您的当前位置：首页 2009-Teachers’ stressors and strains A longitudinal study of their relationships

2009-Teachers’ stressors and strains A longitudinal study of their relationships

来源：99网

CORRECTEDJANUARY29,2010;SEELASTPAGE

Teachers’StressorsandStrains:ALongitudinalStudyofTheirRelationships

ArieShirom

TelAvivUniversity

AmalyaOliverandEstherSteinHebrewUniversityofBeitBerl

Theauthorstestedthehypothesizedunidirectionalorbidirectionaleffectsof5typesofteachers’work-relatedstressorsoneachofthe4typesofpsycho-logicalstrain:somaticcomplaints,burnout,andintrinsicandextrinsicjobdissatisfaction.Theauthorsusedstructuralregressionanalysestoanalyzetheresponsesofarepresentativesampleof404highschoolteacherswhohadcompletedbothourT1(beginningofschoolyear)andT2(endofschoolyear)questionnaires.TheauthorsfoundsupportfortheexpectedunidirectionaleffectsoftheT1work-relatedstressorsontherespondents’T2somaticcomplaints(withbaselinesomaticcomplaintscontrolledfor),andalsofortheexpectedunidirectionaleffectsofT1intrinsicandextrinsicjobdissatis-factionontheT2valuesoftheﬁvetypesofstressors(withbaselinestressorscontrolledfor).Onlypartialsupportwasfoundfortheexpectedbidirectionalrelationshipsbetweenthestressorsandteachers’burnout.Theauthorssuggestthatthedirectionalityacrosstimeoftherelationshipsbetweenstressorsandstrainsmaydependontheintrinsicpropertiesofthestrainunderconsideration.

Keywords:teachers’stress,strain,somaticcomplaints,burnout,jobsatisfaction,longitudinaldesign

Acornerstoneofseveralmajortheoriesintheareaofstressresearchisthepostulatethatwork-relatedstressorsareunidirectionalpredictorsofpsychologicalstrain.Thesetheories—discussedbelow—oftendeﬁnepsy-chologicalstrainsratherbroadly,toincludeburnout,somaticcomplaints,and

ArieShirom,TelAvivUniversity;AmalyaOliverandEstherStein,HebrewUniversityofBeitBerl.

CorrespondenceconcerningthisarticleshouldbeaddressedtoArieShirom,FacultyofManagement,Tel-AvivUniversity,POB39010,Tel-Aviv69978,Israel.E-mail:ashirom@post.tau.ac.il

312

InternationalJournalofStressManagement2009,Vol.16,No.4,312–332

StressandStraininTeachers313

differenttypesofjobsatisfaction(e.g.,see:Karasek&Theorell,1990;Siegrist&Rodel,2006).Ourobjectiveinthecurrentstudyistoexamine,againstthistheoreticalview,analternativetheoreticalperspective.Inouralternativetheoreticalperspective,wearguethatthereisaspeciﬁcpathofinﬂuencebetweenwork-relatedstressorsandeachpsychologicalstrain.Thispathofinﬂuencecouldbeeitherofthefollowing:stressorsa`strain,straina`stressors,orareciprocalrelationshipamongthemacrosstime(i.e.,bidirec-tionalinﬂuence).Ourexpectationisthatthecentralrelationalmeaningofeachpsychologicalstrainwilldeterminethedirectionalityofitsrelationshipswithstressorsacrossallstressors.Forbrevity’ssake,wewillrefertoouralternativetheoreticalviewasthe“speciﬁcdirectionalitymodel.”Wetestedourhypothesesamongagroupofprofessionalteachersusingalongitudinaldesign.

TheTheoreticalModel

Theﬁeldofstressresearchisdominatedbytheoreticalmodelsthatpostulateaunidirectionalpositiveeffectofstressorsonpsychologicalstrains:anincreaseinstressorswillbefollowedbyanincreaseinpsychologicalstrains.KarasekandTheorell’sDemand–ControlModel(cf.Karasek&Theorell,1990),accordingtoitscurrentinterpretation(deLange,Taris,Houtman,&Bongers,2003),postulatesthatjobdemands,deﬁnedanalo-gouslytostressorsinthecurrentstudy(cf.Karasek,1979),havemain(direct)effectsonbothpsychologicalandphysiologicalstrain.InthePerson–EnvironmentFittheory(French,Caplan,&Harrison,1982),perceivedstress,deﬁnedasasubjectiveappraisalindicatingthatsuppliesareinsufﬁcienttofulﬁlltheperson’sneeds,leadstopsychological,physical,andbehavioralstrain(Edwards,Caplan,&VanHarrison,1998).IntheJobDemands–Resourcestheory,Demeroutiandcolleagues(Demerouti,Bakker,Nachre-iner,&Schaufeli,2001)refertojobdemandsasthosephysical,psycholog-ical,andorganizationalaspectsofthejobthatrequiresustainedphysical,emotional,andcognitiveefforts.Jobdemandsoftenturnintostressors,whenmeetingthedemandsrequiressustainedefforts(Demeroutietal.,2001),andhaveaunidirectionalinﬂuenceonavarietyofpsychologicalandphysiolog-icalstrains.Eachoftheabovemodelshasstimulatedalargebodyofempiricalresearch.Severalauthorspointedoutthatoneofthemajorlimi-tationsofthesemodelsisthattheyallowonlygeneralpredictions(e.g.,Kasl,1996;Houkes,Janssen,deJonge,&Nijhuis,2001).

ThespeciﬁcdirectionalitymodelwasbasedonLazarus’swork(2001).Lazarus(1999)proposedthateachdiscreteaffectivestatehasacentralrelationalmeaningassociatedwithit.Acentralrelationalmeaningforan

314Shirom,Oliver,andStein

affectivestatereﬂectsasynthesisofpeople’sappraisalsofitsintrinsiccharacteristics(cf.Lazarus,1999,pp.95–97).Forexample,happiness(andprobablyalsosatisfaction)isassociatedwithone’smakingreasonableprogresstowardtherealizationofone’sgoals,whilefeelinganxiousreﬂectsone’sexposuretouncertaintyandexistentialthreats.Wearguethatforeachofthestrainsunderconsideration,itscentralrelationalmeaningwilldeter-minethedirectionalityassociatedwithitsrelationshipswithstressors.Thisargumentiselaboratedbelow,butweﬁrstdeﬁnewhatwemeanbystressorsandstrains.Wefocusedonaspeciﬁcoccupation—professionalteaching—followingBacharachandBamberger’s(1992)suggestionforanoccupation-speciﬁcfocusofresearchexaminingtherelationshipsamongstressorsandstrains.Therefore,inreviewingpastempiricalstudies,weshallfocusonthoseconductedintheteachingprofession.

