ML102500633

2010/09/07-NRC000027-The Nature of Safety Cultua Review of Theory and Research

ML102500633
Person / Time
Site:	Prairie Island
Issue date:	09/07/2010
From:	Guldenmund F Delft Univ of Technology
To:	Atomic Safety and Licensing Board Panel
SECY RAS
Shared Package
ML102500629	List: IR 05000282/2008008 IR 05000282/2008009 IR 05000282/2009008 IR 05000282/2009010 IR 05000282/2010003 IR 05000306/2009013 ML102500631 ML102500632 ML102500633 ML102500636 ML102500638 ML102500639 ML102500640 ML102500641 ML102500644 ML102500646 ML102500647 ML102500648 ML102500650 ML102500653
References
50-282-LR, 50-306-LR, ASLBP 08-871-01-LR-BD01, RAS 18573
Download: ML102500633 (43)

Text

Thenatureofsafetyculture:areviewoftheoryandresearchF.W.GuldenmundSafetyScienceGroup,DelftUniversityofTechnology,Kanaalweg2b,NL-2628EBDelft,TheNetherlands AbstractThispaperreviewstheliteratureonsafetycultureandsafetyclimate.Themainemphasisisonappliedresearchcustomaryinthesocialpsychologicalororganisationalpsychologicaltraditions.Althoughsafetycultureandclimatearegenerallyacknowledgedtobeimportantconcepts,notmuchconsensushasbeenreachedonthecause,thecontentandtheconsequencesofsafetycul-tureandclimateinthepast20years.Moreover,thereisanoveralllackofmodelsspecifying eithertherelationshipofbothconceptswithsafetyandriskmanagementorwithsafetyperfor-mance.Inthispaper,safetycultureandclimatewillbedi*erentiatedaccordingtoageneralframeworkbasedonworkbySchein(1992Schein)onorganisationalculture.Thisframeworkdistinguishesthreelevelsatwhichorganisationalculturecanbestudiedbasisassumptions, espousedvaluesandartefacts.Atthelevelofespousedvalueswendattitudes,whichareequatedwithsafetyclimate.Thebasicassumptions,however,formthecoreoftheculture.Itisarguedthatthesebasicassumptionsdonothavetobespecicallyaboutsafety,althoughitisconsidereda goodsigniftheyare.Itisconcludedthatsafetyclimatemightbeconsideredanalternativesafetyperformanceindicatorandthatresearchshouldfocusonitsscienticvalidity.Moreimportant,however,istheassessmentofanorganisation'sbasicassumptions,sincetheseareassumedtobe explanatorytoitsattitudes.

1. 2000PublishedbyElsevierScienceLtd.Allrightsreserved.

Keywords:Safetyculture;Organisationalculture;Safetyclimate;Safetyattitudes;Questionnaireresearch;Safetyperformanceindicator1.IntroductionInthelasttwodecadesempiricalresearchonsafetyclimateandsafetyculturehasdevelopedconsiderablybut,unfortunately,theoryhasnotbeenthroughasimilarprogression.Althoughmostoftheresearchreportedisconductedaccordingtothe familiarroutinesofsocialscienticespeciallysocialandorganisationalSafetyScience34(2000)215+/-257www.elsevier.com/locate/ssci0925-7535/00/$-seefrontmatter

1. 2000PublishedbyElsevierScienceLtd.Allrightsreserved.PII:S0925-7535(00)00014-XE-mailaddress:frank.guldenmund@wtm.tudelft.nl(F.W.Guldenmund).

psychologicalresearch,littleconsensushasbeenreachedonthedi*erentaspectscommonlyassociatedwithaconceptwithinthisscienticdiscipline.Forinstance, whiletheimportanceoftheconceptofsafetyclimateorcultureisstressedbymostauthors,veryfewhaveattemptedtosupporttheirclaimbyreportinganindicationofitsconstructvalidityorpredictivevalidity.Moste*ortshavenotprogressed beyondthestageoffacevalidity.Basically,thismeansthattheconceptstillhasnot advancedbeyonditsrstdevelopmentalstages.Thepresentpaperreviewstheresearchonsafetyclimateandsafetyculture.Itwilltrytoseparateoutdi*erentschoolsofthoughtandviews.Specialattentionwillbe giventothepresenceofatheoreticalmodelinanapproach,becauseitisthought thatsuchamodel,howeversimpleitmaybe,shouldbethestartofanyscienticenterprise.MostofthepapersthathavebeenconsideredforthisreviewarelistedinTable1.Whilenotanexhaustivelist,itisthoughtthatitisrepresentativeofthis researcheld.Researchoncultureingeneralandorganisationalcultureinparti-cularhasbeenofinterestnotonlytosocial,personnelandorganisationalpsychol-ogistsbutalsotosociologists,anthropologistsandpoliticalscientists.Themain emphasishere,however,isonappliedresearchinthesocialpsychologicalororga-nisationalpsychologicaltraditions.Oneimportantassumptionassociatedwiththese traditionsisthatalargegroupoforganisationalculturescanbedescribedwithalimitednumberofdimensions.Suchdimensionsareusuallysoughtthroughlarge,organisation-widequestionnairesurveyswiththeultimatepurposeofdescriptionor diagnosisandpossiblyintervention.Itisacknowledgedthatthisisnotthe wholestory,though.Therefore,someotherapproachesandviewsarealsodis-

cussed.Noreviewofsafetyclimateorsafetycultureiscompletewithoutasummaryofthoseaspectsofthediscussiononorganisationalcultureandclimatethatarerele-vantforthepresentreview.Theseaspectswillbereviewedrst.Next,thedi*erentdenitionsgivenforsafetyclimateandsafetyculturearediscussed.Aswillbeshown,mostauthorsaimatthesameconceptbutdi*eronwhatthisconceptmight encompass,i.e.theiroperationalisationsoftheconceptdi*er.Asamatterofcourse thisleadstoadiscussionofthedimensionalityoftheconceptandthecausalmodel underlyingit.Unfortunately,notmanyauthorshaveputforwardatheoretical modelthatcanbetestedandultimatelybefalsied.Are-ectiononthe importantissueoflevelofaggregationwillroundo*thispartofthereview.There-upon,aframeworkwillbeoutlinedthatintegratesthereviewndings.2.OrganisationalcultureandclimateTheconceptsoforganisationalcultureandclimategainedmuchattentioninthe1970sand1980s.Clearly,theappealofsuchintegrative`umbrella'concepts,espe-ciallyformanagers,isgreat.Theprospectofobtaininganoverallhelicopterviewof one'sorganisationisindeedattractive.However,becauseofthefactthatthesecon-ceptsaresoglobalandabstract,theycanalsoruntheriskofbecomingvirtuallymeaningless.

216F.W.Guldenmund/SafetyScience34(2000)215+/-257 Table1Overviewofthesources,causalmodelsandgoalsofsafetycultureandclimateresearches Reference SourceCausalmodel GoalZohar(1980)Literaturereviewofcharacteristicsthatdi*erentiatebetweenhighversuslowaccident-ratecompaniesBasedonavarietyofcuespresentintheirworkenvironment,employeesdevelopcoherentsetsofperceptionsandexpectationsregardingbehaviour-outcomecontingenciesandbehave accordingly1.Describeaparticulartypeoforganizationalclimate2.ExamineitsimplicationsGlennon(1982a,b)Areviewoftheorganisationalclimateliteratureandsafetymanagement

literatureOrganisationalclimatecanbeviewedasabridgebetweenformalorganisationalcharacteristicsandindividualbehaviour.[

...]Thee*ectsofthecharacteristicsaremediatedthroughtheperceptionsandbeliefsofsignicantindividualsandgroups amongmanagementandemployeesPutahandleonthewithin-company variablesthatconstitutethenecessarysafetyclimateinwhichdesiredbehavioursforhazardreductionandeliminationare

fosteredBrownand Holmes(1986)Zohar(1980)Itisacknowledgedthatnosinglestudycanestablishcausalrelationships

(``chicken+/-eggposition)1.ReplicateZohar'sfactorstructure 2.Establishvalidfactorstructure 3.Exploredi*erencesinclimateperceptionsbetweenpre-andpost-traumaticemployeesLutness(1987)Self-developedNotexplicitlystated1.Revealasafetyprogram'sstrengthsand weaknesses2.Findsolutionstotheproblemsunearthed 3.Establishbenchmarksforasafetyprogram4.DenesafetytrendsCoxandCox(1991)FrameworkbyPurdham(1984)NotexplicitlystatedStudywassetupastherstpartofaprogrammetofurtherdevelopsafetycultureaspartofone meansofimprovingonthecompany's[...]safetycultureDedobbeleerand

BeÂland(1991)BrownandHolmes(1986)NotexplicitlystatedTestofBrownandHolmesthree-factormodel(continuedonnextpage)F.W.Guldenmund/SafetyScience34(2000)215+/-257 217 Table1(continued)ReferenceSourceCausalmodelGoalOstrometal.(1993)Self-developedImproveoveralllevelofsafety1.Discussionoftheconceptofsafetyculture2.Presentingasurveyinstrumentdevelopedtoassessthesafetyculturesoforganisations3.Discussinghowtheresultsofthesurvey instrumentcanbeusedtoimprovesafetycultureSafetyResearchUnit

(1993)Facettheoreticapproach(a.o.

Zohar,1980)Theclimateofanorganisation representsthecontextinwhich behaviouroccursandthebasisofpeople'sexpectations;itprovidesaframeofreferencetoguidebehaviour (cf.Zohar,1980)CooperandPhilips

(1994)Zohar(1980)Notexplicitlystated1.ReplicateZohar'sfactorialstructure2.Ascertainifthefactorstructureisreplicatedinapost-test3.Assessanychangesintheperceptionsof safetyclimatethatmayhaveoccurredasaresultofagoal-settingandfeedbackinterventionNiskanen(1994)ReviewofsafetyclimateliteratureSafetyclimateprovidesalinkbetweenattributesoccurringatthe individual(workersandsupervisors)levelandtheorganisationalleveland maybeinducedbythepoliciesandpracticesthatorganisationsimposeupontheirworkersandsupervisors1.Developanapproachtosafetyclimate 2.Exploredi*erencesinsafetyclimateperceptions3.DevelopnewandsupplementalperspectivesforsafetyperformanceGeller(1994)Engineering(i.e.equipmentdesign)andpsychology(i.e.behaviouralandsocial sciences)ABC-model(A=Activator,B=BehaviourandC=Consequence)``Asafetyprofessional'sultimategoalisto achieveatotalsafetyculture(p.18).(continuedonnextpage) 218F.W.Guldenmund/SafetyScience34(2000)215+/-257 Table1(continued)ReferenceSourceCausalmodelGoalCoyleetal.(1995)Nominalgrouptechniquetodeterminetherelativeimportanceofhealthandsafetyissues+Zohar(1980)andGlennon(1982a,b)Assafetyclimatefactorsarea representationoftheemployees'perceptionoftheirworkenvironment,theiridenticationindicatesthoseareas whereemployeesfeelanalysisand/orchangemaybenecessaryAssesswhetherananalysisoftwohighlysimilarorganisationswouldidentifythesamesetsofsimilarfactorsBerends(1996)Literaturereview,interviewsandclusteringofitsresultsSafetycultureshapestheemployees'safetybehaviour1.Developanalternativemeasureforsafety

performance2.Identifystrengthsandweaknesses3.Assesse*ectivnessofsafetyprogrammes 4.Assesstheopportunitiesandthreatsfortheimplementationofanewsafetymanagement tool 5.Revealdi*erencesinperceptionsbetweenshop-oorworkersandmanagementLee(1996)BasedontheoutcomeofdiscussionsofvefocusgroupsOrganisationswithapositivesafetyculturearecharacterisedbycommunicationsfoundedonmutualtrust,bysharedperceptionofthe importanceofsafetyandbycondenceintheecacyofpreventivemeasures(cf.ACSNI)1.Toidentifythoseaspectsofthesafetyculture thatareinneedofimprovement2.Tocomprehendtheunderlyingorderandstructureofsafetyculturesandtheirdynamic linkswithotherorganicvariables,suchasmanagementstyleCabreraetal.(1997)Zohar(1980)+additionsSafetyclimateisconsideredtoproducea``collectiveschemaofmeaning,thatisthoughttohaveastrongerimpactonbehaviourthansafetypolicies1.Thedevelopmentofaseriesofevaluationmeasuresfortrainingprogrammesaimedat improvingsafetylevelsatairportramps2.Evaluatetherelationshipbetweensafetyclimate,safetylevelandsafebehaviours 3.Evaluatethein-uencesofvariouspersonalandorganisationalfactorsonattitudeandsafetyWilliamsonetal.

(1997)Literaturereviewforaspects,Coxand Cox(1991)andDeDobbeleerand BeÂland(1991)SafetyclimateisthoughttopredictthewayemployeesbehavewithrespecttosafetyinaworkplaceTodevelopameasureofperceptionandattitudesaboutsafetyasanindicatorofsafetycultureforusewithworkingpopulationsF.W.Guldenmund/SafetyScience34(2000)215+/-257 219 Theliteratureon,aswellastheconceptof,organisationalcultureand/orclimatehasalreadybeenreviewedanddiscussed(e.g.JamesandJones,1974;Schneider,1975; Glick,1985;VanHoewijk,1988;Schein,1992).Aswillbecomeclear,severalpointsemanatingfromthesediscussionsareequallyrelevantforthepresentreviewofsafetycultureandclimate.Beforedeningsafetycultureandclimate,thedistinctionbetweencultureandclimatehastoberesolved,i.e.whetheritisusefultomakesuchadistinctionandif so,whythatdistinctionshouldbemadeorifnot,whynot.2.1.OrganisationalclimateversuscultureInthe1970s,muchresearchwasundertakenunderthetitleoforganisationalcli-mate,whichnaturallyalsoresultedinseveraldebatesontheconcept(e.g.JamesandJones,1974;JonesandJames,1979;Glick,1985;DeCocketal.,1986).Gradually, duringthe1980s,thetermculturereplacedthetermclimateinthistypeofresearch.

Hence,thedevelopmentoftheseconceptshasbeensuccessiveratherthaninparallel.

