,Taffarel,M.,&Camargo,C.(2014).Embeddednessestruturale espacialemredesestratégicas:efeitosatitudinaisnoníveldasdíades.Revista deAdministra¸cãoMackenzie,15(3),166–190.
Maciel, C. O., & Machado-da-Silva, C. L. (2009). Práticas estratégicas em uma redede congregac¸ões religiosas: valores e instituic¸ões, inter-dependência e reciprocidade.Revista de Administra¸cão Pública, 43(6), 1251–1278.
Makani,J.,&Marche,S.(2012).Classifyingorganizationsbyknowledge inten-sity –Necessarynext-steps.Journalof KnowledgeManagement,16(2), 243–266.
Mizruchi, S. M. (1993). Cohesion, equivalence, and similarity of behav-ior: A theoretical and empirical assessment. Social Networks, 15(2), 275–307.
Mizruchi,M.S.,&Marquis,C.(2006).Egocentric,sociocentric,ordyadic: Identifyingtheappropriatelevelofanalysisinthestudyoforganizational networks.SocialNetworks,28(1),187–208.
Mura,M.,Lettieri,E.,Radaelli,G.,&Spiller,N.(2013).Promoting profession-als’innovativebehaviorthroughknowledgesharing:Themoderatingrole ofsocialcapital.JournalofKnowledgeManagement,17(4),527–544. Peng,H.(2013).Whyandwhendopeoplehideknowledge?Journalof
Knowl-edgeManagement,17(3),398–415.
Piazza,A.,&Castellucci,F.(2014).Statusinorganizationandmanagement theory.JournalofManagement,40(1),287–315.
Phelps,C.,Heidl,R.,&Wadhwa,A.(2012).Knowledge,networks,and knowl-edgenetworks:Areviewand researchagenda.Journalof Management, 38(4),1115–1166.
Podolny,J.M.(2005).Statussignals.Princeton,NJ:PrincetonUniversityPress. Ravlin,E.C.,&Thomas,D.C.(2005).Statusandstratificationprocessesin
organizationallife.JournalofManagement,31(6),966–987.
Reinert,M.,&Maciel,C.O.(2012).Análisedasdíadesparacompreendera semelhanc¸adaac¸ãoestratégica:umaaplicac¸ãodaregressãomúltiplaQAP (MRQAP).Redes–RevistaHispanaparaelAnálisisdeRedesSociales, 22(5),81–105.
Riahi-Belkaoui,A.(2009).Socialstatusmatters.UnitedStates:BookSurge Publishing.
Rivera,M.T.,Soderstrom,S.B.,&Uzzi,B.(2010).Dynamicsofdyadsin socialnetworks:Assortative,relationalandproximitymechanisms.Annual ReviewofSociology,36(1),91–115.
Sauder,M.,Lynn,F.,&Podolny,J.M.(2012).Status:Insightsfrom organiza-tionalsociology.AnnualReviewofSociology,38(1),267–283.
Santos,L.G.A.,Rossoni,L.,&Machado-da-Silva,C.L.(2011). Condicio-nantesestruturaisdosrelacionamentosIntraorganizacionais:umaanáliseda influênciasobrerelac¸õesdecomunicac¸ãoedecisão.Revistade Adminis-tra¸cãoMackenzie,12(1),139–168.
Sorokin,P.A.(1927).Socialmobility.NewYork:Harper&Row.
Swift,M.L.,&Virick,M.(2013).Perceivedsupport,knowledgetacitness,and providerknowledgesharing.Group&OrganizationManagement,38(6), 717–742.
Tsoukas,H.(1996).Thefirmasadistributedknowledgesystem:A construc-tionistapproach.StrategicManagementJournal,17(1),11–25.
Turner,B.S.(1988).Status.Minneapolis:UniversityofMinnesotaPress. Verburg,R.M.,&Andriessen,E.J.H.(2011).Atypologyofknowledgesharing
networksinpractice.KnowledgeandProcessManagement,18(1),22–34.
Wasserman, S., & Faust, K. (2009). Social networkanalysis. Cambridge: CambridgeUniversityPress.