Astressoroccurswhenanemployeeperceivesanenvironmentaldemandasbeingexceeding,taxing,orthreateningtohisorheradaptiveresources(Lazarus&Folkman,1984).Thus,astressorimpliesbothperceptualandcognitiveoperations.Measuresofgeneraltypesofstressorsbasedonroletheory,likeroleambiguityandconﬂict,wereoftencriticizedasconceptuallyoverlappingwithmeasuresofpsychologicalstrain,renderingtheircorrela-tionstheoreticallyvaluelessandpracticallyinsigniﬁcant(Ganster&Schaubroeck,1991),andthereisempiricalsupportforthistheoreticalposi-tion(e.g.,Jex,Beehr,&Roberts,1992).Weviewtheoccupationalnexusasdeterminingthetypesofchronicstressorsthatanemployeeexperiences(Bacharach&Bamberger,1992;Fletcher&Jones,1993),aviewsupportedinseveralstudiesonteachers’stressors(Kyriacou,2001).Therefore,weconstructedoccupation-speciﬁcstressorscales,describedbelow(seetheMethodsection)forthecurrentstudy.Weassumethatdespiteindividualdifferencesregardinghowstressorsareperceived,individualsrespondinaconsistentwaytoexperientiallysimilarworkenvironments(Bliese&Halv-erson,1996).Basedonthistheoreticalposition,weexpectthesamedirec-tionalityorbidirectionaleffectonstrainacrossthestressorsunderstudy.Wetestedthistheoreticalexpectationbysystematicallycomparinghoweachofthestressorsrelatestoeachindividualstrain.Consequently,foreachstrainunderstudy,weimplementedaconstructivereplicationstrategy.

Thetermpsychologicalstrain(forbrevity’ssake,wewillrefertoitsimplyasstrain)isdeﬁnedasanyenduringdeviationfromnormalresponses,includingnegativeaffectivestates(Frenchetal.,1982,pp.5)thatmayleadtoimpairedphysicaland/ormentalhealth(Suls&Bunde,2005).Wein-cludedburnoutasamajorstrainbecausethereisabodyofevidencelinkingitwithavarietyofdiseasestates(Melamed,Shirom,Toker,Berliner,&Shapira,2006),andwithreducedjobperformance(Taris,2006).Addition-ally,burnoutappearstobeprevalentamongteachers(e.g.,Halbesleben&Buckley,2004).Burnoutisconceptualizedasreferringtoindividuals’feel-

StressandStraininTeachers315

ingsthattheyhavedepletedorexhaustedtheircognitive,physical,andemotionalresources(Melamedetal.,2006).Weincludedsomaticcomplaintsasastrainbecauseitisoftenusedasanindictorofpoormentalhealth(cf.Frese,1985).Jobdissatisfactionisoftenregardedasafocalvariableintheevaluationofthequalityofworkinglife,aswellasapotentpredictorofmentalandphysicalhealth(Faragher,Cass,&Cooper,2005).Jobsatisfac-tionreferstoone’scognitive(evaluative),affective(oremotional),andbehavioralresponsestoone’sjob(Hulin&Judge,2003).Adistinctionwasmadebetweenintrinsicjobdissatisfaction,whichreferstoone’sevaluativeassessmentofone’sself-mediatedrewardsinteaching,suchasdoinganinterestingjob,andextrinsicjobdissatisfaction,referringprimarilytore-wardsprovidedbymanagement,suchaswagesandworkingconditions.ThisdistinctionfollowsHackmanandOldham’s(1980,pp.83–88)theoreticalpropositionthatjobcharacteristicshavedifferenteffectsonintrinsicversusextrinsicjobsatisfaction,apropositionsupportedbyaccumulatedevidence(cf.Judge&Larsen,2001).

MajorHypotheses

Ourﬁrsthypothesiswasthatthestressorsunderstudyareunidirectionalpredictorsofsomaticcomplaints.Somaticcomplaintsisoftenconceptualizedasreferringtoexperiencedunpleasantandpainfulbodilysensations(Watson&Pennebaker,19);therefore,ourreasonsforformulatingthishypothesisfocusonphysiologicalreactionstostressors.First,aperson’sstressors,representingthreats,areoftendirectlyassociatedwithphysiologicalarousal,whichmakesapersonmorevulnerabletotheoccurrenceofbodilysymptomsassessedbyallmeasuresofsomaticcomplaints(Miller,Chen,&Zhou,2007).Second,experiencingstressorsmayaffectthewaypeoplemonitortheirbodiesaswellastheirreadinesstoidentifybodilysymptoms(Watson&Pennebaker,19).Third,stressorsmayimpactavarietyofhealthbehav-iorsandthesechangescouldincreasesubsequentreportingofsomaticcomplaints(Mechanic,1980).Severalpaststudiesonteachers’stressorsandstrainsprovidesupportforourhypothesis.Forexample,inastudythatfollowed63teachersthroughoutthedurationofoneschoolyear,stressorswerefoundtoleadtosubsequentstrain,includingsomaticcomplaints(Bren-ner,So¨rbom,&Wallius,1985).Inyetanotherlongitudinalstudy,Frese(1985)foundworkstressorstohaveaunidirectionalinﬂuenceonlevelsofsomaticcomplaintsamongfactoryworkers.Therefore,weformulatedourﬁrsthypothesisasfollows:

Hypothesis1:Stressorsareunidirectionalpredictorsofsomaticcom-plaints.