Below,ashortsummaryofthisdevelopmentwillbegiven.JonesandJames(1979,p.205)talkaboutclimatewhichtheydescribe``asasetofperceptuallybased,psychologicalattributes.Toseparateclimatefromjob-relatedattitudesandsatisfaction``thedescriptive and cognitivenatureofpsychologicalcli-mateisstressedandcontrastedwiththea*ectiveandevaluativeaspectsofatti-tudes.Theyneverthelessconcludethatbetweenthetwoconceptsa``dynamic interrelationshipmightbeassumed.Thisdistinctionbetweendescriptiveand a*ectiveattributesisbroughtupbySchneider(1975)intermsof``perceptionsof organizationalpracticesand``reactionstothosesamepracticesandprocedures

(p.464)respectively,althoughheacknowledgesthatitisquitediculttodistinguish thetwo.Ekvall(1983)emphaticallydistinguishesorganisationalclimatefromculture.Hedividesanorganisation'ssocialsysteminto:(1)organisationalculture,i.e.beliefs andvaluesaboutpeople,work,theorganisationandthecommunitythatareshared bymostmemberswithintheorganisation;(2)socialstructure,i.e.especiallythe informalorganisation;(3)organisationalclimate;i.e.commoncharacteristicsof behaviourandexpressionoffeelingsbyorganisationalmembers;and(4)workrela-tions,i.e.especiallythenatureoftherelationshipbetweenmanagementand employees.Ekvallarguesthatallfoursegmentsaremutuallyrelatedbutdistin-guishable.Glick(1985)considersthedistinctionintermsofappliedmethodology,particu-larlybecausethetwoconceptsstemfromdi*erentdisciplines.Hearguesthatresearchonorganisationalclimatedevelopedprimarilyfromasocialpsychological framework,whilecultureisrootedrmlyinanthropology.Evidently,bothdis-ciplinescontributedi*erentresearchparadigms,theformeramorequantitative approachwhilethelatterusesmainlyqualitativetechniquestostudyitsresearch objects.Moreover,researchoncultureismuchmorefocusedonthedynamicpro-cessesatworkinanorganisationalculture,continuouslycreatingandshapingit.Inaddition,Glickconsiderscultureresearchassucceedingclimateresearch.Although 220F.W.Guldenmund/SafetyScience34(2000)215+/-257 initiallydistinguishingclimatefromcultureGlick(1985,p.612)concludesthat``[t]heminorsubstantivedi*erencesbetweencultureandclimatemayprovetobe moreapparentthanreal.VanHoewijk(1988,p.9)describesorganisationalclimateasatermcomprising``severalcorrelatingviews,habitsandtheatmosphere,buttheconceptoforgani-sationalcultureremainsundenedalthoughseveralconvergentanddivergentviews fromvariousauthorsaregiven.Oneofthemostrenownedscholarsintheeldof,especiallynational,cultureresearchisHofstede.Henarrowsorganisationalclimatedowntojobsatisfactionand tosomethingthatistypicallytheconcernoflowerandmiddlemanagement.Organi-sationalcultureisconsideredtobetop-management'sbusiness(Hofstede,1986).DeCocketal.(1986)attempttodistinguishorganisationalclimatefromculture.Theyarguethatorganisationsarecharacterisedbyacoherenceofnumerouspro-cesses.Organisationalclimate,then,istheperceptionofthiscoherencebyallthe members.Ontheotherhand,organisationalcultureistheunderlyingmeaninggiven tothiscoherence,whichformsapatternofsignicanceandvalues.Schein(1992)conceivesofclimateasprecedingculture,i.e.climateiscultureinthemaking.Furtheron,Scheinwritesthat``climatewillbeare-ectionandmanifesta-tionofculturalassumptions(p.230).Climateisreplacedbycultureandculturethenconveysabroaderandmoreprofoundmeaning.Soinitially,thetermorganisationalclimatemighthavesigniedthebroadcon-structenvisionedbyresearchersbut,successively,ithasbeenrestrictedtoattitudinalor`psychological'phenomenawithinanorganisation,whichishowitwasinitially operationalised.Climatewasreplacedbythetermculture,whichnowadayshasthis comprehensivemeaningformerlycoveredbythetermclimate.Ontheotherhand,withintheeldofsafetycultureandsafetyclimateresearch,bothtermsarestillnotablyinuse.Berends(1995a,1996)considersculturesimplyareplacementofclimate.Otherauthors,however,restrictthemselvestothetermsafetyclimateandconsiderthistobethe``psychologicalorattitudinalclimatewithregard tosafetywithinanorganisation(e.g.DonaldandCanter,1994;Niskanen,1994).Forthepresentthefollowingcanbeconcluded.Thetermorganisationalclimatewascoinedtorefertoaglobal,integratingconceptunderlyingmostorganisational eventsandprocesses.Nowadays,thisconceptisreferredtobythetermorganisa-tionalculturewhereasthetermorganisationalclimatehascometomeanmoreand moretheovertmanifestationofculturewithinanorganisation.Therefore,climatefollowsnaturallyfromcultureor,putanotherway,organisationalcultureexpressesitselfthroughorganisationalclimate.Thisisalsoclearfromthewayinwhichbothconceptsarecurrentlyoperationalisedandassessedassumingofcourse,thattheparticularresearcherstilldistinguishesthe two.Organisationalclimateiscommonlyconceivedofasadistinctcongurationwith limiteddimensionalitysurveyedthroughself-administeredquestionnaires.Suchmea-suresare,uptoacertainpoint,objectiveandsemi-quantitative.Organisationalcultureis oftendeterminedphenomenologically,i.e.throughobservationsandinterviews,throughtrial-and-error,mutualcomparisonandthelike.Suchmeasuresareregardedasqualitativeandthusdiculttoquantify.F.W.Guldenmund/SafetyScience34(2000)215+/-257 221 Portrayedthisway,organisationalclimateassessmentshowsalotofsimilaritywithattitudemeasurement.Attitudesareconceptuallydenedas``apsychologicaltendencythatisexpressedbyevaluatingaparticularentitywithsomedegreeoffavorordisfavor.Withinthisdenitionevaluatingrefersto``allclassesofeva-luativeresponding,whetherovertorcovert,cognitive,a*ective,orbehavioral

(EaglyandChaiken,1993,p.1).Anorganisationalclimate,then,wouldbedened orgivenbytheaggregatedattitudesofitsmembers.

1Amongattitudetheoristsitiscommonlyassumedthatbeliefs 2areinsomesensethebuildingblocksofattitudes(EaglyandChaiken,1993).Projectingthisassump-tiononthecurrentdiscussionoforganisationalcultureandclimate,certainstrong organisationalbeliefscouldbeassociatedwithorganisationalculture.Or,putinanotherway,certainbeliefs,orbetterstilldogmasorconvictions,formthecorethatisassociatedwithorganisationalculture.Similarly,thisdistinctioncanbeappliedtosafetycultureandsafetyclimate,withthelatterdenotingattitudestosafetywithinanorganisationandsafetyculturebeing thestrongconvictionsordogmasunderlyingsafetyattitudes.Theselatterbeliefsdo nothavetobespecicallyaboutsafety,butunderlieallorganisation'sattitudes.2.2.CharacteristicsoforganisationalcultureandclimateNowthatcultureandclimatehavebeendistinguished,themostimportantnd-ingsandthelessonslearnedfromtheresearchonorganisationalcultureandclimatearesummarised,whichareconsideredrelevantforthepresentreview.Organisa-tionalculturehasbeengiventhefollowingcharacteristics(needlesstosay,mostof thesecharacteristicsequallyapplytoclimate):1.Itisaconstruct(e.g.Guion,1973;James,1982;Berends,1996).Basicallythismeansthatcultureisanabstractconceptratherthanaconcretephenomenon.This characteristicalreadysetsthestageforsignicantdisagreement,becauseitallows theresearcherconsiderabledegreesoffreedomtobothdeneandoperationalise culture.Whenoperationalisingaconstruct,itisgenerallyassumedthatthereare severalvariablesthatcovaryorttogethertoformanuniedwhole(seealso3).2.Itisrelatively stable.DeCocketal.(1986)havefoundaperiodofstabilityofatleast5yearsfororganisationalculture.3.Ithasmultipledimensionality(e.g.Guion,1973;JonesandJames,1979).Again,thischaracteristicisthecauseofmanydi*erencesbetweenresearchers.Because dimensionsarenearlyalwayscomposites,comprisedofseveralvariables,the labellingofadimensionbecomesverymuchapersonalmatter,re-ectingboth 1Pleasenotethatthecurrentconceptionofattitudesismuchbroaderthan,forinstance,inthe1970s.Atthattime,attitudeswereconsideredtobeprimarilya*ective,notcognitive.ThisledJonesandJames(1979)todistinguishbetweencognitiveanda*ectiveprocessesanddescriptiveandevaluativeresponsesasdescriptorsoforganisationalclimateandjob-satisfaction,respectively.Presently,botha*ectiveandcog-nitiveprocessesandresponsesareconsideredtounderlieattitudes(EaglyandChaiken,1993).

2``Mentalassenttooracceptanceofaproposition,statement,orfact,astrue,onthegroundofauthorityorevidence.(TheShorterOxfordEnglishDictionary,3rdedition,1959).

222F.W.Guldenmund/SafetyScience34(2000)215+/-257 acommondenominatorandtheresearcherlookingforit.Clearly,apre-denedmodelmightguidearesearcherhere.Additionally,asJonesandJames (1979)assert,theremightexistbotha``centralcoreofdimensionsaswellasspecicdimensionsapplyingtosomeparticularsituation.Theextenttowhichthisistrue,orthatparticularculturalmanifestationsaresimplylocalvariants ofacentralcoreisstillopentoinvestigation.4.Itissomethingthatis sharedby(groupsof)people(e.g.DeCocketal.,1986;Hofstede,1986;Schein,1992).Cultureissomethingthatismutualandreci-procal.Consequently,itisholistic(e.g.Hofstede,1991)orreferstomolarper-ceptions(Schneider,1975).Cultureisasynergisticaggregatecomposedof severalparts.Somewouldarguethatitisawholethatismorethanthesumofitsparts.Thisattribute,however,highlightsthefactthatnotonlythosecon-stitutingpartsofculturehavetobedened,butalsothecompositionrule whichbindsthemalltogether(cf.Glick,1985).Others(e.g.DeCocketal.,

1986)considerculturetobeanintegrativeconcept,contributingtoahelicopter visionthatmanagementcravesfor.Thischaracteristicisthebasisforassuming multiplecultureswithinalargeorganisation,inthatsuchanorganisationcan bedividedintodivisions,departments,units,etc.,thatwillallhavedeveloped theirownculture.DeCocketal.(1986,p.6and7)mention6levels:nationalculture,corporateculture,organisationalculture,departmentalculture,groupcultureandpsychologicalclimate.Morefundamental,however,isthecon-siderationofdistinguishingculturesandmakingstatementsaboutthesedif-ferences.Again,thischaracteristicdrawsattentiontothequestionofwhat makesupaculture.Schein(1992,p.14)argues``[

...]behavioralregularityshouldnot,therefore,bethebasisfordeningculture.And,``whenwe observebehavioralregularities[ataparticularinstance],wedonotknow whetherwearedealingwithaculturalmanifestation.Bymakingthisinitialexceptionforbehaviour,Scheinwantstoprevent``behavioralregularitieselicitedbysituationalcharacteristicsbeingconsideredmanifestationsofcul-ture.Theissueofwhatconstitutesagroupshouldnotbeoverlooked.For instance,acommonawarenessorunderstandingbetweenafewpeoplecannot beconsideredamanifestationofasub-culture.ThisisalsowhatScheinis aimingatabove.WithregardtogroupsScheinstressestheimportanceof stablemembership,commonhistory,sharedlearningandleadership.Thisissue willbetakenuplater,whenthetopicofaggregationisdiscussed.5.Itconsistsofvariousaspects;thismeansthatseveral,di*erentculturesorcli-matescanbedistinguishedwithinanorganisation,e.g.a``serviceclimate

(Schneider,1975),a``creativeclimateor``innovativeclimate(Ekvall,1983) orasafetyculture.Thesedistinctionshaveonlybeenmadeforanalyticalor practicalreasonstonarrowtheconceptandthusmakeitmoretangible.6.Itconstitutespractices;thischaracteristicissuppliedbyHofstede(1991).Hediscussesorganisationalcultureprimarilyinrelationtonationalculture.Hofstede, butotherauthorsaswell,conceivesculturesashavingmultiplelayers 3,not 3Theselayersshouldnotbeconfusedwiththedimensionsmentionedabove.F.W.Guldenmund/SafetyScience34(2000)215+/-257 223 unlikethelayersofanonion.Ateachoftheselevels,culturehasitsmanifes-tationswhichcanbestudiedseparately.Hofstedelocatesnormsandvaluesat thecentralcore.Hisnextlayerconsistsofrituals,thefollowingofheroeswhilsttheouterlayerconsistsofsymbols.Inconsideringorganisations,onlythelastthreelayers-rituals,heroesandsymbolsarerelevant,accordingtoHofstede.

Hecallsthesethreelayers``practicesincontrasttothenormsandvaluesofthe core.Thesepracticesaremoreeasilychangedthanthenormsandvalues,while themoreoutwardalayerissituated,themoresupercialitis.Normsandvalues arelearnedduringchildhoodthroughparentalupbringingandschoolingand remainrelativelystableduringtherestofourlives.Thischaracteristicalsoimplies thatcultureislearned.Howeverobvious,thisfacto*ersamajorjusticationforcontemporarycultureresearchinthatitexplainsthequestforculture'sin-uences, ingredientsandconsequences.Wewishtoin-uenceandchangeit.Althoughauthorsarerelativelyconsensualaboutthegeneralorderingofthelayers,thereisconsiderabledisagreementaboutwhatthedi*erentlayersmight encompass(Table2).Schein(1992)iscarefulininterpretingthemeaningofthe outerlayers,whichisre-ectedinhisphrasing,i.e.espousedvaluesandartefacts ,herebyclearlyindicatingthatwhatisseenandheardisnotalwaysatrueexpressionofculture.Scheinisthereforeveryreluctanttocountbehaviourasaculturalexpressionperse.Healsoremovesvaluesfromthecore,whichhe replaceswithbasicassumptions.Hence,whatseemstobethecoreofmost authors'onionsisspreadovertwolayersinSchein's 4.Anyothermanifestationofcultureis,forhim,anartefact,whereastheotherauthorsmakeseveraldis-tinctionswithinthoseartefacts.Table2LevelsofcultureReferenceCentralcoreLayer1Layer2Layer3DealandKennedy(1982)ValuesHeroesRitesandritualsCommunication networkHofstede(1986,1991)ValuesRitualsHeroesSymbolsSandersandNuijen(1987)ValuesandprinciplesRitualsHeroesSymbolsSchein(1992)Basicunderlyingassumptions Espoused values AftefactsVanHoewijk(1988)FixedconvictionsNormsand valuesMyths,heroes,symbols,storiesCodesofconduct, rituals,procedures 4ItisstressedagainthatHofstede'sonionisbasedonhisresearchintonationalcultures.Withregardtothebasicassumptionsoforganisations,thenormsandvaluesthatdistinguishnationalculturesareobviouslyfarmoresubstantial.IthereforeagreewithHofstedethatSchein'sbasicassumptionsareless

`basic'thannationalnormsandvalues.However,IalsoagreewithScheinthatwithinorganisationscer-tainbeliefsaremorepervasivethanHofstede'spractices.

224F.W.Guldenmund/SafetyScience34(2000)215+/-257 7.Itisfunctional;thisattributeisdiscussedbySchneider(1975)butisalsoimpliedbyHofstede(1991)andSchein(1992).Cultureprobablyclimatealso(SafetyResearchUnit,1993)isfunctionalinthesensethatitsuppliesaframeofreferenceforbehaviour.Schein(1992)considersculturetobetheproductofadaptive(orexternal)andintegrative(orinternal)processesofa group,steeredbyitsleader.Asimpleandwell-knowndenitionof(organisa-tional)culturereads,``Thewaywedothingsaroundhere,whiche*ectively capturesthisfunctionalaspect.Overall,organisationalcultureisarelativelystable,multidimensional,holis-ticconstructsharedby(groupsof)organisationalmembersthatsuppliesaframeofreferenceandwhichgivesmeaningtoand/oristypicallyrevealedincertainpractices.2.3.TheconceptualisationoforganisationalcultureandclimateOrganisationalcultureandclimatearecomplexconcepts.Guion(1973)declares,``[t]heconceptoforganizationalclimateisundoubtedlyimportant,butitalsoseems tobeoneofthefuzziestconceptstocomealonginsometime``(p.121).Glick(1985)actuallytalksabouta``conceptualmorass(p.601)andstatesthat``[organisational]climateisagenerictermreferringtoaclassofdimensionsthatmanyhavearguedis sobroadanddiverseastomaketheconceptuseless(p.605).Douglas(citedinDe Cocketal.,1986)writes:``Cultureisablankspace,ahighlyrespected,empty pigeonhole.Scheinstatesintheprefaceofhis1992book(p.xi):``Theconcept[of organisationalculture]ishardtodene,hardtoanalyzeandmeasure,andhardto manage.Healsomentionsthefollowingusesofthetermorganisationalculture(p.

8*.):``observedbehavioralregularitieswhenpeopleinteract(language,customsandtraditions,rituals),groupnorms,espousedvalues,formalphilosophy,rulesofthe game,climate,embeddedskills,habitsofthinking/mentalmodels/linguisticparadigms, sharedmeaningsand``rootmetaphorsorintegratingsymbols,toillustratethefactthatbehindthetermculturealotofdi*erentmeaningsarehiding.Organisationalclimatewasstudiedinitiallyasacausalfactorin-uencingjobper-formanceandsatisfaction(e.g.FriedlanderandMargulies,1969;PayneandPheysey, 1971)andwasestablishedthroughthemeasurementofindividualperceptionsof attributesofthatclimate.Inhis``note,Guion(1973)wonderswhetherclimateactuallyreferstoattributesoforganisationsorattributesofpeople.Toidentifygen-uinelyobjectiveorganisationalattributesheproposestopresentallmemberswith statementsaboutsuchattributesthatcanbeansweredbyasimpleyesorno.Thetruly descriptiveattributeswillallhaveaveryhighfrequencyofendorsement.Thisconfusion aboutwhetherorganisationalclimateisanorganisationalattributeoranindividual attributemadeJamesandJones(1974)suggestadistinctionbetweenorganisational climate(organisationalattribute)andpsychologicalclimate(individualattribute).DieterlyandSchneider(1974)conceiveorganisationalclimateasintermediate,``locationaryperceptions[...]whichhelpindividualsto``xorlocatethemselvesintheirlargerenvironmentpriortobehaving(p.317).MoreauthorshavestressedtheF.W.Guldenmund/SafetyScience34(2000)215+/-257 225 functionoforganisationalclimateorcultureasaframeofreferenceforthemembersoftheorganisationthatdirectsbehaviour(e.g.Schneider,1975;SafetyResearch Unit,1993).Consequently,members'behaviourwithintheorganisationbecomesmorepredictable,whichpossiblyalsoreducesanxiety(VanHoewijk,1988).Inlikemanner,culturefunctionsasadefencemechanism(Schein,1992)creatingbothsta-bilityandcontinuitywithintheorganisation(VanHoewijk,1988).Accordingly, organisationalculturenotonlyfunctionsasaconceptualumbrellabutalsoasa

`real'umbrella,shieldingfromtheprecipitationoftheunknownortheunwanted.Overall,researchersdonotdisagreeonthegeneralfunctionoforganisationalcultureorclimateasapatterningconceptprovidingacoherentstructuretoorgani-sationallifeorcertainpartsthereof.However,thereisnorealconsensusonhowtodescribetheclimateorcultureofanorganisation,i.e.whatisitsbasicstructure,howcanitbetypiedandhowshoulditbedetermined?Thisleavesonetowonderwhyit issodiculttoobtainthatconsensus.Firstly,itmightbethatissuesrelatingtothecausesande*ectsoforganisationalculturehavebecomeintertwined.Thisrelatestothelayersofculturementioned earlier.Itispostulatedthatthecoreisexplanatoryfortheouterlayers.Whenthese layersareconfused,onemixescauseswithe*ects,independentvariableswith dependent.

5However,thelayeredconceptofcultureintroducedabovegivesthepossibilitytodistinguishclimatefromcultureintermsoftheselayers.Culturethen, wouldbebestassociatedwiththecoreorSchein's``basicassumptions.Thenext layerwouldbeculture'sprimarymanifestationorclimate.Secondly,thereseemstoexistacertaintensionbetweentheholisticcharacteristicofcultureandclimateandthereductionisticapproachofmostresearchers.

Researchersfromsociologyora(social)psychologicalresearchtraditionareinclined toassumethatagivencultureorclimatecanbedescribedbyalimitednumberof dimensions.Theresearchobjectivebecomestouncoverthatstructure,whichisusuallyaccomplishedbyaquestionnairesurvey.Thestructureofthecultureorcli-matefollowsthenfromanalysisofresults.Obviously,otherapproachesarecon-ceivablebutalsootherwaysofrepresentingculture.Forinstance,thereisthe unresolveddebateofwhetheranorganisation hasacultureor isaculture.Further-more,itispossibletodepictcultureasaseparateentitywithinanorganisation-usuallyexistingbesideorganisationalstructureandprocesses-orasanaspectsys-tem,permeatingthewholeoftheorganisation(e.g.Frissen,1986).Moreover,itisof majorsignicancewhetheroneconsidersorganisationalcultureacollectionofobservablepractices(e.g.Hofstede,1991),anitesetofconsciousattitudes(e.g.JonesandJames,1979)orasmallamountofunconsciousbasic assumptions(e.g.Schein,1992).Clearly,suchdivergingviewswillresultindi*erent researchquestions,paradigms,methodsandoutcomes.Thirdly,thereistheissueofthelevelofaggregation.Severalauthorshavetriedtoshedsomelightonthisaspectoforganisationalculture(e.g.Guion,1973;Jamesand Jones,1974;JonesandJames,1979).Itisquestionedtowhatextentindividual 5Itisacknowledgedthatthisisarathertheoreticaldistinctionwhichmightbehardtosubstantiateinpracticewheresuchrelationshipsaremoreinteractiveanddynamic.

226F.W.Guldenmund/SafetyScience34(2000)215+/-257 measurescanbeusedtosaysomethingaboutorganisationallevelshigherthantheindividualone.Clearly,thisisanissueofgreatimportancebecause,veryoften, aggregatedindividualmeasures,fromquestionnairesurveysforinstance,areusedtosaysomethingaboutthefullorganisationorcertainpartsthereof.Or,asJonesandJames(1979)stateintheirstudyofUSNavyenlistedpersonnelonvariousships:

``[...]aggregationsofsuchdatacarrythepotentialforerroneousinference(p.205).Especiallyso,when``perceptionsarecombinedacrossgroupsofincreasinglyhet-erogeneouscontextorstructure(p.207).Theirstudyenabledthemtoaggregate theirmeasuresobtainedatindividualandhigherlevelstoorganisationally meaningfulgroupsorunitslikeship,divisionanddepartment.Aggregationto representship-wideordepartmental-wideconditionsdidnotappearwarrantedbutaggregationtodivisionalorfunctionallevellikeNavigation,MaintenanceandRadioCommunicationdid.Suchstudiesshowthatseeminglyobviousaggrega-tionallevelswithinorganisationsmightnotbesohomogenousinpractice.Otherauthors(e.g.Guion,1973;James,1982;Glick,1985)havetriedtodenecriteriaforthedegreeofhomogeneitythatjustiesaggregation.Guion(1973,p.

124)proposesahighlysignicantamountofagreementordisagreementwithinthe organisationstudiedonasetofdichotomousquestions.BothGlick(1985,p.607*.)

andJames(1982,p.221*.)proposeotherindiceslike Z 2andmodicationsthereof,whicharesuppliedbyone-wayanalysisofvariance(ANOVA)designs.Overall,the levelofaggregationand``aggregationbias(James,1982,p.225)areimportant methodologicalissuesthatmeritseriousattentionandthatcouldbethecauseof someoftheproblemsencounteredinorganisationalcultureandclimateresearch.Someoftheseissueswillbetakenupagainwhenresearchonsafetycultureandsafetyclimateisdiscussed.Thisdiscussionwillbeconductedunderthefollowingheadings:

denitionofsafetycultureandclimate,dimensionalityofbothconstructs,models underlyingtheseconstructs,levelofaggregationandthenatureofsafetyculture.3.SafetyclimateandsafetycultureTheearliestlocatedpaperonsafetyclimateisKeenanetal.(1951).Thisstudywasbasedonintrospectiveratingsfromprimaryindividualsinanautomotiveplant.Subsequently,theoryandresearchparadigmshaveimprovedbutnottotheextent thatacomprehensivetheoryonsafetycultureexists,northatameasurementapproachhasbeendevelopedthathasunanimouspreference.3.1.DenitionAlthoughfairlyeasilygivenusuallyitisjustonelinethedenitionofahypotheticalconstructsetsthestageforensuingresearch,i.e.itisthebasisforhypotheses,researchparadigmsandinterpretationsofthendings.Itdemarcates theboundariesoftheconceptandfocusestheresearch.Denitionscanbeexplicitorimplicit,thelatterleavingmuchmoreroomforinterpretations.DenitionsofsafetycultureandclimatearelistedinTable3.F.W.Guldenmund/SafetyScience34(2000)215+/-257 227 Table3Denitionsofsafetyclimateandsafetyculture ReferenceDenitionofsafetyculture/climateZohar(1980)Asummaryofmolarperceptionsthatemployeesshareabouttheirworkenvironments(safetyclimate)Glennon(1982a,b)Employees'perceptionsofthemanycharacteristicsoftheirorganisationthathaveadirectimpactupontheirbehaviourtoreduceoreliminatedanger(safetyclimate)and, safetyclimateisaspecialkindoforganisationalclimateBrownandHolmes(1986)Asetofperceptionsorbeliefsheldbyanindividualand/orgroupaboutaparticularentity(safetyclimate)Lutness(1987)Notexplicitlystated(safetyclimate)CoxandCox(1991)Safetyculturesre-ecttheattitudes,beliefs,perceptions,andvaluesthatemployeesshareinrelationtosafety(safetyculture)DedobbeleerandBe

Âland(1991)Molarperceptionspeoplehaveoftheirworksettings(safetyclimate)InternationalSafetyAdvisoryGroup(1991)Safetycultureisthatassemblyofcharacteristicsandattitudes inorganizationsandindividualswhichestablishesthat,asan overridingpriority,nuclearplantsafetyissuesreceivetheattention warrantedbytheirsignicance(safetyculture)Pidgeon(1991)Thesetofbeliefs,norms,attitudes,roles,andsocialandtechnicalpracticesthatareconcernedwithminimisingtheexposureof employees,managers,customersandmembersofthepublictoconditionsconsidereddangerousorinjurious(safetyculture)Ostrometal.(1993)Theconceptthattheorganisation'sbeliefsandattitudes,manifestedinactions,policies,andprocedures,a*ectitssafetyperformance(safetyculture)SafetyResearchUnit(1993)Notexplicitlystated(safetyclimate)CooperandPhilips(1994)Safetyclimateisconcernedwiththesharedperceptionsandbeliefsthatworkersholdregardingsafetyintheirworkplace(safetyclimate)Geller(1994)Inatotalsafetyculture(TSC),everyonefeelsresponsibleforsafetyandpursuesitonadailybasis(safetyculture)Niskanen(1994)Safetyclimatereferstoasetofattributesthatcanbeperceivedaboutparticularworkorganisationsandwhichmaybeinduced bythepoliciesandpracticesthatthoseorganisationsimpose upontheirworkersandsupervisors(safetyclimate)Coyleetal.(1995)Theobjectivemeasurementofattitudesandperceptionstowardoccupationalhealthandsafetyissues(safetyclimate)Berends(1996)Thecollectivementalprogrammingtowardssafetyofagroupoforganisationmembers(safetyculture)Lee(1996)Thesafetycultureofanorganisationistheproductofindividualandgroupvalues,attitudes,perceptions,competencies,and patternsofbehaviourthatdeterminethecommitmentto,and thestyleandprociencyof,andorganisation'shealthandsafety management(safetyculture)(continuedonnextpage) 228F.W.Guldenmund/SafetyScience34(2000)215+/-257 Mostdenitionsareveryglobalandthereforehighlyimplicit.TheACSNI(1993)denitionemployedbyLeeisthemostexplicit,outliningmostoftheassumedcontentsofsafetyculture.Ofthe16denitionsgiveninTable3,nineareaboutsafetyclimateandsevenaboutsafetyculture.Ninementionorganisationmember's perceptionswhereassixdenitions(also)refertobeliefsandsix(also)toattitudes.

Fiveoftheseareaboutsafetyculture.Roughly,perceptionsaremoreassociated withclimatewhereasattitudesareconsideredtobeapartofculture.Theholisticaswellasthesharedaspectofcultureandclimatearestressedinmostdenitionswithtermslike``molar(Zohar,1980;DeDobbeleerandBe

Âland,1991),``shared(CoxandCox,1991;CooperandPhilips,1994;Cabreraetal.,1997),``summary(Williamsonetal.,1997),``group(BrownandHolmes,1986;Berends,1995a,b,1996;Lee,1996),``set(Pidgeon,1991,1997,1998),``assembly(Interna-tionalSafetyAdvisoryGroup,1991),``employees'perceptionsor``organisation's beliefsandattitudes(Glennon,1982a,b;Ostrametal.,1993).Theobjectoftheseperceptions,beliefsorattitudesareoftenidentiedwith``workenvironments(Zohar,1980;DeDobbeleerandBe

Âland;Cabreraetal.,1997)orsimplyspeciedwith``safety(CoxandCox,1991;CooperandPhillips,1994; Berends,1995a,b,1996;Coyleetal.,1995;Williamsonetal.,1997).Sometimesalso,theseobjectsaremorecomplexlike``organisationalcharacteristics(Glennon,1982a,b),``actions,policies,andprocedures(Ostrometal.,1993)or,equivalently,

``organisationalsafetypolicies(Cabreraetal.,1997)orevenmoreabstractlike

``entity(BrownandHolmes,1986)or``attributes(Niskanen,1994).Thecharacteristics``constructand``dimensionalityofcultureandclimatedescri-bedinSection2.2areeitherimplicit(CoxandCox,1991;Ostrometal.,1993;Cooper andPhilips,1994;Coyleetal.,1995;Williamsonetal.,1997)orexplicit(Glennon, 1982a,b;BrownandHolmes,1986;Niskanen,1994;Lee,1996)inmostdenitions.Thee*ectofclimateorcultureontheorganisationanditsmembersissometimesstatedaswell(Glennon,1982a,b;Ostrometal.,1993;CooperandPhilips,1994; Geller,1994;Pidgeaon,1991,1997;Lee,1996).Totheextentthattheparticulardenitionhasfocusedresearch,Table1mightyieldananswer.Undertheheading``GoalinTable1thegoalsdenedexplicitlyby theresearchersaresummarised.Mostresearchershaveformulatedquitepractical goals,althoughtheobjectivesofsome(BrownandHolmes,1986;DeDobbeleerand

BeÂland,1991;Niskanen,1994;Coyleetal.,1995)alsohaveamoretheoretical-a-vour,whichmightbetraytheabsenceofaparticularassignmentfromacompany.