316Shirom,Oliver,andStein

Weviewburnoutasrepresentingthecorerelationalmeaningofhavingexperiencedcontinuousdepletioninone’senergeticresources(cf.Melamedetal.,2006).Therefore,wearguethatstressorsandburnoutareprobablycontinuouslyinﬂuencingoneanotheracrosstime.AnanalogouspropositionwasformulatedbyUrsinandErikson(2004),whosuggestedthatcumulativeexposuretochronicstressorsatworkmaydrainanindividual’senergeticresources.WeconsidertherelationshipsbetweenstressorsandburnoutfromtheperspectiveofHobfoll’sConservationofResources(COR)theory(Hob-foll,19;Hobfoll&Shirom,2000)becauseburnoutrepresentsanaffectivereactiontothegradualdepletionofemployees’intrinsicenergeticresources.CORtheoryalsosuggeststhatindividualswhoalreadylackastrongbaseofintrinsicresourcesaremorevulnerabletotheresource-depletioneffectsofexperiencingchronicstressors,andthatthereforeaviciouscycleofresourcedepletionmayarise.Thesetheoreticalviewsledustoexpectthatstressorsandburnoutarebothrelatedinabidirectionalmannerovertime.Mostlongitudinalresearchonburnoutlendssupporttothethesisthatchronicstressorsatworkandburnoutarerelatedinabidirectionalmannerovertime(e.g.,deJongeetal.,2001;deLange,Taris,Kompier,Houtman,&Bongers,2004).Followingtheabovetheoreticalconsiderationsandempiricalevi-dence,weexpectthat:

Hypothesis2:Stressorsandburnoutarerelatedinabidirectionalmannerovertime,continuouslyinﬂuencingoneanother.

Jobsatisfactionisoftenviewedasaconstructwhichsynthesizesandsummarizesemployees’responsestoworkingconditionsandjobcharacter-istics(Staw&Cohen-Charash,2005).Thecoremeaningofjobsatisfactionreﬂectsexperiencedprogresstowardtherealizationofgoals(cf.Lazarus,1999,pp.95–97).Therefore,itisplausibletoassumethatexistinglevelsofdissatisfactionmaycauseindividualstobeselectivelyattentiveto,ortoredeﬁne,environmentaldemandstomakethemconsistentwiththeirbeliefsregardingthesatisfyingordissatisfyingqualityoftheirjobs(James&Tetrick,1986).Forexample,“Iamdissatisﬁed;therefore,myjobmustbestressful”(James&Tetrick,1986).Severallongitudinalstudieshavepro-videdsupportforourhypothesis(Bacharach&Bamberger,1992;Bateman&Strasser,1983).Forexample,BatemanandStrasser(1983)foundthatintrin-sicjobdissatisfactionpredictedaglobalmeasureofjobstressors.However,otherlongitudinalstudiesofstressor–jobsatisfactionrelationsfoundthemtoberelatedinabidirectionalmanner,eachaffectingtheotherovertime(e.g.,Gelsemaetal.,2006),notrelatedovertime(terDoest&deJonge,2006),andwiththestressorasaunidirectionalpredictorofjobsatisfaction(deJongeetal.,2001).Consequently,theevidencesupportingourexpectationisnot

StressandStraininTeachers317

conclusive.Primarilybasedontheabovetheoreticalconsiderations,weformulatedthefollowinghypothesis:

Hypothesis3:Intrinsicandextrinsicjobdissatisfactionareunidirectionalpredictorsofstressors.

Aboveweassumed,basedonBlieseandHalverson’s(1996)study,thatworkstressorswhicharerelevanttotheemployeesunderstudy—thatis,reﬂecttheiractualdemandingandthreateningworkexperiences—maintainaconsistentpatternofrelationshipswitheachstrain.However,eachstressorcouldbecharacterizedbyauniquepatternofrelationshipswithstrainsrelativetootherstressors(cf.referencesin:McCarthy,Lambert,O’Donnell,&Melendres,2009).Therefore,wetestedourassumptionaccordingtothefollowinghypothesis:

Hypothesis4:Theunidirectionalandbidirectionalexpectationsfortherelationshipsamongstressorsandthestrainsunderstudy—somaticcom-plaints,burnout,andextrinsicandintrinsicjobsatisfaction—areconsis-tentacrossrelevantstressorsthatrepresentthreateningdemandscontex-tualizedintheworkexperiencesofrespondents.

METHOD

ResearchSettingandRespondents

Arandomprobabilitysampleof30highschoolswasdrawnfromthesamplingframeof240Hebrew-language,nonvocational,andnonboardinghighschoolslocatednorthofBeer-ShevainIsrael.Eachschoolinthesampleprovideduswithastafflist.Allteacherswhotaughtlessthan5hrweeklyortaughtinmorethanoneofthesampleschoolswereexcludedfromtheﬁnalsample.Outoftheremainingteachersemployedinthe30highschoolsatthebeginningoftheschoolyear,about60%ofthosewhosenamesappearedonthestafflists—1,048respondents,representing95%ofthosewhopartici-patedintheﬁrstwhole-schoolmeetingatTime1(T1)—completedthestudy’sT1questionnaire.AtTime2(T2),762teachers,representing45%ofthosewhosenamesappearedonthestafflists,completedtheT2question-naire.BecausebothT1andT2questionnaireswerecompletedanonymously,wematchedsomehighschoolT1andT2respondentsbasedontheirpos-sessingtheexactsociodemographicproﬁleinbothwaves,excludingalldubiousorunclassiﬁableresponsesduetomissingdata.ThesematchingproceduresresultedinNϭ404,thesizeoftherepeatedmeasuresgroup.

318Shirom,Oliver,andStein

BecauseoftheattritionfromeitherT1orT2tothematchedsampleusedhere,wesystematicallycomparedtherepeatedmeasuresgroupwiththeone-timerespondentsatbothT1andT2forthecharacteristicsofage,gradestaught,professionalandschooltenure,subjectmatterandnumberofclassestaught,jobduties,joblevel,numberofweeklyhourstaught,familystatus,gender,andeducationalattainment.ThesesamecharacteristicswerealsousedtomatchT1andT2responses.Wefoundthattherepeatedmeasuresgroupshowednodifferences(basedontwo-sidedt-tests)fromeitherofthetwosamplesofone-timerespondentsin20ofthe22comparisons.