Hence,mostresearchershaveexecutedtheirresearchwithregardtocertainquestionsTable3(continued)ReferenceDenitionofsafetyculture/climateCabreraetal.(1997)Thesharedperceptionsoforganisationalmembersabouttheirworkenvironmentand,moreprecisely,abouttheirorganisationalsafetypolicies(safetyclimate)Williamsonetal.(1997)Safetyclimateisasummaryconceptdescribingthesafetyethicinanorganisationorworkplacewhichisre-ectedinemployees' beliefsaboutsafety(safetyclimate)F.W.Guldenmund/SafetyScience34(2000)215+/-257 229 Table4Overviewoftheamountofquestions,surveyedpopulationanddimensionsofsafetycultureandclimateresearches ReferenceNo.ofquestions/instrumentPopulationTypeofanalysisClimate/culturedimensionsZohar(1980)40,questionnaireisadministeredduring interview20Israelianindustrialorganisations (steel,foodprocessing,chemicaland textileindustry);400respondentsExploratory-Importanceofsafetytrainingprogrammes-Managementattitudestowardssafety

-E*ectsofsafeconductonpromotion-Levelofriskatworkplace-E*ectsofrequiredworkpaceonsafety

-Statusofsafetyocer

-E*ectsofsafeconductonsocialstatus

-StatusofsafetycommitteeGlennon(1982a,b)68,self-administeredquestionnaire(SAQ)LinemanagersfromeightAustralian companies(bauxite,mining, sawmillingandloggingmetalrening, petroleumrening,cement manufactureandgeneralengineering andmanufacturing);198respondentsNoformal testing-Perceivedin-uenceofsafetyandhealth

legislation

-Perceivedcorporateattitudetosafety andhealth

-Perceivedorganizationalstatusofsafety advisoryocer

-Perceivedimportanceofsafetyandhealth

training

-Perceivede*ectivenessofencouragement (vs.discipline)inpromotingsafety-Perceivede*ectofdepartmental/sectionsafetyrecordonpromotion

-Perceivedrisklevelofworkplaces

-Perceivedstatusofsafetytargetsrelative toproductionpressuresBrownand Holmes(1986)40,SAQ10Americanmanufacturingandproducecompanies;425respondentsConrmatory-Employeeperceptionofhowconcernedmanagementiswiththeirwell-being

-Employeeperceptionofhowactive managementisinrespondingtothisconcern

-Employeephysicalriskperception(continuedonnextpage) 230F.W.Guldenmund/SafetyScience34(2000)215+/-257 Table4(continued)ReferenceNo.ofquestions/instrumentPopulationTypeofanalysisClimate/culturedimensionsCoxandCox(1991)18(+4),SAQEmployeesofanEuropeancompanyinvolvedintheproductionanddistributionofindustrialgasses; 630respondentsExploratory-Personalscepticism-Individualresponsibility

-Safenessofworkenvironment

-E*ectivenessofarrangementsforsafety

-PersonalimmunityDeDobbeleerand

BeÂland(1991)9,SAQ9constructionsites;272respondentsConrmatory-Management'scommitmenttosafety-Worker'sinvolvementinsafetyOstrometal.(1993)88,SAQEmployeesoftheDepartmentofEnergyinIdahoanditseight contractors;4000administeredNoformaltesting-Safetyawareness-Teamwork

-Prideandcommitment

-Excellence

-Honesty

-Communications

-Leadershipandsupervision

-Innovation

-Training

-Customerrelations

-Procedurecompliance

-Safetye*ectiveness

-FacilitiesSafetyResearch Unit(1993)65,SAQWorkersfromsteelandchemicalindustries;atotalof1475

respondentsExploratory-Management/supervisorsatisfaction(M1)-Management/supervisorknowledge(M2)

-Management/supervisorencouragementandsupport(M3)-Management/supervisorenforcement(M4)

-Personalmanagementcontact(M5)

-Managementsupport:meetings(M6)

-Shop-oorsatisfaction(A)

-Shop-oorenvironment:hardware(B)

-Workgroupsupport/encouragement(C)

-Shop-oortraining(D)(continuedonnextpage)F.W.Guldenmund/SafetyScience34(2000)215+/-257 231 Table4(continued)ReferenceNo.ofquestions/instrumentPopulationTypeofanalysisClimate/culturedimensions-Globalselfsafety(E)-Meetings(F)

-Safeworkingprocedures(G)

-Safetyinformation(H)

-Safetyrepresentatives:practice(SR1)

-Safetyrepresentatives:authority(SR2)Cooperand Philips(1994)50,SAQPersonnelofapackagingproductionplant;374(pre)and187(post)

respondentsExploratory-Managementattitudestowardssafety-Perceivedlevelofrisk

-E*ectsofworkpace

-Managementactionstowardssafety

-Statusofsafetyocerandcommittee-Importanceofsafetytraining-SocialstatusofsafetyandpromotionNiskanen(1994)22(workers)and21(supervisors),SAQWorkersandmanagementin maintenance,constructionand centralrepairshops;1890(workers) and562(supervisors)respondentsExploratoryWorkers:-Attitudetowardssafetyintheorganisation

-Changesinworkdemands

-Appreciationofthework

-Safetyaspartofproductivework

Supervisors:

-Changesinjobdemands

-Attitudetowardssafetywithinthe organisation-Valueofthework

-SafetyaspartofproductiveworkGeller(1994)+/-+/-+/--Personi.e.knowledge,skills,abilities,intelligence,motives,personality-Behaviouri.e.complying,coaching,recognising,communicating,demonstrating activelycaring-Environmenti.e.equipment,tools,machines,housekeeping,heat/cold,engineering(continuedonnextpage) 232F.W.Guldenmund/SafetyScience34(2000)215+/-257 Table4(continued)ReferenceNo.ofquestions/instrumentPopulationTypeofanalysisClimate/culturedimensionsCoyleetal.(1995)30(organisation1)and32(organisation2)Workforceoftwoorganisations``involvedintheprovisionofhealth careandsocialservicestotheelderly

(incl.oce,nursingandsocialwork duties);340(org.1),540 (org.2)respondentsExploratoryOrganisation1:-Maintenanceandmanagementissues

-Companypolicy

-Accountability

-Trainingandmanagementissues

-Workenvironment

-Policy/procedures

-Personalauthority Organisation2:-Workenvironment-Personalauthority

-TrainingandenforcementofpolicyBerends(1996)34,SAQThreeindustrialorganisations(twochemicalprocessindustriesandone steelcompany);atotalof434

respondentsExploratory-Condenceinthearrangementsforsafety

-Compliancewithsafeworkingpractices-Perceived prioritygiventosafety-Ownactivee*ortputinsafetymatters

-CommunicationaboutsafetyLee(1996)172,SAQEmployeesatBritishnuclearindustry site;5295respondentsExploratorySafetyprocedures:-Condenceinthesafetyprocedures

-Safetyrules:

-Personalunderstandingofsafetyrules

-Perceivedclarityofsafetyrules

-Permittoworksystem:

-Condenceine*ectivenessofPTW

-GeneralsupportforPTW

-PerceivedneedforPTW

Risks:

-Personalcautionoverrisks-Perceivedlevelofriskatwork-Perceivedcontrolofrisksintheplant

-Personalinterestinjob

-Jobsatisfaction:(continuedonnextpage)F.W.Guldenmund/SafetyScience34(2000)215+/-257 233 Table4(continued)ReferenceNo.ofquestions/instrumentPopulationTypeofanalysisClimate/culturedimensions-Contentmentwithjob-Satisfactionwithworkrelationships

-Satisfactionwithrewardsforgoodwork

Participation/ownership:

-Self-participationinsafetyprocedures

-Perceivedsourceofsafetysuggestions

-Perceivedsourceofsafetyactions

-Perceivedpersonalcontroloversafety

Design:

-Satisfactionwithdesignofplant

-Training:

-SatisfactionwithtrainingSelection:

-Satisfactionwithsta*suitability Cabrera etal.(1997)69,SAQEmployeesofseveralcompaniesatthreeEuropeanairports(groundhandlingdivisionsfromfourairlines,onefuelcompany,twoairport authorities);totalling389respondentsExploratory-Organisationalempasisonsafety-Communicationchannelaboutsafety-Safetylevelperceivedonthejob-Feedbackperformanceonsafety

-Specicstrategiesofaccidentprevention Williamson etal.(1997)67,SAQ7workplaces,coveringheavyandlightindustryandoutdoorworkers, totalling660responsesExploratory-Personalmotivationforsafety-Positivesafety+/-Practice

-Riskjustication+/-Fatalism

-Optimism 234F.W.Guldenmund/SafetyScience34(2000)215+/-257 posedtothembyone,ormore,companiesorinstitutions,whichhasgiventheirresearchaparticularfocus,e.g.implicationsofsomesafetyclimateorculture (Zohar,1980;SafetyResearchUnit,1993;Carbreraetal.,1997;Lee,1996)indicatorofsafetyclimateorculture(Ostrometal.,1993;Niskanen,1994;Berends,199a,b,1996;Williamsonetal.,1997)ordevelopmentofamethodforimprovement (Glennon,1982a,b;Lutness,1987;CoxandCox,1991;Ostrometal.,1993;Cooper andPhilips,1994;Lee,1996).Noneoftheresearchers,however,seemstohavea pre-denedtargetpopulationinmind.InTable4thesurveyedpopulationsare enumerated;ascanbeseenfromTable4bothhomogeneousandheterogeneous populationswithvarioustypesofoccupationsareusedinthesestudies.Insummary,mostresearchershavedenedeithersafetyclimateorsafetycultureintheirpublicationsaswellaswhytheywanttoexploreit.Thesedenitionscontainsome ormostofthecharacteristicsdenedearlier.Thepurposeofthesestudiesisoftenquite practical,althoughtheoreticalmotivesarealsoputforward.Theaccentoneitherper-ceptions,beliefsorattitudesaswellasoneoranotheraggregate(e.g.``molar,``group,

``summary)suggestsaself-administeredquestionnaireresearchparadigm.Table4 showsthatthisisbyfarthemostcommonapproach.Ludborzs(1995)andKennedy (1997)haveoptedforalternativeapproaches,i.e.anauditandaSCHAZOP(Safety CultureHAZOP),respectively.Theselatterapproachesarediscussedbelow.Thisparticularoperationalisation,i.e.aself-administeredquestionnaire,generallyfollowsacharacteristicpathofdevelopment.First,onedemarcatestheparticular areaofinterest,whichisthenthoroughlyinvestigated,mostlythroughaliterature survey.Thisusuallyresultsintheidenticationofaspectsrelevantfortheareaof interest.Giventhefactthatmostresearchersfocusonbeliefs,perceptionsandatti-tudes,theserelevantaspectsarethentheobjectsofthosementalprocesses.With regardtotheseaspects,questionsareformulated,whicharethenpre-testedinapilot studyonarelevantpopulation.Ifthepilotstudygoessatisfactorily,theques-tionnairecanbedistributedamongthetargetpopulation.Theresultsofthissurveyarethensubjectedtocertainstandardanalysismethodslikefactoranalysis(FA)or principalcomponentsanalysis(PCA)(TatsuokaandLohnes,1988;Tabachnickand Fidell,1989),wherelinearrelationsbetweenthequestionsorvariablesareassumed, ortechniqueslikeHOMALSorPRINCALS(VandeGeer,1993a,b),wheresuch linearityisnotassumed.Theseanalysesresultinfactors,principalcomponentsor dimensions,whicharethesubjectofthenextsection.Ascanbegarneredfromthecolumnlabelled``SourceinTable1,thisistheapproachfollowedbymostresearchers,i.e.manystartthewholeprocessfromscratch again,althoughthe1980Zoharstudyhasinspiredsomeresearchersinmore(Brown andHolmes,1986;CooperandPhilips,1994)orlessprofoundways(DeDobbeleerand

BeÂland,1991;SafetyResearchUnit,1993;Coyleetal.,1995;Cabreraetal.,1997;Wil-liamsonetal.,1997).Theresultsoftheseandmoreinvestigationsarereviewednext.3.2.DimensionalityCommonly,socialscienticconstructsaremulti-dimensional.Forinstance,aconstructlikeintelligencemightnotonlyshowintheperformanceonparticularF.W.Guldenmund/SafetyScience34(2000)215+/-257 235 arithmetictestsbutalsoonvisuo-spatialtasksoroncertainlanguageexercises.Therangeofactivitieswhichareshowntobein-uencedmightevenbecomesolargethat theconstructissubdividedintoseparatetypeslikearithmeticintelligenceorsocialintelligence.Thisisnotonlytrueforsocialscienticartefacts,butappliestothephysicalworldaswell.Forinstance,anyobject'scolourcanbedescribedalongthe threedimensionsoftheprimarycoloursred,yellowandblue.Cultureandclimatehavebeencharacterisedaboveasmulti-dimensional.AnalysistechniquessuchasFA,PCA,PRINCALSandHOMALSproducesuchdimensions whentheyareusedforanalysingsurveyresults.Thesedimensionsaretheresultof inter-andintra-respondenttendenciestoevaluatecertainquestionsinasimilarway.

Suchtendenciesarecalledcorrelations,i.e.whentwoquestionsareansweredoverallinasimilarway,itissaidthatthesequestionscorrelate.Itisassumedthen,thatthesequestionshaveacertainrelationship,forinstancebecausetheyrefertoa similarobject.Thisrelationshipmightbeobviousbutthisdoesnothavetobethe case.Forinstance,Hofstedeperformedasecondarydataanalysisoninformation collectedamongemployeesatIBM,originallycollectedtodeterminetheirattitudes (Hofstede,1991,p.251).Withthesedata,however,hewasabletoproducehis famous4-Dmodel.AndSchumanandPresser(1981,p.153*.)describeacorrela-tionbetweenevaluationsofeconomicpolicyandanon-existinglaw,whichtheycanonlyexplainwithanoverall(lackof)condenceexpressedinthegovernment.InTable4resultsfromtheanalysesperformedonthesurveyresultsaresum-marised.Atrstsight,thereisnotmuchcorrespondencebetweentheresearchesreported.Forone,thisisbecausetheresearcherhasconsiderablefreedomtolabelher orhisdimensions.Obviously,mostresearchersdidnothavetheneedtoconnectto previousresearchintermsoftheirdimensions 6.Moreover,thenumberofdimensionsfounddi*ersenormously.Theserangefromtwo(DeDobbeleerandBe

Âland,1991)to16(SafetyResearchUnit,1993)oreven19(Lee,1996,whentakenliterally).Althoughthislatterndingmightseemstriking,afewexplanationscanbeputforwardtoexplaintheseresults.AscanbeseenfromTable4thesurveyswerecar-riedoutindi*erentorganisations,rangingfromindustry(Zohar,1980;Glennon, 1982a,b;BrownandHolmes,1986;CoxandCox,1991;SafetyResearchUnit,1993; CooperandPhilips,1994;Berends,1995a,b,1996;Williamsonetal.,1997)tocon-struction(DeDobbeleerandBe

Âland,1991;Niskanen,1994)toenergy(Ostrometal.,1993;Lee,1996)toairports(Cabreraetal.,1997)andtohealthcareandservice (Coyleetal.,1995).Obviously,employeeswithintheseorganisationshavequitedi*erentobjectsfortheirattitudes.Additionally,whatisdistinguishedbysomeisconsideredsimilarbyothers 7,obviouslyresultinginlesscomplexattitudestructures,i.e.fewerdimensions(EaglyandChaiken,1993,p.89*.).CoxandFlin(1998) arguethatinstrumentsdevelopedinonedomain(oil)maynotgeneralisetoothers 6Interestingly,moststudiesreviewedhereareexploratory.OnlytheresearchreportedbyBrownandHolmes(1986)andDeDobbeleerandBe

Âland(1991)areconrmatorystudies(Table4).Bothstudiesfailedtoconrmfactorstructuresthathadbeenfoundpreviously 7Forinstance,attheshop-oor``managementmightbeeverybodyintheocebuildingwhereasintheocespeoplemighthaveamorenuancedviewof``management.

236F.W.Guldenmund/SafetyScience34(2000)215+/-257 (construction).Interestingly,evenanattemptaimedatreplicatingapreviouslyfoundfactorstructureinasimilarkindoforganisationfailed(Coyleetal.,1995).However,additionalmethodologicalissuesmightbeimportanthere.Forinstance,thetechniquescommonlyusedFAorPCA(TatsuokaandLohnes,1988; TabachnickandFidell,1989)areneverquestionedfortheirapplicability.The appropriatenessofFAorPCAcouldbequestionedwithregardtotheassumed measurementlevelofthedata.Althoughforquestionnairedataanintervallevelof measurementisusuallyassumed,thisassumptionmightnotbeappropriateand couldresulttheoreticallyindimensionswhicharenotactuallythere.OnlytheSafety ResearchUnitappliesadi*erenttechniquecalledSmallestSpaceAnalysis(SSA; Guttman,1968).Althoughthistechniqueusesamathematicaltransformationcom-parabletotheothertechniquesnamelysingularvaluedecompositionorSVD(GreenandCarroll,1978)thenalapproachisquitedi*erent.Hence,thedimen-sionsoftheSafetyResearchUnitinTable4arenotdimensionsinthesensethatthe othersare,theyaremoreappropriatelyreferredtoasscales,toavoidconfusion.WhenaFAorPCAhasbeenperformed,thenalsolutionisoftenrotatedtofacilitateinterpretation(e.g.TabachnickandFidell,1989).Thisrotationisnearly alwaysorthogonal,meaningthattheinitialsolutionofuncorrelateddimensionsis preserved.However,thisdoesnothavetobethecase;theattitudeobjectsre-ectedinthedimensionsmightbeunrelatedintheanalysisbutdonothavetobesoinreality.Itshouldbepointedoutthatthemethodologicalpointsmadeabove,are meretheoreticalconsiderations.However,inmostofthepapersreviewedthemeth-odologicalargumentationifitisdiscussedatallisnotparticularlystrong, whichiswhytheseconsiderationsaremadehere.Thereisanotherinterestingdiscussionrelatedtotheissueofmethodology.ResearchbyKerlinger(citedinEaglyandChaiken,1993)hasshownthatcertaindimensionsare notbipolarbutunipolar.Forinstanceadimensionlike``politicalorientationdoesnothave``conservatismand``liberalismatitsouterpolesbutisinsteadsplitintotwodimensions,onedenoting``conservatismandtheother``liberalism.Kerlingerfound thatconservatistsarenotsomuchopposedtotheidealsofliberalismbutrather indi*erenttotheseideals(EaglyandChaiken,1993,p.98).Thisimportantdistinction hasbeenobservedbyothersaswell(e.g.VanSchuurandKiers,1994).Inaddition,thelevelofaggregationmightplayanimportantpartheretoo.DeCocketal.(1986)arguethattheorganisationallevelatwhichthestudyisdirected andaboutwhichstatementswillbemadeshouldbeconsistentwiththeinstructionsandthequestioning.Withregardtothestudiesrevieweditisnotclearwhetherthisisalwaysthecase.ThelevelofaggregationwillbediscussedlaterinSection3.4.Despitethesemethodologicalconsiderations,arenamingandgroupingexercisemightyieldsomesolaceaswell.Thatis,onecoulddeneasmallsetofcommon denominatorstoclassifycomparabledimensionsunder.Forinstance,alldimensions re-ectingsafetye*ortsofmanagementcouldbeclassiedasManagement'sSafety Activity.Clearly,whenthedimensionsfoundinsafetycultureandclimateresearch arerenamedaccordingtothiscommonclassicationsystem,thetotalamountofdimensionswillreducesignicantly.Moreover,whenthenumberoftimesadimensionisfoundisalsotakenintoaccount,itwillbecomeobviousthatcertaindimensionsF.W.Guldenmund/SafetyScience34(2000)215+/-257 237 arementionedmoreoftenthanothersare.Suchanindexmightserveasanindica-tionofimportanceorubiquitousness.Insummary,alotofdi*erentdimensionshavebeenfoundtounderliesafetycultureandclimate.Somemethodologicalargumentshavebeensuppliedtoexplainthisabun-danceandtosuggestalternativemethodologicalapproachesforapplicationinfuture researchonsafetycultureandclimate.Inaddition,whenmanyofthesedimensionsare relabelledtheirnumberissignicantlyreducedandmayalsoyieldsomeinsightintothe relativeimportanceorubiquitousnessofthesedimensions.Additionalresearchand/or secondarydataanalysisisneededtosubstantiatethesemethodologicalissuesandto shedmorelightontheirsignicance,theirconsequencesandpossiblesolutions.Decidingonthenumberofdimensionsandtheirlabellingisoftenfacilitatedwhenamodelhasbeenusedtopreparethequestions.Thenextsectionreviewsthemodels usedforsafetycultureandclimate.3.3.CausalmodelTheelementmissinginmanypublicationsonsafetycultureisanexplicit,theore-ticalmodeloutliningthemannerinwhichsafetycultureisthoughttobeembeddedinthewholeofanorganisation'spracticesandsystemstructure;Table1revealsthis undertheheading``Causalmodel.Ideally,thismodelshouldbeaboutthecause , the contentandtheconsequenceofsafetycultureorclimate.Ingeneral,itispossibletodistinguishtwotypesofmodels:(1)normativeorpre-scriptivemodels,whichseektodescribeandspecifysafetyclimateorcultureperse; and(2)descriptiveorempiricalmodels,whichattempttosummarisendingsfrom oneorseveralorganisationsstudied.TherstactualmodelofsafetyclimatefunctioningwasputforwardbyGlennon(1982a,b).Ine*ect,thisnormativemodeloutlinesthecause,contentandcon-sequencesofsafetyclimate,althoughinaveryglobalway(Fig.1).Fig.1.Glennon's(1982a,b)modeloforganisationalclimatefunctioning.

238F.W.Guldenmund/SafetyScience34(2000)215+/-257 Glennonoperationalisessafetyclimateastheperceptionoforganisationalreality,whichseemstosuggestakindofattitudemeasurement,butonlypartlybecauseperceptionsarenotidenticalwithattitudes.CoxandCox(1991)basedtheirmodelonworkdonebyPurdham(1984;citedinCoxandCox,1991).Thismodel(Fig.2)appearstobedescriptiveandtheaccom-panyingfactorstructureisgiveninTable4.Inthisstudy,safetycultureisprimarily discussedinthecontextofattitudestowardssafetyandtheirobjects,i.e.whathas beendenedassafetyclimateabove.Themodeldistinguishesseveralattitude objectshardware,software,people/livewareandrisks.Theattitudestowards hardwareandphysicalhazardsthough,werenotincorporatedintheirstudyanditis alsonotclearhowitisthoughtthattheya*ecttheotherattitudes.CoxandCox'smodelisnotworkedoutwell,buttheideaseemstobethatthemajorattitudestosafetywithinanorganisationaredirectedatfourcategoriesofobjects:1.hardware,i.e.safetyhardwareandphysicalhazards; 2.software,i.e.rulesandprocedures,legislation,safetymanagementandpolicy;3.people/liveware,i.e.allclassesofpeopleinvolvedlikeworkers,supervisors,management,safetycommittees,specialists,authorities,unions;and4.risks,i.e.riskybehaviouranditsregulation.Whentalkingaboutattitudestosafety,theobjectsoftheseattitudescouldalwaysbeclassiedwithinoneofthesefourmajorcategories.ThemodelunderlyingtheapproachtakenbytheSafetyResearchUnit(1993)isestablishedinthe``mappingsentence.Themappingsentencecontainsalltheaspectsor``facetsconsideredrelevantfortheissueunderstudy(seeShyeetal.,

1994,fordetailsonfacettheory).Consequently,thestartingpointoffacettheoryisFig.2.CoxandCox's(1991)suggestedarchitectureofattitudestowardssafety.F.W.Guldenmund/SafetyScience34(2000)215+/-257 239 normative,althoughthefacetstructureisusedtogenerateaquestionnaire.Sub-sequentanalyses,however,willeventuateinadescriptiveresult.Theactualmappingsentencehasthefollowingform:Theextenttowhichrespondent(x)reportsthat{People}{Attitudebehaviour}

{Locus}{Activity}{Context}under{Operatingconditions}

!{verymuch...notatall}.wherethebracketedwordsareparticularslotsforthefacetsmentionedinTable5.Based onthismappingsentencesome432(432233)questionscanbegenerated,whichcanbeevaluatedbyrespondentsonaseven-point(verymuch/not)atallresponse scale.Theirstudyresultedina16-scalesolution,arrivedatthroughSSA(Guttman,1968).SSAisnotsomuchaimedatanorthogonalsolutioninalowdimensionalCarte-sianspaceas,forinstance,isFAorPCA,butmoreatsomecongurationinalowdimensionalspace(seeBorg,1981,forexamplesofthesecongurations).Asalreadyremarked,thescalesfromthisstudyarethereforenotdimensionsandiftheyare,they areoblique,whichmeansthattheyarecorrelated.Inthisway,theSafetyResearchUnit identiesafewmajorcategoriesaroundwhichsafetyattitudesareformed.TheformulationofthemodelputforwardbyBerends(1995b)startedwithopen,unstructuredinterviewsnotunlikefreeassociationwithpersonnelatseveral companiesaroundtheissueofsafety.Recurringthemesorstatementsfromthe interviewsweregroupedbyseveralindependentjudgesintocategories.Theircorre-spondingcategoriesformedthebuildingblocksforthenalmodel.Inthismodel, twobroadclassesofstatementsunderlieallothercategories;normsandbeliefs.Normsaresubdividedintoindividual,interactionalandorganisationalnorms.Thesecategoriesarebrokendownagainintoseveralsub-categories.Beliefsontheother hand,areimmediatelybrokendownintosub-categories(Fig.3).Theremarksand statementscollectedineachofthesub-categoriesarethereuponreworkedintoquestions.

Inasubsequentsurveythemodelwasonlypartlyveried.TheFAyieldedmostly norm-factorswhilethebeliefs-factorswerenotconrmedinthestudy.Factors resultingfromthisstudyareshowninTable4.AtrulynormativeorbetterstillprescriptivemodelofsafetycultureisputforwardbyGeller(1994).Gellerdistinguishesthree``dynamicandinteractivefac-tors(p.18+/-19):Table5FacetsintheSafetyResearchUnit(1993)studyPeopleAttitudebehaviourLocusActivityContextOperatingconditions1.self1.knowsabout1.yourjobin particular1.passive1.preparations1.normal2.supervisor2.issatisedwith2.safetyin general2.active2.actions2.maintenance3.manager3.carriesout3.checks/revisions3.special4.workmates 240F.W.Guldenmund/SafetyScience34(2000)215+/-257 Fig.3.Berends'(1995b)safetyculturemodel.F.W.Guldenmund/SafetyScience34(2000)215+/-257 241 NORMS BELIEFS individual interactional organisational controllability of safety controllability by individual causes of accidents human nature results of safe working evaluation of situation passive active support communication design of environment controlling behaviour 1.person,i.e.knowledge,skills,abilities,intelligence,motives,personality;2.behaviour,i.e.complying,coaching,recognising,communicating,demonstrat-ingactivecaring;and3.Environment,i.e.equipment,tools,machines,housekeeping,heat/cold,engi-neering.Moreover,heputsforward10principlesthatformthefoundationforatotalsafetyculture.Through``veprocessesorinterventiondomainstheseprinciples shouldbeimplemented.Basically,Gellerappliesprinciplesofbehaviourismand sociallearningtheorytotheeldofsafety.Therelationshipbetweenallthecom-ponentsofhismodelisnotdened,noraretheyprioritised.Despitetheobviouslydi*erentapproaches,severalsimilaritiescouldbepointedout.Forinstance,especiallyboththeCoxandCoxandtheSafetyResearchUnitstudies focusonattitudes,butalsothestudiesbyBerendsandGelleryieldattitudeobjects.Thisisanappropriateplacetosaysomethingaboutattitudesandattituderesearchingeneral.Attitudesweredenedaboveas``apsychologicaltendencythat isexpressedbyevaluatingaparticularentitywithsomedegreeoffavorordisfavor

(EaglyandChaiken,1993,p.1).InFig.4theprocessesprecedingandtheresponsesresultingfromattitudesaredepicted(adaptedfromEaglyandChaiken,1993).Withregardtoattitudes,itis theoreticallypossibletoseparateantecedentsfromconsequences,althoughboth maybeofthesameorder.Perceptionsandbeliefsareonlyoneprocessorresult withinthismodel,namelyacognitiveone.Hence,neitherperceptionsnorbeliefsare attitudes.Attitudesarealwaysdirectedatanobject,i.e.theentityinthedenitionabove.Thisentitycouldbevirtuallyanything,aslongasitissomehowdiscriminable,e.g.

abstractobjectslikepoliciesorsafety;concreteobjectslikepersonalprotectiveequipmentorreextinguishers;behaviourslikerisktakingorruleviolations.MostFig.4.Simpleattitudemodel(EaglyandChaiken,1993).

242F.W.Guldenmund/SafetyScience34(2000)215+/-257 ofthemodelsdescribedabovesuggestsuchattitudeobjects.Forinstanceallmodelsincludeapeoplecategory.UsingCoxandCox'scategoriesofattitudeobjects,i.e.

hardware,software,peopleandrisksitwouldbepossibletolinkBerends'andGeller'smodelsbutnotthemodelputforwardbytheSafetyResearchUnit.Interestingly,thedemarcationbetweennormsandbeliefsinBerends'modelcouldbetracedbacktothedistinctionmadeearlierbetweendescriptiveanda*ectivestatementsaboutorganisationalclimate,althoughinhismodelthisdistinctionis notworkedoutthisway.Anotherwayoflookingatthisdistinctionwouldbein Schein'stermsoflevelsofculture,wherethecategorydenedasnormswould pertainto`espousedvalues'andthebeliefscategorythenwouldcorrespondtohis

`basicassumptions'.Measuringnorms,i.e.`espousedvalues'throughaself-admi-nisteredquestionnairewouldbefeasibleaccordingtoSchein,buttryingtomeasurebeliefsinthisway,i.e.`basicassumptions'wouldbeboundtofail,whichisexactly whathappenedinBerends'study.Allinall,themodelsonsafetycultureareunsatisfactorytotheextentthattheydonotembodyacausalchainbutratherspecifysomebroadcategoriesofinterestand tentativerelationshipsbetweenthose.Inmyproposedterminology,atbesttheyare aboutthe contentofsafetyclimate,i.e.theobjectsofsafetyattitudes.However,thisisnottosaythattheissueisnotalsoaddressedelsewhere.Forinstance,thedominomodelunderlyingtheInternationalSafetyRatingSystem (versionV)positionsthemeasurementofsafetyattitudes,i.e.safetyclimate,atthe frontofsafetyaudits.Withinthismodel,safetyattitudesaretheprimarycause underlyingallincidents.Fromthedomainofriskanalysisamodelhasbeenput forwardthatrepresentssafetyculture,alongwithotheraspects,asanall-pervading in-uence(TuliandApostolakis,1996).Nevertheless,safetycultureitselfisstilliso-latedand`uncaused'.Withregardtotheselasttwomodelsitcouldbesaidthatthey arenormativemodelsthatfocusontheconsequencesofsafetyculture.Researchersintheeldofsafetymanagementhavealsore-ectedonsafetyculture.Reason(1997)spendsafullchapteronsafetycultureitscomponentsand engineeringasdoHaleandHovden(1998),whodeliberatesafetycultureincon-siderabledetail.However,adiscussionoftheseviewsisconsideredbeyondthescope ofthispaper.Nevertheless,itshouldbenotedthattheconceptofsafetyculturehas raisedtheinterestofresearchersinmanyrelatedelds.Additionalapproachesof interestarereviewedinSection3.5.Inconclusion,atpresentthereisnooverallsatisfyingmodelofsafetyclimateorsafetyculture.However,throughoutthepaperseveralbuildingblocksforsucha modelhavebeenindicated.Forinstance,adistinctionwasmadebetweenculture andclimate,whichwasassociatedwiththelayeredmodelbySchein(1992).Climate wasequatedwithespousedvalues,whichwerethereuponidentiedasattitudes.As aresult,anorganisation'ssafetyclimateismadeupofitsmembers'safetyattitudes.