InIsrael,highschoolteachersthroughoutthecountryarecoveredbythesamecollectiveagreement;consequently,thesamecontractualobligationsandjobrulesappliedtoallparticipants,regardlessofwhopaidtheirsalaries(inmostcases,itwasthemunicipalityinwhichtherelevanthighschoolwaslocated).Respondentstaughtanaverageof20hrperweek(thefull-timeteachingloadforahighschoolteachervariesbetween20and24weeklyhr).Onaverage,respondentswere36yearsold,85%ofthemweremarried,andofthelattercategory86%weremarriedwithchildren.Respondentswerepredominantlywomen(82%);theirmeanseniority(lengthofserviceinteaching)was11years,andmost(60%)ofthemwereeitherstudyingtowardoralreadyhadabachelor’sdegree.

Procedure

T1dataweregatheredbyaskingallteacherswhoattendedtheﬁrstall-schoolpedagogicalcouncilatthebeginningoftheschoolyeartocom-pletethestudy’squestionnaire,towhich95%ofthemagreed.Respondentswereguaranteedanonymityandconﬁdentialityofindividualresponses.TheT2questionnairewasadministeredapproximately7monthslater.Itwasmailedtothehomeaddressesofalltheteacherswhowerestillonthestafflistsaboutamonthbeforetheendoftheschoolyear.TeacherswereaskedtocompletetheT2questionnairesontheirowntimeandreturnthemtotheprincipalinvestigators’universityaddress,usingtheprepaid,preaddressedenvelopeenclosedwiththequestionnaire.

Measures

Allitemsinthequestionnairewereoftheclosed-endtype.Unlessotherwisenoted,a5-pointresponseformatwasused.Multipleitemindiceswereconstructedbycombiningsingleitemsthatmeasurethesamevariable.Arespondent’sscoreoneachindexwasobtainedbycomputingthemeanof

StressandStraininTeachers319

Table1.IntercorrelationsAmongStressVariablesandTheirMeans,StandardDeviations,

andReliabilityCoefﬁcients

T1Variable

1.Heterogeneousclasses2.Discipliningstudents3.Home-workconﬂict4.Physicalconditions5.ExtracurriculardutiesT2MeanT2SDT2Alphaa1()333927283.191.05.55

237(66)1826Ϫ04ء2.49..80

32825(51)22273.51.90.62

4193225(59)292.961.05.55

524143040(67)2.81.93.66

Mean3.382.693.072.942.87

T1SD1.02.901.231.11.97

T1Alphaa.52.81.62.61.66

Note.DataabovethediagonalcamefromtheT1survey.Parenthesizedentriesaretest-retestreliabilitiesoracross-timestabilitycoefﬁcients.DatabelowthediagonalcamefromtheT2survey.Nϭ400.Withtheexceptionofthestarredcoefﬁcient,allcorrelationsarestatisticallysigniﬁcantatthepϽ.05level.Decimalpointshavebeenomittedfromthecorrelationcoefﬁcients.aAlpharepresentsCronbach’sinternalconsistencyreliability.

her/hisresponsestotheitemsintheindex.Internalconsistencyreliabilities(Cronbach’sAlpha)forallstressorvariablesarepresentedinTable1;thoseforthestrainvariablesarepresentedinTable2.Belowwedescribehowwedevelopedthemeasuresofstressorsusedinthecurrentstudy.

DevelopingtheMeasuresofStressors

Toaccordwithourconceptualizationofstressors,theyshouldrefertoafrequentlyexperiencedenvironmentaldemandthatislikelytoexceedthe

Table2.IntercorrelationsAmongStrainVariablesandTheirDescriptiveStatistics

T1Variable1.Somaticcomplaints2.Burnout3.Intrinsicjobdissatisfaction4.ExtrinsicjobdissatisfactionT2MeanT2SDT2Alphaa1(61)4520162.50.86.66

240(70)35243.351.07.86

31638(56)392.67.71.56

4113231(67)3.08..54

Mean2.543.342.543.01

T1SD.971.12.66.92

T1Alphaa.70.86.59.55

320Shirom,Oliver,andStein

respondents’adaptiveresources(cf.Lazarus,2001).Anexaminationofavailablemeasuresofstressorsintheteachingprofession(cf.Brenneretal.,1985;Kyriacou,2001)ledustoconcludethatnoneofthemprovidesanadequateoperationalizationofourconceptualizationofperceivedstressors.Therefore,basedonboththeoreticalconsiderationsandpreviousresearch,weconstructedtheteachers’stressorsmeasuresspeciﬁcallyforthisstudy.

Considerabledevelopmentaleffortwentintotheconstructionofthemeasuresofteachers’stressors.Inapilotstudy,53juniorhighschoolteacherscompletedopen-endedquestionnairesinwhichtheywererequestedtodescribestressfulsituationsexperiencedatwork.The520situationsthusobtainedwereexamined,andduplicate(identical)itemswereremoved.Nineveteranhighschoolteachersthenevaluatedtheremainingitemsonthesolecriterionofrelevancytotheirteaching,andonlyitemsconsistentlyjudgedas“veryhighlyrelevant”or“highlyrelevant”bysevenormoreofthesejudgeswereretained.Eachoftheremaining141situationswereevaluatedbyinformedjudges(21highschoolteacherswith5ormoreyearsofseniority),accordingtothecriteriaoffrequency,intensity,andimpactonteachers’performance.Onlythestressoritems(nϭ44)thatreﬂectedongoingcondi-tionsintheschools,ratherthanstressincidents,andwhichwereconsistentlyratedasoccurringfrequentlyorveryfrequently,wereselectedforfurtheranalysis.Theresultantvariableswereintercorrelatedandthecorrelationmatrixwasthenfactoranalyzed.Thisledtotheconstructionofthestressormeasuresdescribedbelow.Wecomparedthelistofstressors—describedbelow—thatemergedfromourconstructionprocesswithotherinventoriesofteachers’stressors.Theseinventorieswerebasedonqualitative(Cox,Boot,Cox,&Harrison,1988),quantitative(Boyle,Borg,Falzon&Baglioni,1995),orsynthesizing(Borg,1990)researchmethodologies.Overall,foreachofthestressorslistedbelow,therewasacloselyrelatedscalecon-structedinatleastoneoftheseinventories,thusprovidingsupportforthevalidityofourprocessofconstructingmeasuresofteachers’stressors.