Also,theobjectsofattitudeswerementioned.Followingthecurrentlineofreason-ingthesewouldmakeupthecontentofsafetyclimate,whilesafetyculturecouldbe denotedastheircause.Finally,safetyclimate'sconsequenceswouldbetheevalua-tiveresponses,whethercognitive,a*ectiveorbehavioural.InSchein'sterminologythesewouldbecalled``artefacts.F.W.Guldenmund/SafetyScience34(2000)215+/-257 243 3.4.LevelofaggregationIntheabovediscussionoforganisationalcultureandclimate,itwasindicatedthatthelevelofaggregationisanimportantpointforre-ection.Therefore,itissomewhatsurprisingthatthispointhasnotbeengivendueattentioninsafetycultureandcli-materesearch.Forinstance,whentalkingabouttheobjectsofattitudesonecan seriouslyquestionwhethertheseobjectsremainthesameatdi*erentorganisational levels.Thatis,itisatleastdoubtfulthattheattitudeobjectsofindividualsarethe sameasthoseforgroupsororganisations.Hence,whenaggregatingindividualdata tothelevelofanorganisationalgrouporunit,itisopentoquestionwhetherthe combineddataactuallycorrespondtoanattitudeobjectexistingatthatlevel.Moreover,asdiscussedpreviously,ateachparticularlevelofaggregationtheissueofcommunalityarisesacertainamountofhomogeneityofopinionisneededin ordertobeabletospeakofsharedattitudesorassumptions.Thisisnottosaythatitisnotpossibletocompareaggregateddata,butbyaggre-gatingdataonedoesnotnecessarilygetinformationaboutattitudeobjectspertinent atthatlevelofaggregation.Toobtaindatavalidatacertainlevelofaggregation,one shouldaskquestionsaboutobjectspertinentatthatlevel.Inallothercasesonestill hasdataabouttheindividuallevel.Toensurethattheirrespondentsconsistentlygiveanswersaboutthewholeorganisationandnotaboutthework-groupDeCocketal.(1986,p.7)includedthisrequirementintheinstructionsoftheirquestionnaire.Atrstsight,theserecommendationsseemsensibleandeasytofollow.However,whenworkingwithquestionnaires,oneisconfrontedwithseveralphenomenalike ambiguity,poly-interpretability,thelackofclarityoflongsentences,etc.,which threatenthevalidityoftheresults.Obviously,givinganswerstoquestionsnotabout one'sownpersonalenvironmentbutaboutthewholeorganisationrequiresasig-nicantcognitivee*ort.Itistobeexpectedthatrespondentswillreplacesuchabstractreferenceswiththeirpersonalsubstitutions.Ashasbeensaidearlier,thisissuehasnotgottheattentionitwarrants.Additionalresearchisneededtoshedmorelightonthisissue,theseverityofitsconsequencesandpossiblesolutions.3.5.OtherapproachesUptonow,themainfocushasbeenonappliedresearchconductedinthetradi-tionofsocialororganisationalpsychology.TheTotalSafetyCulture(TSC)advocatedbyGeller(1994)isnotadiagnosticorevaluativequestionnaire-basedapproach,butisactuallyaimedatchangingthesafetycultureinadesired,pre-deneddirectionthroughmostlybehaviour-directed processes.ATSCcanbedevelopedwhenemployeesunderstandandacceptthe10 principlesoutlinedbyGeller.Thisobjectiveisaccomplishedbytheapplicationof veactionplans.Gellerdoesnotindicatetowhatkindofindustrialorganisations hisTSCappliesorwhatkindofpreconditionsarenecessaryforsuccessfulapplica-tion.ItseemsthatitisarguedthatjusttherigorousutilisationoftheveactionplanswillresultinaTSC.

244F.W.Guldenmund/SafetyScience34(2000)215+/-257 ThereportonsafetyculturebytheInternationalSafetyAdvisoryGroup(1991)alsofollowsanormativeapproach.Accordingtothem,safetycultureconsistoftwoelements,a``necessaryframeworkwithinanorganisationand[

...]theattitudeofsta*atalllevelsinrespondingtoandbenetingfromtheframework(p.5)andappliesto bothorganisationsandindividualswithinthoseorganisations.Establishingasafety culturemeansspecifyingdemandsatseverallevels,i.e.requirementsatpolicylevel, requirementsonmanagersandresponsesofindividuals.Foreachoftheselevels requirementsarespecied.Thesepertain,amongotherthings,toknowledgeandcom-petence,commitment,motivation,supervision,individualawarenessandresponsibility.

With``frameworktheInternationalSafetyAdvisoryGroupimplies``organizational policyand``managerialaction(p.2).Althoughattitudesareconsidered``generallyintangibletheyhavemanifestoutcomesandparticularsatisfactoryindicatorsareprovidedbytheInternationalSafetyAdvisoryGroup.ThesafetycultureauditmethodoutlinedbyLudborzs(1995)istoalargeextentcomparablewiththesurveyapproachdiscussedextensivelyinthispaper,inthatit attemptstoquantifyparticularsafetycultureindicators.However,thisisnot attainedbyanextensivesurvey,butratherthroughinterviewswithkeyindividuals andemployeesandthroughobservations,asiscommonpracticein(safety)man-agementaudits.Throughtheanalysisofboth``documentedandlivedstructuralorganisationand``documentedandlivedoperationalorganisation,shortcomingsinimplementationareassessed.Themethoddenestenbroadareasofanalysis, whichareinvestigatedbymeansofchecklistswithdetailedindicators,whichhaveto bescoredseparately.Safetycultureisusedherenormativelyinthatitisapplied exclusivelytoorganisationswithapositiveassessmentforall10areaswithabove-averagefrequency.Varyingresultsareconsideredanindicationoftheexistenceof subculturesorcounter-cultures.Inthatcasetheterm``livedsafetyisapplied.Ludborzscorrectlyrecommendsnevertolosesightofthe``culturalsuper-structurewheresafetycultureisonlypartofacorporateculture,whichitselfispart ofanindustrialcultureandanationalculture.Asamatteroffact,itisremarkable howfewresearcherspointoutthissubdivision.Itmightbeverywellassumedthat safetyculturesnotonlydi*erbetweenthemselves,butalsobecauseofdi*erences betweenindustrialandnationalcultures.Itisthereforestrikingthatthesesourcesof variancedidnotgetanyattentionintheappliedresearchesreported.Foronething, thisisbecausetheorganisationisnormallythehighestlevelofaggregationinorga-nisationalpsychologicalresearch.Foranother,includingbothindustrialcultureandnationalcultureinone'sinvestigationwouldcomplicatemattersbeyondwhatisconsideredpracticalresearch.Therefore,approachesotherthanthosethathave beenreviewedherearemoreofatheoreticalandre-ectivenature.TheSafetyCultureHAZOPbyKennedy(1997)isamodicationoftheHazardandOperabilityStudy(HAZOP),whichisoneoftheestablishedtechniquesto identifyhazardsincomplexengineeringsystems.AHAZOPsessionand,like-wise,aSCHAZOPsessionisagroup-basedmethodology.Thisgroupconsistsof achairman,asecretaryandaselectionofpersonnelknowledgeableaboutthesafetymanagementprocessbeingstudied.Throughaprocessofbrainstormingandanensuing,structureddiscussion,asafetymanagementprocessrepresentedinF.W.Guldenmund/SafetyScience34(2000)215+/-257 245 diagrammaticalformisexaminedbymeansofguidewords(e.g.``missing,``skip-ped,``mistimed)andpropertywords(e.g.``person,``action,``procedure/speci-cation).Theresultofsuchasessionisasetofsafetymanagementareavulnerabilities.Interestingly,themethodproposedbySchein(1992,p.147*.)togetaninitialviewofacompany'sorganisationalcultureisnotunliketheSCHAZOP approachdiscussedhere.Inhisevaluationofthe(sociological)signicanceoftheconstructMijs(1992)viewsorganisationalcultureaspartofatrinitythatalsoencompassesorganisational regimeandorganisationalstructure.Organisationalcultureandstructureaswellas regimeareaspectsystemsthatcanbedistinguishedanalytically.Quiterightly, Mijswarnsthatoneshouldguardagainstreication,inthatthesesystemsarecon-sideredactualsub-systemsthatcanbeisolatedandmanipulatedseparately.Oneshouldtakeaccountofthefactthatthesesystemsareembeddedintoaeldofforces consistingofnationalculture,industrialandoccupationalculturesandsituational factorsliketechnology,typeoflabour,ageoforganisationandthelike.These in-uencescertainlyputalimitonwhatisfeasibleintermsofchangeof,forinstance, organisationalculture.Whenindustrialandnationalculturesarealsoembraced,wendourselvesinthecompanyofsociologists,politicalscientistsorconceptuallyorientedpsychologists.AnexampleofthelattercategoryisPidgeon.Hisscopeisapparentfromthefactthatheconsidersorganisationsultimatelyassub-cultureswithinsocieties(Pidgeon, 1991).Inseveralpublications(Pidgeon,1991,1997,1998)heembedsorganisational safetycultureinitsindustrialandpoliticalenvironment,wheretheoccasional``man-madedisaster,e.g.Chernobyl,Challenger,Exxon-Valdez,hasaprofoundimpact onbothpoliticalandsocietalviewsonsafety.AccordingtoPidgeon(1991),a

``goodsafetyculturecanbecharacterisedbythreeattributes:``normsandrules forhandlinghazards,attitudestowardsafety,andre-exivityonsafetypractice(p.135).Althoughallthreeattributesareimbuedwithpoliticalorsocietalthinking, especiallythelastisfacilitatedwhenitisconsideredatanindustry-widelevel,where learningisincreasedsubstantiallythroughthecollectionanddisseminationofinci-dentandaccidentdata.Summarising,theapproachestowardsafetyculturethathavebeendiscussedinthissectiontosomeextentdenetwoextremesofthecontinuumdescribingtheinter-pretationoftheconceptofsafetyculture.Atoneextreme,safetycultureisnormative, havingdistinctfeatures(Geller,1984;InternationalSafetyAdvisoryGroup,1991).Whenthesefeatureshavebeenimplemented,asafetycultureisestablished.Attheotherextreme,safetycultureisseenasjustasmallelementinaeldofdistinctforces, i.e.safetycultureisrelative(Pidgeon,1991,1997,1998;Mijs,1992).Clearly,the approachthathasbeenthemainfocusofthispaperfallssomewhereinbetween,with particularresearchersincliningtowardsoneortheotherextreme.4.ThenatureofsafetycultureThecurrentliteraturereviewofsafetycultureandsafetyclimatehasshownthat:

246F.W.Guldenmund/SafetyScience34(2000)215+/-257 1.theconceptsofsafetycultureandsafetyclimatearestillill-denedandnotworkedoutwell;2.therelationshipbetweensafetycultureandsafetyclimateisunclear;3.thereisconsiderableconfusionaboutthecause,thecontentandthecon-sequenceofsafetycultureandclimate,i.e.:

.thecauseofsafetycultureandclimatehasnotbeenaddressedseriously;

.thereisnoconsensusonthecontentofsafetycultureandclimate;and

.theconsequencesofsafetycultureandclimateareseldomdiscussed.4.thereisnosatisfyingmodelofsafetyculturenorsafetyclimate;and5.theissueofthelevelofaggregationhasnotreceivedtheattentionitwarrants.However,thisisnottosaythatnothinghasbeenaccomplished,onthecontrary.Nevertheless,itisregrettablethatfewauthorshaverelatedtheirworktoresearchby othersorhavetriedtoestablishanintegrativeframework.Suchaframeworkisthe subjectofthenextparagraphsandwill,hopefully,beusefulinsteeringfutureresearch.4.1.SafetyattitudesMostresearchersofculturewhethernational,organisationalorsafetyculturedistinguishseverallevelsatwhichmanifestationsofculturecanbeobserved(e.g.DealandKennedy,1982;Hofstede,1991;Schein,1992).Examplesof suchlevelshavebeendiscussedinSection2.2above.ForthepresentframeworkthethreelevelsofSchein(1992)arechosen,mainlybecauseoftheirintuitiveappealandtheconvenienceofjustthreelevels.Asa reminder,histhreelevelsare:I.Basicassumptions.

II.Espousedvalues.

III.Artefacts.Also,averyglobalmodelofattitudeswasdiscussedinSection3.3.Itstatedthatattitudesareprecededeitherbycognitive,a*ectiveorbehaviouralprocessesandthat attitudesyieldcognitive,a*ectiveaswellasbehaviouralresponses.Again,threestages canbediscerned.However,thismodelisstillveryroughandundistinguished.Eagly andChaiken(1993,p.209)proposeacompositemodeloftheattitude+/-behaviourrela-tion,whichismorespecic.Therearetwoimportantpointstobemadeaboutthismodel(Fig.5).Firstly,thisisstillanormativemodel,althoughitisbasedonmodels thathavealreadybeentested.Secondly,themodelfocusesonbehaviourandthere-foreneglectstheothertwoattitudinalresponses,namelya*ectiveandcognitive responses.Theadvantageofthismodelisthatitspeciesalbeitverybroadlythepro-cessesthatprecedeattitudeformation.Earlierinthispaper,attitudeswerecompared withthesecondlayerinthethree-layeredculturemodel.Manifestationsofcultureatthislevelwerecalled``espousedvalues.Itisnowsuggestedtoequateattitudeswithespousedvalues.Clearly,theprocessesthatprecedeattitudeformationthenshouldF.W.Guldenmund/SafetyScience34(2000)215+/-257 247 beequatedwiththecoreofculture,namelythebasicassumptions.Iwillhavemoretosayabouttheselater.Uptohere,thebuildingoftheframeworkhasbeenneutralwithregardtoitsobjective.Thisframeworkcouldbeappliedtoanyaspectoforganisationalculture.However,thepresentfocusisonsafetyandsafetyculture.Attitudesalwayshaveobjects.Althoughtherearecountlessobjectswithregardtosafetyimaginable,theseobjectswillpresumablyfallintoafewcategories.Duringthediscussionofthemodelsthathavebeendevisedtoaccountforsafetycultureand climatephenomena,suchcategorieshavealreadybeenmentioned.Themodelby CoxandCox(1991)forinstance,refersto``hardware,``software,``people(live-ware)and``risks.Asatruebehaviourist,Geller(1994)suggests``people,

``environmentand``behaviour,leavingout``software.Alsotheaccident/incidentinvestigationmanualfortheUSDepartmentofEnergy(DOE)mentions``plant-personnel,``plant-hardwareand``proceduralsystemsandtheinterfacesbetween thesethree(Johnson,1985).Itissafetosaythenthatthefollowingfourbroadcategoriesofsafetyattitudeobjectshavesomesubstance:1.hardware/physicalenvironment; 2.software; 3.people;and 4.behaviour.Fig.5.Processesprecedingattitudeformation(adaptedfromEaglyandChaiken,1993).

248F.W.Guldenmund/SafetyScience34(2000)215+/-257 Specicexamplesofhardwareattitudeobjectswouldbesafetymeasuresandarrangementsorpersonalprotectiveequipment.Safetyprocedures,trainingandknowledgewillcomeundertheheadingofattitudestowardsoftware.Thecategoryofattitudestowardpeoplewillencompassalldi*erentkindsofpeopleandgroupsthatcanbedistinguishedwithinacompany,likemanagement,supervisors,collea-guesandsoon.Finally,attitudetowardbehaviourwillincludeallactswithregard tosafety(orlackofsafety)likeresponsibility,safeworking,scepticismandcom-municationaboutsafety.4.2.Safetyculture:basicassumptionsAbove,somethingwasalreadysaidaboutthecoreofsafetyculture,namelythebasicassumptions.Itwouldbeobvioustoassumethatthesebasicassumptionswillalsobeformedaroundthecategoriesdenedforsafetyattitudesabove.However,in thesensethatSchein(1992)denesthemtheydonothavetobe.Hedenesbasic assumptionsas``theimplicitassumptionsthatactuallyguidebehaviour,thattell groupmembershowtoperceive,thinkabout,andfeelaboutthings.Such assumptions``havebecomesotakenforgrantedthatonendslittlevariationwithin aculturalunit.[

...][M]emberswillndbehaviorbasedonanyotherpremiseinconceivable(p.22).Hence,suchpremisesmightbespecicallyaboutsafetybut donotnecessarilyhavetobeso.Forexample,ifinsomeorganisationwrittenrules orproceduresareconsideredfutile,safetyruleswillbetoo.Therefore,onemight ndnegativeattitudestowardsoftware(rulesandprocedures)inthisorganisation.

Thisndingdoesnotmean,however,thatthebasicassumptionisthatonlysafety rulesarefutilebutthatrulesingeneralare.

8Envisionedthisway,basicassumptionscanonlyfunctionasexplanatoryvari-ables,i.e.theyexplaintheattitudestructurefound.Moreover,basicassumptionshaveamorepervasivein-uencethanattitudes,inthatbasicassumptionstranscendparticularorganisationalunitslikegroupsordepartmentsorparticulartypesof culturelikesafetyculture.Nowwecanalsoseehowthesehypothesisedbasicassumptionslinktothepre-attitudecomponentsofFig.5.Thesecategoriesarehabits,attitudestowardtargets, utilitarianoutcomes,normativeoutcomesandself-identityoutcomes.Itisnothardto considersomeoftheseasbasicassumptions.Forinstance,utilitarianornormative outcomes,whichpertaintorewardorpunishmentandtheapprovalofsignicantoth-ers,respectively(EaglyandChaiken,1993,p.209),areusuallydeep-rootedorganisa-tionalprocesses(Schein,1992).Habitsalsore-ectdeeporganisationalconvictions aboutwhatworksandwhatdoesnot.Theattitudetowardthetargetmightalsobeastrong,basicassumption.Forinstance,inaproductioncompanyahighproductionisusuallyconsideredthe greatestgood.Itshouldbe,otherwisethecompanywillgobrokeintheshortrun 8Forinstance,becausetheyprovideameansforexcuseslike``Ididn'tdoit,becausetherulesdidn'tsayIshouldastheauthorencounteredinoneparticularcompany.F.W.Guldenmund/SafetyScience34(2000)215+/-257 249 thewholeorganisationisgenerallyleavenedwiththisfact.Therefore,individualsmightbreakcertainsafetyrulesbecauseofthegreatestgoodwhichisproduction.Inthisrepresentation,anorganisation'sbasicassumptionsarecompletelymouldedintothecastofattitudesandtheircorrespondingmodels.Thisisjustonewayofget-tingarmergriponanorganisation'sbasicassumptions.Schein(1992)alsomentions particulardimensions,aroundwhichsharedbasicassumptionsform(p.95+/-96):1.Thenatureofrealityandtruththeseassumptionsgenerallydenewhatisrealandwhatisnot,or,morespecically,whatissafeandwhatisnot;2.Thenatureoftime; 3.Thenatureofspacethesedimensionsdenetheimportanceoftimeandspacewithinanorganisation,howtheyareusedandlled.Whenrelatedto safety,thesedimensionscouldrevealtheassumptionsaboutworkplaces, theirhazardsandtheirhousekeepingandthetimespentonsafety,prepara-tionofworkandworkitself;4.Thenatureofhumannaturethisdimensionre-ectsassumptionsaboutpeople'sintrinsicqualitiesandwhatcanbedoneaboutthem,e.g.whether somepeopleareaccidentproneorlikelytoengageinriskybehaviour;5.Thenatureofhumanactivitytheseassumptionsdenewhatis`work'andtherightthingforpeopletodoinrelationtotheirenvironment;towhat extentpeopleshouldtakeinitiativeorawaitinstruction;6.Thenatureofhumanrelationshipsthisdimensionisallabouthowpeoplerelatetoeachother,e.g.competition,individualism,co-operation,authorityofindividuals,includingissueslikewhetheritisacceptabletocorrectotherpeople'sunsafebehaviour.Clearly,Schein'sdimensionsarethemselvesratherabstractconcepts,incontrastwithmoreconcretecategoriesoftheattitudemodel.Attitudemodelsliketheoneabove,are usuallytestedinalaboratorysetting.Here,subjectslloutsomequestionnairesand theirresponsesaresubjectedtoalinearstructuralrelationsanalysisdemandinga numericalinput.Thequestionnairesthereforecontainsomewell-delineatedconstructs, assumedtobeofrelevance,thatarecoveredbyseveralquestions.Ontheotherhand, Schein'sdimensionshavemoreofananthropologicalnatureseekingunderstanding ratherthanreduction.Althoughbothcategorysystemscannotbereducedtoone another,itappearsthatthereisstillconsiderableconceptualoverlap.Forinstance,thebasicassumptionsabouthumannaturewillcertainlyencompasshabitformationandbeliefsaboutself-identity.Or,attitudestoward(theapprovalof)signicantothers(Fig.

5)mostcertainlyre-ectbasicassumptionsabouthumanrelationships.4.3.SafetycultureredenedSchein(1992)denesorganisationalcultureas:apatternofsharedbasicassumptionsthatthegrouplearnedasitsolveditsproblemsofexternaladaptationandinternalintegration,thathasworkedwell 250F.W.Guldenmund/SafetyScience34(2000)215+/-257 enoughtobeconsideredvalidand,therefore,tobetaughttonewmembersasthecorrectwaytoperceive,thinkandfeelinrelationtothoseproblems(p.12).Scheinhasincludedtwoofthethreeresponsescommonlyassociatedwithattitudes, i.e.cognitive(``perceive,``think)anda*ective(``feel)responses.Hehasdeliber-atelyleftbehaviourout,whichhereservesfortheouterlayers,i.e.espousedvalues andartefacts.Scheinalsolimitshisdenitiontowhatheassumesisthecoreof organisationalculture.Actually,inthewayScheinconceivesanddenes(organisa-tional)culture,thereisnoneedforaspecicdenitionforsafetyculture.Thebasic assumptionspermeatethroughouttheorganisation,includingitsaspectofsafety.In thisway,Scheinremainsfaithfultotheoriginalconceptionoforganisationalcultureasanoverall,integrativeconcept.WhentalkingaboutclimateandassumingthatclimateconformstotheespousedvaluesinSchein'smodelwhicharethenoper-ationalisedasattitudes,itisnecessarytodeneobjectsfortheseattitudes.

Obviously,dependingontheobjectsoftheattitudes,di*erentclimatesexist.Asobservedabove,thedi*erenttypesofculturethataretobefoundinthelit-erature,haveonlybeendenedforanalyticalorpracticalreasonsandtofocusthe research.Hence,forthesamepracticalreasons,adenitionofsafetyculturewillbe given.Safetycultureisdenedas:thoseaspectsoftheorganisationalculturewhichwillimpactonattitudesandbehaviourrelatedtoincreasingordecreasingrisk.Insummary,thefollowingframeworkisproposed(Table6).Safetycultureisconceptualisedashavingthreelayersorlevelsatwhichitmightbestudiedsepa-rately.Thecoreisassumedtoconsistofbasicassumptions,whichareunconsciousTable6Levelsofculture,theirvisibilityandexamplesthereofLevelsofcultureVisibilityExamples 1.OuterlayerartefactsVisible,buthardtocomprehendintermsofunderlyingcultureStatements,meetings,inspectionreports,dresscodes,personal protectiveequipment,posters,bulletins2.Middlelayerespousedvalues/attitudesregarding:

-hardware,-software,-people/liveware,

-risksRelativelyexplicitand consciousAttitudes,policies,training manuals,procedures, formalstatements,bulletins,accidentandincidentreports,jobdescriptions, minutesofmeetings3.Corebasicassumptionsregarding:-thenatureofrealityandtruth,-thenatureoftime,

-thenatureofspace,

-thenatureofhumannature,-thenatureofhumanactivity,-thenatureofhumanrelationshipsMainlyimplicit:obviousforthemembers,invisible,pre-consciousHavetobededucedfromartefactsandespousedvaluesaswellasthrough observationF.W.Guldenmund/SafetyScience34(2000)215+/-257 251 andrelativelyunspecicandwhichpermeatethewholeoftheorganisation.Thenextlayerconsistsofespousedvalues,whichareoperationalisedasattitudes.Attitudes havespecicobjectsandthereforethislayeris,necessarily,specicwithregardtotheobjectofstudy.Forsafetyculturefourcategoriesofobjectsaresuggested;hardware,software,peopleandbehaviour.Finally,theoutermostlayerconsistsof particularmanifestations,whicharealsospecictotheobjectofstudy.Withregard tosafetyonemightthinkofinspections,posters,wearing(ornot)ofpersonal protectiveequipment,accidentsorincidents,near-missesordi*erenttypesof behaviour.Theappealofthisframeworkisthatitfusessafetyclimateandsafetycultureandthatitalsodoesjusticetotheintegrative,holisticconceptofcultureasadvocatedby,forinstance,culturalanthropologists.Inaddition,anotherelaborationcanbemade.Ashasbeenclaimedabove,thebasicassumptionsdonothavetobespeci-callyconcernedwithsafety.Althoughtheydonothavetobespecicallyso,itis quiteconceivablethatsomeoftheorganisation'sbasicassumptionsinfactare, whensafetyistakenseriouslywithintheorganisationandre-ecteduponbyallofits members.Thiswouldcertainlyleadtoananchoringofsafetywithinthebasic assumptions.Thissuppositioncouldbeconvertedintoahypothesisstatingthatitis agoodsignthat,amongthebasicassumptionsofanorganisation,referencestosafetyaremade.Conversely,itissuspectwhensuchreferencescannotbefound.Itmightverywellbethatonehastoconcludethatsuchanorganisationdoesnotyield sucientevidencefortheexistenceofasafetyculture.5.DiscussionThisreviewofsafetyclimateandsafetycultureresearchhasbeenlargelyfromasocialpsychologicalpointofviewandhasfocusedprimarilyonresultsfrom20yearsofresearchinthiseld.Anintegrativeframeworkhasbeenproposed,mergingsafetycli-matewithsafetycultureanddeliveringcategoriesforbothsafetyattitudesandbasic assumptionsthatareopentoinvestigation.However,aquestionthathasnotbeen posedyetpertainstotheuseandutilityofthesafetycultureandclimateconstruct.AscanbeseeninTable1,allresearchershavedenedcertaingoals,oftenbeingofboththeoreticalandpracticaluse.BothLutness(1987)andBaileyandPetersen (1989)outlineparticulargoalsthatgobeyondthemeredeterminationofsafetycul-ture,inthattheyconsidersuchmeasurementaperformanceindicator.Forinstance,Lutness(1987,p.20)aimstoreveal``asafetyprogram'sstrengthsandweaknesses.

BaileyandPetersen(1989,p.20)wanttodevelopanalternativemeasureforsafety performancewhile``thee*ectivenessofsafetye*ortscannotbemeasuredbytradi-tional(procedural-engineering)criteria.Theseresearchersarereferringtosafety attitudemeasurement,i.e.whathasbeencalledsafetyclimateinthispaper.Hence,thedeterminationofsafetyclimatehasbeenputforwardbysomeauthorsasanalternativeperformanceindicator,inadditiontothemoreestablishedoneslikesafetymanagementaudits,accidentsandincidentsandnear-misses(seealsoBud-worth,1996).Thismeansthatthereshouldexiststrongrelationshipsbetweenall 252F.W.Guldenmund/SafetyScience34(2000)215+/-257 thesemeasures.Ashasbeenassertedbefore,suchrelationshipshavenotbeenreportedoften.Atpresent,therearefewstudieswhichhavetriedtoestablishsuch correlations,i.e.arelationshipbetweensafetyperformancemeasuresandsafetycultureorclimateassessments(e.g.Cabreraetal.,1997;Erickson,1997).Hurstetal.(1996)reportarelationshipbetweencertainauditedmanagementareasand attitudinalmeasures.Also,themodicationfactorsresultingfromtheirauditanda self-reportedaccidentmeasurearehighlycorrelatedintheirstudy.Clearly,these relationshipsneedtobeexploredinmoredetailtobeabletomakesubstantive statementsoftheusefulnessofasafetycultureorclimatemeasureasanalternative performanceindicator.Throughtheirempirical,questionnaire-basedstudy,SimardandMarchand(1996)illustrateconvincinglythein-uenceofwhattheycall``microorganisationalfactors

onsafetyinitiatives.Theirresultsshowthat,especially,participatorysupervision shapesthepropensityofworkgroupstotakesuchinitiatives.Possibly,suchatypeof leadershipisaproductofanunderlyingculture.Relationships,correlationsand,ingeneral,comparisons,bringtheissueofquan-ticationupfrontagain.Withregardtosafetyclimatethisshouldnotcauseamajor problem,becauseattitudesareusuallysurveyedthroughself-administeredques-tionnaireswhichgenerallyprovidesuchmeasuresinsemi-quantiedform.Theonlypointtoworryaboutthenisgettingenoughdatatobeabletomakestatisticallysoundgeneralisations,herebykeepingawakeful,methodologicaleyeonthemea-surementlevelofthedatacompared.Theassessmentofsafetyculture,however,as conceptualisedinthispaperasasmallsetofimplicitbasicassumptions,doesnot haveanumericalcounterpart.Comparisonswillhavetobemadeinhypotheticalif

...then...statements,like-ifsuch-and-suchbasicassumptionsareuncovered thenwewillalsondincidentsandaccidentswithsuch-and-suchcauses.Clearly,tobe abletomakesuchstatements,afairamountofcasestudieshavetobeconductedaccordingtotheframeworkpresentedinthispaper.However,whenagivensafetycultureorclimatehasbeenassessed,thenextquestionwillcertainlybesowhat?Mostofthesafetyclimateresearchreportedherewouldyieldsomescoresoncertaindimensions.However,thosescoresdonotspeakfor themselves,i.e.themeaningofthescoreswillnotbeobvious.Moreover,because mostresearchersworkwiththeirowndimensionsorscales(Table4),itisimpossible torefertogeneralnormsorbenchmarks 9Evenifitisassumedthatscoresoncertaindimensionsareconspicuouslylow,thequestionwillremainsowhat?Asub-sequentstrong(management)focusonthecontentofthoselowscoreddimensionscommunication,forexamplewouldviolatetheholisticcharacter ofculture.Again,onerunsintothismethodologicalparadoxthattheanalysis methodsimposeonthedata.Usually,theresultsofthedataanalysesareuncorre-lateddimensions,butinactualitythisishardlyeverthecaseandisalsoincontrast withtheholisticcharacterthatisattributedtocultureandclimate.