MeasuresofStrain

Allstrainitemsinthequestionnairewereprecededbyarequestthatrespondentsreporthowoften,sincethebeginningoftheschoolyear,theyhadexperiencedeachofthefeelingsinthiscategoryofitems.Toassessburnout,weusedtheShirom-MelamedBurnoutMeasure(SMBM),validatedinseveralpaststudies(Melamedetal.,2006;Shirom&Melamed,2006).IntheSMBM,burnoutisdeﬁnedasanaffectivestate,comprisedofemotionalexhaustion,physicalfatigue,andcognitiveweariness,whichdenotethedepletionofenergeticresourcesresultingfromcumulativeexposureto

StressandStraininTeachers321

chronicworkandlifestresses(cf.Melamedetal.,2006).Forexample,respondentswererequestedtoindicatehowoftentheyhadfeltthefollowing:(a)feeling“usedup”attheendoftheworkday;(b)lackingtheenergytogotoworkinthemorning;(c)incapableofdealingwiththeirpupils’problems;(d)burnedoutintheirteachingjobs;and(e)emotionallydrainedatwork.TheSMBMconsistedof12items,andtheresponseformatrangedfromalmostnever(scalevalueof1)toalmostalways(scalevalueof7).ThemeasureofsomaticcomplaintsaswellasthemeasuresofintrinsicandextrinsicjobsatisfactionwerebaseduponthoseusedattheInstituteofSocialResearch,UniversityofMichigan(Frenchetal.,1982),extensivelyusedinpastresearch(cf.Judge,Bono,Thoresen,&Patton,2001)andvalidatedinpaststudiesonteachersinIsrael(e.g.,Shirom&Mazeh,1988).Themeasureofsomaticcomplaintsincludedﬁveitems.Respondentswereaskedtoreport,ona7-pointscale,howoftentheyhadexperienced(a)headaches,(b)inabilitytoconcentrate,(c)colds,(d)sorethroats,and(e)dizziness.Anintrinsicjobdissatisfactionindexwasformedbycombiningthreeitems:dissatisfactionwith(a)theautonomylevel,(b)theopportunitiesforprofes-sionalgrowth,and(c)theopportunitiesfortheimplementationofvaluededucationalgoalsintheirjobs.Anextrinsicjobdissatisfactionindexwasformedsimilarlybycombiningthreeitems,forexampledissatisfactionwith(a)thesalaryand(b)theworkingconditionsintheirjob.Alljobdissatis-factionitemsweregaugedona5-pointratingscale:1ϭverysatisﬁed;5ϭverydissatisﬁed.

MeasuresofStressors1Respondentswererequestedtoindicatetowhatextenteachofthefollowingstressfulsituationsexistedintheirteachingjob,accordingtoa5-pointresponseformatrangingfromtoaverysmallextent(scalevalueof1)totoaverylargeextent(scalevalueof5).Heterogeneousclasseswasassessedbytwoitemsthatreceivedhighscoresforteachersreportingthattheyhadtoteach(a)inheterogeneousclassesinwhichitwasdifﬁculttoadaptthelevelofinstructiontostudents’instructionalneeds,and(b)inlargeclassesinwhichitwasdifﬁculttoprovideindividualattentiontostudents.Discipliningstudentswasmeasuredbyﬁveitemsassessingtheamountoftimeandeffortthatrespondentshadtoinvestin(a)discipliningstudents,(b)teachingunmotivatedstudents,(c)dealingwithdisruptivestudents,and(d)and(e)maintainingorderandquietnessintheirclasses.Home–workconﬂictincludedtwoitemsdealingwiththeextenttowhichrespondentshadtograde

1Theexactwordingoftheitemsusedisavailablefromtheauthorsuponrequest.

322Shirom,Oliver,andStein

examsorotherstudentassignmentsathome,whileothermembersofthefamilyweredemandingattention.Physicalconditionsweremeasuredbytwoitemsaskingrespondentstoindicatetowhatextentclassroomtemperatureoroutsidenoiseinterferedwiththeirteachingactivities.Extracurriculardutiesweregaugedbythreeitemsassessingtheextenttowhichrespondentshadto(a)supervisestudentsduringbreaks,(b)investaninordinateamountoftimeinpaperwork,and(c)handleadditionaldutiesatschool.Twoadditionalstressormeasures,adaptabilityrequirementsandparent–teacherconﬂict,wereomittedfromthestructuralregressionanalysisreportedherebecauseof(a)relativelylowinternalconsistencyreliabilities,and(b)signiﬁcantlylowertest–retestorstabilitycoefﬁcients,relativetotheotherstressormeasures.

StatisticalAnalyses

AlltestsofstatisticalsigniﬁcancereportedinthisstudyusedapϽ.05levelofsigniﬁcance,andallttestsusedatwo-tailedtest.Thehypothesesweretestedbystructuralregressionanalysis(SRA),whichisconsideredsuperiortocross-laggedcorrelationanalysis(Finkel,1995;Twisk,2003).OnedisadvantageofSRAisthatsynchronouscausaleffectscannotbeidentiﬁedbythisapproach(Zapf,Dormann,&Frese,1996);however,inthecurrentstudytheseeffectswerenotafocusofconcern.InTable3,weprovidefourcorrelationsandtwostructuralregressioncoefﬁcientsforeachpairofstressor͑xi͒andstrain͑yi͒measures.Theﬁrsttwocorrelations͑x1y1andx2y2͒arethestatic(contemporaneous)correlationsforT1andT2,respectively.Next,weprovidethetwocross-laggedcorrelations(rx1y2andrx2y1)betweentheT1valueofonevariableandtheT2valueoftheothervariable.Thesecorrelationsareprovidedasacross-methodvalidationofthestructuralregressions.Finally,werepresenttheeffectofastressoronastrainbythestandardizedregressioncoefﬁcient(b2intheequationbelow)resultingfromtheregressionofaT2strainmeasureonaT1stressor,theT1strainmeasurebeingtheﬁrstpredictor͑Y2ϭaϩb1y1ϩb2x1ϩe͒intheregressionequation.Similarly,theinﬂuenceofastrainonastressorisrepresentedbythestandardizedregressioncoefﬁcientresultingfromtheregressionofaT2stressoronaT1strain,withtheT1stressormeasureintroducedastheﬁrstpredictorintheregressionequation.Inouranalyses,weassumedthattheﬁrstpredictorofaT2criterionshouldbeitsT1value.Undertheusualassumptionsunderlyingregressionanalysis(e.g.,linearity,additivity,homoscedasticity),weinterpretedthestructuralregressioncoef-ﬁcientsinamannersimilartoBatemanandStrasser(1983);thatis,anonzerovalueofarelevantstructuralregressioncoefﬁcientparameterisindicativeofasigniﬁcantunidirectionaleffect.Signiﬁcantnonzerovaluesofbothstruc-