9However,bothLeeandtheSafetyResearchUnithavelargedatabasesmakingcomparisonsandrelativestatementspossible.F.W.Guldenmund/SafetyScience34(2000)215+/-257 253 Thepresentpaperhasnotpromotedthisapproach,though.Theassessmentofsafetyattitudesorsafetyclimatethroughquestionnaires,isonlypartoftheadvocatedapproach.Thebasicassumptions,whichareexplanatorytosafetyattitudes,alsohavetobeassessedforarecommendationorfruitfulintervention.Actually,sub-sequentinterventionsshouldonlybeundertakenwithdetailedknowledgeofa company'sparticularbasicassumptionsasexplanatoryvariablesparexcellence.In thatcase,therearetwoalternativesforaction.Either,anattemptatchangingthe basicassumptionsisundertakenoranattemptatchangingthesafetyattitudesis undertaken, givenaparticularsetofbasicassumptions.Clearly,therste*ortmightturnouttobethemostdiculttoattain,ifitisindeedfeasible.Thelatterone, althoughthemostfeasible,mightstilltakeafewyears.Forinstance,DeCocketal.(1986)mention5years.Obviously,assessingsafetyclimateorsafetyculturewiththeobjectofchangingitisbothambitiousandtimeconsuming,spanningaperiodalot ofmanagerswillnotevenseetheendof.Consequently,themeasurementofsafetyclimatecouldbeconsideredanalter-nativesafetyperformanceindicator,whereastheassessmentofsafetyculturepro-videsmoreinsightintotheparticularattitudesfound,henceparadoxically yieldingthesubstrateforbothsafetyimprovementsandunforeseenmajoraccidents (Pidgeon,1998).Asthepresentreviewillustrates,researchshouldnotbeundertakentodevelop`new'safetyclimatemeasurementinstruments,butshouldratherfocusonthevalidityoftheconstructandwhetheritindeedyieldsarobustindicationofanorga-nisation'ssafetyperformance.Inaddition,increasingresearche*ortsshouldbedirec-tedatdevelopingmeanstoassessanorganisation'sbasicassumptions,forgettinga muchdeeperunderstandingof``thewaywedothingsaroundhere.ReferencesAdvisoryCommitteeontheSafetyofNuclearInstallations(ACSNI),1993.StudyGrouponHumanFactors,ThirdReport:OrganisingforSafety.HSMO,London.Bailey,C.W.,Petersen,D.,1989.Usingperceptionsurveystoassesssafetysysteme*ectiveness.Profes-sionalSafetyFebruary,22+/-26.Berends,J.J.,1995a.Organisatie,cultuurenveiligheid:Eenliteratuurstudienaarorganisatiecultuurenveiligheidscultuur(Unpublishedinterimreport).EindhovenUniversityofTechnology,Eindhoven.Berends,J.J.,1995b.DevelopingandUsingaWidelyApplicableMeasurementToolforSafetyCulture(Unpublishedinterimreport).EindhovenUniversityofTechnology,Eindhoven.Berends,J.J.,1996.OntheMeasurementofSafetyCulture(Unpublishedgraduationreport).EindhovenUniversityofTechnology,Eindhoven.Borg,I.(Ed.),1981.Multidimensionaldatarepresentations:Whenandwhy.MathesisPress,AnnArbor.Brown,R.L.,Holmes,H.,1986.Theuseofafactor-analyticprocedureforassessingthevalidityofanemployeesafetyclimatemodel.AccidentAnalysisandPrevention18(6),455+/-470.Budworth,N.,1996.Indicatorsofperformanceinsafetymanagement.TheSafetyandHealthPractitionerNovember,23+/-29.Cabrera,D.D.,Isla,R.,Vilela,L.D.,1997.Anevaluationofsafetyclimateingroundhandlingactivities.In:Soekkha,H.M.(Ed.),AviationSafety,ProceedingsoftheIASC-97InternationalAviationSafety Conference,Netherlands,27+/-29August,pp.255+/-268.Cooper,M.D,Philips,R.A.,1994.ValidationofaSafetyClimateMeasure.PaperpresentedattheBritishPsychologicalSociety,AnnualOccupationalPsychologyConference,Birmingham,January3+/-5.

254F.W.Guldenmund/SafetyScience34(2000)215+/-257 Coyle,I.R.,Sleeman,S.D.,Adams,N.,1995.Safetyclimate.JournalofSafetyResearch26(4),247+/-

254.Cox,S.,Cox,T.,1991.Thestructureofemployeeattitudestosafety:anEuropeanexample.WorkandStress5(2),93+/-106.Cox,S.,Flin,R.,1998.Safetyculture.Philosophers'stoneormanofstraw?WorkandStress12(3),189+/-

201.Deal,T.E.,Kennedy,A.A.,1982.Corporatecultures.Addison-Wesley,Reading,Mass.DeCock,G.,Bouwen,R.,deWitte,K.,1986.Organisatieklimaat:Eenopdrachtvoorhetpersoneelsbe-leid?PraktischPersoneelsbeleid,CapitaSelecta16,1+/-20.DeDobbeleer,N.,Be

Âland,F.,1991.Asafetyclimatemeasureforconstructionsites.JournalofSafetyResearch22,97+/-103.Dieterly,D.L.,Schneider,B.,1974.Thee*ectoforganizationalenvironmentonperceivedpowerandclimate:alaboratorystudy.OrganizationalBehaviorandHumanPerformance11,316+/-337.Donald,I.J.,Canter,D.V.,1994.Employeeattitudesandsafetyinthechemicalindustry.JournalofLossPreventionintheProcessIndustry7(3),203+/-208.Eagly,A.H.,Chaiken,S.,1993.ThePsychologyofAttitudes.HarcourtBraceJovanovich,FortWorth.

Ekvall,G.,1983.Climate,StructureandInnovativenessofOrganisations.WorkingpaperofTheSwedishCouncilforManagementandOrganisationalBehaviour.Erickson,J.A.,1997.Therelationshipbetweencorporatecultureandsafetyperformance.ProfessionalSafetyMay,29+/-33.Friedlander,F.,Margulies,N.,1969.Multipleimpactsoforganizationalclimateandindividualvaluesystemsuponjobsatisfaction.PersonnelPsychology22,171+/-183.Frissen,P.,1986.Organisatiecultuur:eenoverzichtvanbenaderingen.M&O,TijdschriftvoorOrganisa-tiekundeenSociaalBeleid6,532+/-544.Geller,E.S.,1994.TenprinciplesforachievingaTotalSafetyCulture.ProfessionalSafetySeptember,18+/-

24.Glennon,D.P.,1982a.Measuringorganisationalsafetyclimate.AustralianSafetyNewsJanuary/Feb-ruary23+/-28.Glennon,D.P.,1982b.Safetyclimateinorganisations.Proceedingsofthe19thAnnualConferenceoftheErgonomicsSocietyofAustraliaandNewZealand,17+/-31.Glick,W.H.,1985.Conceptualizingandmeasuringorganizationalandpsyhologicalclimate:pitfallsinmultilevelresearch.AcademyofManagementReview10(3),601+/-616.Green,P.E.,Carroll,J.D.,1978.Mathematicaltoolsforappliedmultivariateanalysis.Secondedition.AcademicPress,NewYork.Guion,R.M.,1973.Anoteonorganizationalclimate.OrganizationalBehaviorandHumanPerformance9,120+/-125.Guttman,L.,1968.Ageneralnonmetrictechniqueforndingthesmallestcoordinatespaceforacong-urationofpoints.Psychometrika33(4),469+/-506.Hale,A.R.,Hovden,J.,1998.Managementandculture:thethirdageofsafety.Areviewofapproachestoorganisationalaspectsofsafety,healthandenvironment.In:Feyer,A.M.,Williamson,A.(Eds.),

OccupationalInjury;Risk,PreventionandIntervention.TaylorandFrancis,London,pp.129+/-166.Hofstede,G.R.,1986.Werkenaandeorganisatiecultuur.Bedrijfskunde58(2),102+/-106.

Hofstede,G.R.,1991.Culturesandorganisations:softwareofthemind.McGraw-Hill,London.Hurst,N.W.,Young,S.,Gibson,H.,Muyselaar,A.,1996.Measuresofsafetyperformanceandattitudestosafetyatmajorhazardsites.JournalofLossPreventionintheProcessIndustry9(2),161+/-172.InternationalSafetyAdvisoryGroup,1991.SafetyCulture(SafetySeriesNo.75-INSAG-4).InternationalAtomicEnergyAgency,Vienna.James,L.R.,1982.Aggregationbiasinestimatesofperceptualagreement.JournalofAppliedPsychology67(2),219+/-229.James,L.R.,Jones,A.P.,1974.Organizationalclimate:areviewoftheoryandresearch.PsychologicalBulletin81(12),1096+/-1112.Johnson,W.G.,1985.Accident/IncidentInvestigationManual,2ndEdition.PreparedfortheUSDepartmentofEnergy,DOE/SSDC76-45/27.F.W.Guldenmund/SafetyScience34(2000)215+/-257 255 Jones,A.P.,James,L.R.,1979.Psychologicalclimate:dimensionsandrelationshipsofindividualandaggregatedworkenvironmentperceptions.OrganizationalBehaviorandHumanPerformance23,201+/-

250.Keenan,V.,Kerr,W.,Sherman,W.,1951.Psychologicalclimateandaccidentsinanautomotiveplant.JournalofAppliedPsychology,108-111.Kennedy,R.J.,1997.TheDevelopmentofaHAZOP-basedMethodologytoIdentifySafetyManagementVulnerabilitiesandTheirAssociatedSafetyCulturalFactors.DoctoralthesisUniversityofBirming-hamLee,T.R.,1996.Perceptions,attitudesandbehaviour:thevitalelementsofasafetyculture.HealthandSafetyOctober,1+/-15.Ludborzs,B.,1995.Surveyingandassessing``safetyculturewithintheframeworkofsafetyaudits.In:Mewis,J.J.,Pasman,H.J.,DeRademaecker,E.E.(Eds.),LossPreventionandSafetyPromotioninthe ProcessIndustries1,83+/-92.Lutness,J.,1987.Measuringup:assessingsafetywithclimatesurveys.OccupationalHealthandSafety56, 20+/-26.Mijs,A.A.,1992.Eensociologischevisieopdebetekenisenpopulariteitvanorganisatiecultuur.In:R.vanderVlist(Ed.),Visiesoporganisatiecultuur.Eenmultidisciplinairperspectief.UitgeverijLemmaB.V.,Utrecht.Niskanen,T.,1994.Safetyclimateintheroadadministration.SafetyScience17,237+/-255.Ostrom,L.,Wilhelmsen,C.,Kaplan,B.,1993.Assessingsafetyculture.NuclearSafety34(2),163+/-172.Payne,R.L.,Pheysey,D.C.,1971.G.G.Stern'sorganizationalclimateindex:areconceptualizationandapplicationtobusinessorganizations.OrganizationalBehaviorandHumanPerformance6,77+/-98.Pidgeon,N.F.,1991.Safetycultureandriskmanagementinorganizations.JournalofCross-CulturalPsychology22(1),129+/-140.Pidgeon,N.F.,1997.Thelimitstosafetyculture,politics,learningandman-madedisasters?JournalofContingenciesandCrisesmanagement5(1),1+/-14.Pidgeon,N.F.,1998.Safetyculture:keytheoreticalissues.WorkandStress13,202+/-216.Reason,J.T.,1997.ManagingtheRisksofOrganizationalAccidents.Ashgate,Aldershot.SafetyResearchUnit,1993.TheContributionofAttitudinalandManagementFactorstoRiskintheChemicalIndustry(FinalReporttotheHealthandSafetyExecutive).PsychologyDepartmentUni-versityofSurrey,Guilford.Sanders,G.,Neuijen,B.,1987.Bedrijfscultuur:Diagnosee

Ânbenvloeding.VanGorcum,Assen.Schein,E.H.,1992.OrganizationalCultureandLeadership,2ndEdition.Jossey-Bass,SanFrancisco.

Schneider,B.,1975.Organizationalclimates:anessay.PersonnelPsychology28,447+/-479.

Schuman,H.,Presser,S.,1981.QuestionsandAnswersinAttitudeSurveys.AcademicPress,Orlando.Shye,S.,Elizur,D.,Ho*man,M.,1994.IntroductiontoFacetTheory;ContentDesignandIntrinsicDataAnalysisinBehaviouralResearch.Sage,NewburyPark.Simard,M.,Marchand,A.,1996.Amultilevelanalysisoforganisationalfactorsrelatedtothetakingofsafetyinitiativesbyworkgroups.SafetyScience21(2),113+/-129.Tabachnick,B.G.,Fidell,L.S.,1989.UsingMultivariateStatistics,2ndEdition.HarperCollins,New York.Tatsuoka,M.M.,Lohnes,P.R.,1988.MultivariateAnalysis.TechniquesforEducationalandPsycholo-gicalResearch,2ndEdition.Macmillan,NewYork.Tuli,R.W.,Apostolakis,G.E.,1996.Incorporatingorganizationalissuesintoroot-causeanalysis.Trans-actionsInstitutionofChemicalEngineers74(B),3+/-16.VandeGeer,J.P.,1993a.MultivariateAnalysisofCategoricalData:Theory.AdvancedQuantitativeTechniquesintheSocialSciences,Vol.2.Sage,NewburyPark.VandeGeer,J.P.,1993b.MultivariateAnalysisofCategoricalData:Applications.AdvancedQuantita-tiveTechniquesintheSocialSciences,Vol.3.Sage,NewburyPark.VanHoewijk,R.,1988.Debetekenisvandeorganisatiecultuur:Eenliteratuuroverzicht.M&O,Tijds-chriftvoorOrganisatiekundeenSociaalBeleid1,4+/-46.VanSchuur,W.H.,Kiers,H.A.L.,1994.Whyfactoranalysisoftenistheincorrectmodelforanalyzingbipolarconceptsandwhatmodeltouseinstead.AppliedPsychologicalMeasurement18,97+/-110.

256F.W.Guldenmund/SafetyScience34(2000)215+/-257 Williamson,A.M.,Feyer,A.-M.,Cairns,D.,Biancotti,D.,1997.Thedevelopmentofameasureofsafetyclimate:theroleofsafetyperceptionsandattitudes.SafetyScience25,15+/-27.Zohar,D.,1980.Safetyclimateinindustrialorganizations:theoreticalandappliedimplications.JournalofAppliedPsychology65(1),96+/-102.F.W.Guldenmund/SafetyScience34(2000)215+/-257 257