StressandStraininTeachers323

Table3.StructuralRegressionResultsforStressandStrainVariables

Staticcorrelations

Variable

Strain:intrinsicjob

dissatisfaction

HeterogeneousclassesDiscipliningstudentsHome-workconﬂictPhysicalconditionsExtracurriculardutiesStrain:extrinsicjob

dissatisfaction

HeterogeneousclassesDiscipliningstudentsHome-workconﬂictPhysicalconditionsExtracurriculardutiesStrain:somaticcomplaintsHeterogeneousclassesDiscipliningstudentsHome-workconﬂictPhysicalconditionsExtracurriculardutiesStrain:burnout

HeterogeneousclassesDiscipliningstudentsHome-workconﬂictPhysicalconditionsExtracurricularduties

rx1y1.26.19.24.13.08ء.17.11.18.14.23.16.17.19.10.06ء.25.28.30.20.25

rx2y2.22.23.17.23.23.16.13.18.13.13.23.20.25.20.20.29.30.19.30.29

Cross-laggedcorrelationsrx1y2.17.21.11.09.04ء.17.07ء.07ء.06ء.17.27.23.28.18.20.27.24.18.15.16

rx2y1.22.22.22.20.03ء.19.13.20.20.28.13.16.13.11.10.21.21.21.20.26

Structuralregressioncoefﬁcientsabx1y2(y1).02(n.s)

—ء.03(n.s).02(n.s)

—.06(n.s.)

—Ϫ.06(n.s.)Ϫ.02(n.s.).02(n.s.).16ء(.36ء,.03ء).12ء(.36ء,.02ء).15ء(.37ء,.02ء).06ء(.37ء,.02ء).14ء(.37ء,.02ء)

by1x2(x1).09ء(.41ء,.01ء)

—ء.09(.26ء,.01ء).13ء(.34ء,.02ء)

—.08ء(.41ء,.01ء)

—

.12ء(.26ء,.01ء).13ء(.34ء,.02ء).13ء(.45ء,.02ء).05(n.s.).06(n.s.).02(n.s.).02(n.s.).05(n.s.)

.08ء(.49ء,.01ء).07(n.s.)

————.00(n.s.).09ء(.35ء,.01ء)Ϫ.03(n.s.).10ء(.44ء,.01ء)

Note.Nϳ400.Two-tailedtestsofsigniﬁcancewereused.Withtheexceptionofthestarred

ones,allcorrelationcoefﬁcientsarestatisticallysigniﬁcantatthepϽ.05level.n.s.ϭnotsigniﬁcant.aTheregressioncoefﬁcientsarethecross-laggedstructural(orcausal)parameters.Foreachcell,theﬁrstreportedvalueisthatofthestandardizedregressioncoefﬁcient,followedbythe⌬R2oftheﬁrstandsecondpredictorsintherelevantregression,ifsigniﬁcantlydifferentfromzero(inparentheses).Emptycellssignifythatthevaluesofallcoefﬁcientsinthatcellwerenotsigniﬁcantlydifferentfromzero.

turalregressioncoefﬁcientsindicatebidirectionalcausation.Forseveralpairsofstressor–strainmeasures,wecalculatedthesigniﬁcanceofthedifferencebetweentwostructuralregressioncoefﬁcientsusingCohen’sFormula9(Cohen,1983,p.81).

RESULTS

Themeanandstandarddeviationvaluesofstressorandstrainmea-suresarepresentedinTables1and2,respectively.Therewereno

324Shirom,Oliver,andStein

differencesinthemeanlevelsofthestressorsandstrainsbetweenT1andT2.WedeterminedthisfactusinganoverallF-Testofthesigniﬁcanceofthedifferencebetweentwopairsofvectorsforthestressorsandstrains,respectively.WhilethelackofasigniﬁcantmeanT1–T2differenceinthestressorsandstrainsisofinterest,thefocusofourstudywasonrela-tionshipsamongvariablesandnotontheirmeanlevels.InTables1and2,wealsopresenttheintercorrelationsamongthestressorvariablesandamongthestrainvariables,respectively.InTable1,themeancorrelationamongtheﬁvestressorswas.25and.27forT1andT2,respectively,thusindicatingthatonaveragetheysharedlessthan7%oftheirvariance.Theminimaloverlapfoundamongtheﬁvestressorsprovidedempiricalsup-portforourdecisionnottocombinethemintoanoverallmeasurerepresentingchronicstressors.Table2displaystheintercorrelationsamongthefourstrains.Themeancorrelationamongthestrainswas.28and.30atT1andT2,respectively.Inaddition,thecorrelationbetweenintrinsicandextrinsicjobsatisfactionwas.31and.39atT1andT2,respectively,empiricallysupportingourdecisiontoregardthemasrep-resentingtwodistincttypesofstrain.Thestaticandcross-laggedcorre-lationsamongthestressorsandthestrainsarepresentedinTable3.

OnthediagonalsofTables1and2,weprovidethestabilitycoefﬁcientsortest–retestcorrelationsamongthesamevariableatT1andT2.Thesestabilitycoefﬁcientsareanalogoustothosereportedinameta-analyticstudy(Conley,1984),andingeneralindicatethatabout60%ofthevarianceinanyT2stressororstrainisnotduetoitsT1value.Therefore,therewasasufﬁcientamountofvariabilityinanyT2variablenotaccountedforbyitsT1level.

InorderforHypothesis1tobesupportedbytheresultsofthestructuralregressiontests,signiﬁcantbetasshouldbefoundforeachofthehypothe-sizedcross-laggedpredictors.Thatis,fortheheadingofTable3labeled“StructuralRegressionCoefﬁcients,”forallstressors,theregressioncoefﬁ-cientsinthecolumnlabeledbx1y2͑y1͒—allcoefﬁcientsshouldbesigniﬁcant.Furthermore,againforallthestressors,thebetasrepresentingaunidirec-tionalinﬂuenceintheoppositedirection,appearinginthecolumnlabeledby1x2͑x1͒,shouldbeinsigniﬁcantlydifferentfromzero.ResultsofthecorrelationandregressionanalysisconductedtotestthishypothesisaresummarizedinTable3.Theresultsprovidedstrongsupportforourﬁrsthypothesis,predictingaunidirectionaleffectofstressorsonsomaticcom-plaints.AllﬁveT1stressormeasuressigniﬁcantlypredictedT2levelsofthecriterion,andconcomitantlyallregressioncoefﬁcientsexaminingthepossi-bilityofreversecausationwerestatisticallyinsigniﬁcant.

WetestedHypothesis2,expectingbidirectionalrelationshipsbetweenthestressorsunderstudyandburnout,usingthesamecomparisonsexempli-

StressandStraininTeachers325

ﬁedabove,andfoundthatitwasnotconﬁrmed.Fortwostressors,physicalconditionsandextracurricularduties,T1burnoutwasfoundtobeasigniﬁ-cantpredictoroftheirT2levels,whiletheoppositepredictionwasnotsigniﬁcant.Forthestressorofheterogeneousclasses,theT1valueofthisstressorsigniﬁcantlypredictedT2burnout,butthereversecausation,fromT1burnouttoT2heterogeneousclasses,wasnotsigniﬁcant.

TheﬁndingspresentedinTable3providestrongsupportforthethirdhypothesisthatexpectedextrinsicjobdissatisfactiontobeaunidirectionalpredictorofthestressors.AsisevidentfromTable3,fouroutoftheﬁvestressorswerepredictedatT2byT1extrinsicjobdissatisfaction,whilenoneofthereversepredictions,fromT1stressorstoT2extrinsicjobdissatisfac-tion,werestatisticallysigniﬁcant.Toasomewhatlesserextent,thethirdhypothesiswasalsosupportedwithregardtotheunidirectionaleffectsofintrinsicjobdissatisfactiononthestressors.Forintrinsicjobdissatisfaction,theexpectedunidirectionaleffectwasfoundforthreeoutoftheﬁvestressors.Wethereforeconcludedthattheseresultsprovidemodest,butnotuniform,supportforourthirdhypothesisinthatbothtypesofjobdissatisfactionwerefoundtobeunidirectionalpredictorsofmosttypesofstressorsconsidered,whilethereversecausaldirectionwasnotfoundtobesigniﬁcant.

AsisevidentfromTable3,thereisconsiderablesupportforthetheo-reticalviewandforourfourthhypothesisthatpredicted,formostofthestressorsunderinvestigation,aconsistentdirectionalityoftheirrelationshipswiththreeoutoffourstrains:intrinsicandextrinsicjobdissatisfactionandsomaticcomplaints.Theresultsoftheconsistencyexpectation,asformulatedinHypothesis4,weremixedinreferencetothefourthstrain,burnout.

Onanexplorativebasis,weconstructedseparateversionsofTable3formaleandfemaleteachers.2Withoneexception,theunidirectionaleffectsdescribedaboveremainedunchanged.Theexceptionwastheroleofextrinsicjobdissatisfactionformaleteachers;whoseresultsappeartosupportthebidirectionalcausationhypothesis,ratherthanHypothesis3.Itshouldbenotedthatmaleteachersrepresentedasmallminorityoftherespondents,andthereforewedidnotregardthissingleﬁndingassupportiveofagenderdifferenceinourresults.

DISCUSSION

Conductingresearchonstressor–strainrelationshipsisacomplexanddifﬁcultendeavor,primarilyduetofourmajorcharacteristicsofthisﬁeldof

Detailedresultsinvolvingthecomparisonofmalevs.femalerespondentsareavailablefromDr.Shiromuponrequest.

2326Shirom,Oliver,andStein

research.First,itisamultidisciplinaryﬁeld,andthenumberofavailablestudiesisveryhigh.Second,acrossthemanydisciplinesrepresented,thereisstillalackofconsensusasconcernsdeﬁningthetermsstressorandstrain(e.g.,Ursin&Eriksen,2004).Third,thereareethicalconstraintsrelatedtoconductingexperimentalstudies.Finally,upuntilnow,researchhascon-sistedmostlyofcross-sectionalstudies.Thesesalientcharacteristicsoftheﬁeldexplainourstrategicdecisiontofocusontestingspeciﬁcpathsofinﬂuencepredictedtocharacterizestressor–strainrelationshipsamongteach-ers.

Ourstudycontributestotheﬁeldofstressresearchinseveralways.First,ourmajorﬁndingssupportthetheoreticalmodel.Thus,wewereabletosupporttheexpectedunidirectionaleffectsoftheT1work-relatedstressorsontherespondents’T2somaticcomplaints(withbaselinesomaticcomplaintscontrolledfor).Moreover,wefoundconsiderablesupportfortheexpectedunidirectionaleffectsofT1intrinsicandextrinsicjobdissatisfactionontheT2valuesoftheﬁvetypesofstressors(withbaselinestressorscontrolledfor).Onlypartialsupportwasfoundfortheexpectedbidirectionalrelation-shipsbetweenstressorsandteachers’burnout,andthereforeweconsideredthishypothesisnottobeconﬁrmed.

Overall,theresultsofourstudytendtosupportthisstudy’sversionofthespeciﬁcitymodel.Besidesburnout,eachofthestressor–strainrelationshipsinvestigatedappearstobegovernedbyaspeciﬁcanduniquedirectionalityofinﬂuence.Wewouldliketoarguethatforsomestrains,stressorsmaypredicttheirsubsequentlevels.Forothertypesofstrains,thereverseholdstrue;andforyetotherstressor–strainlinkagesourresultstendtobesupportiveofastressor–speciﬁcpatternofinﬂuence.Thespeciﬁcityapproachtostressor–strainrelationsisinconsistentwithsomeofthemajortheoreticalframe-worksadvancedinthisﬁeld,liketheKahnetal.framework(Kahn&Byosiere,1992),andtheapproachregardingperson–environmentmisﬁtsasstressorsthathaveaunidirectionalinﬂuenceonstrain(forarecentreviewofthisapproach,seeEdwards,Cable,Williamson,Lambert,&Shipp,2006).Forexample,aswasnotedbyseveralmeta-analyticstudies,mostpaststudiesontherelationsbetweenstressorsandjobsatisfactionwerecross-sectionalstudiesbaseduponthetheoreticalassumptionthatstressorsareunidirectionalpredictorsofjobsatisfaction(cf.Jackson&Schuler,1985).Fromatheoreticalperspective,weargue,basedonourﬁndings,thatresearchinthisarea,basedonnonspeciﬁcprincipleslikethosedescribedearlier,mayobscuremorespeciﬁcrelationsbetweenstressorsandparticulartypesofstrains.Fromanempiricalperspective,ourspeciﬁcityapproachisconsistentwiththemajorﬁndingsofarecentmeta-analyticstudyofstress–strainrelationshipsbasedoncross-sectionalstudies(Ortqvist&Wincent,2006).

StressandStraininTeachers327

StrengthsandLimitations

Thestudy’sdesignhadseveralstrengths.Stressorandstrainweremea-suredonthesameindividualsatboththebeginningandtheendofaschoolyear,whichconstitutesanaturalcycleintheworklifeofteachers.Substantialdevelopmentaleffortwentintoconstructingoccupation-speciﬁcandconcep-tuallyrobuststressormeasures.Incontrasttoseveralpastlongitudinalstudiesonstressor–strainrelations(cf.Zapfetal.,1996),wesystematicallycom-paredunidirectionaleffectsinbothdirections.Ourﬁndingswereobtainedaftertestingmostofourhypothesesacrossﬁvedifferenttypesofstressors,whichonlyslightlyoverlapped;aswenoted,onaveragetheysharedonlyabout7%oftheirvariance.Therefore,theuniformityoftheresultsobtained,asisevidentbytheexamplepresentedforourtestingofHypothesis1,substantiallyincreasesourconﬁdenceintherobustnessoftheourmajorﬁndings.Instressresearch,exclusiverelianceonself-reportmeasuresofbothstressorsandstrainsoftenevokesthecriticismthattheassociationsobtainedmaybeartiﬁciallyinﬂatedduetoavarietyofartifactsandresponsebiases,includingcommonmethodvariance(cf.Doty&Glick,1998).However,arecentsummaryoftheevidence(Spector,2006)castsseriousdoubtoverthebeliefthatthemethoditselfproducessystematicvarianceinobservationsthatinﬂatecorrelationstoanysigniﬁcantdegree.Furthermore,themethodsofanalysisappliedinthisstudy,anditslongitudinaldesign,obviatetoaconsiderableextenttheconfoundingfactorsthatmaycontaminatecross-sectionalstudiesofpercept–perceptrelationships.Finally,ourmethodofanalysis,inwhichtheﬁrstpredictorofeachcriterionwasitsT1value,ineffectcontrolledfordispositionalfactors,likeneuroticismandnegativeaffectivity,whichpresumablyhadalreadyaffectedtheleveloftheT1predictor.

Ourﬁndingsshouldbeconsideredwiththefollowingreservationsandcaveatsinmind.First,amajorcriticismoflongitudinalresearchisthattheinitialmeasurementdoesnotreﬂectano-causebaselineora“steadystate”andthatthetimelagchosenisnotoptimal(Mitchell&James,2001).Inourdesign,theﬁrstassessmentwasconductedsoonaftertheteachersreturnedtoschoolfromtheirlongsummervacation,whilethesecondassessmenttookplaceattheendoftheschoolyear.Therefore,wearguethatourT1andT2representoptimalchoices.Furthermore,theperiodoftimeseparatingourT1andT2wasfoundtobeoptimalforidentifyingstressor–strainpatternsofrelationshipsinseveralmultiwavestudies(deLangeetal.,2004;Dormann&Zapf,1999).Second,theinﬂuenceﬂowsweidentiﬁedmaybespuriousbecausetheywereinducedbyvariablesthatremainexogenoustoourmodel.ThefactthatwecontrolledfortheT1levelofthecriterionineachofourstructuralregressionsprobablyreducedtheinﬂuenceofdispositional

328Shirom,Oliver,andStein

confounders,suchasneuroticismornegativeaffectivity,whichhadalreadyinﬂuencedthecriterionatT1.Still,wedidnotexamineotherpossiblecausesorpotentialmoderatorsofstrain–stressorlinkages.Inthissense,theresultsoftheanalysesperformedprovidesomeempiricalsupportfortheunidirec-tionalinferencesmade;wemakenoattempttoproveanyofthem.Structuralregressionanalysis,theanalyticmethodused,doesnotultimatelyprovecausation.Aswasnoted,weusedthismethodtotesttemporalpatternsinthestressor–strainrelationships.Third,ourtwo-wavedesignisbutaninferioralternativetoatruetime-seriesdesignwithmanydatacollectionpoints,whichallowsfortestingtherateofchangeandnonlinearchangepatterns(Mitchell&James,2001).

Wewouldliketorefertoseveralcaveatsthatneedtobeconsideredinregardtoourﬁndings.First,threeoutofourﬁvemeasuresofstressorshadrelativelylowreliabilities,representinganotherlimitationofourstudy.Theselowreliabilitiesprobablymilitatedagainstourabilitytosupporteitheroftheformulatedhypotheses,becausethestandarderrorofmeasurementisafunctionofscalereliability.Cronbach’salpha,ourcoefﬁcientofreliability,isverymuchafunctionofthenumberofitemsinthescale,andinourstudymustbeinterpretedtakingintoaccountthatforthreemeasuresofstressors,onlytwoitemswereused,andforthem,the␣obtainedrepresentsthelowerboundofwhatcouldbeachievedunderoptimalconditions(Sijtsma,2009).However,therelativeconsistencyoftheresultsobtainedfortheﬁvestressorscouldbeconsideredasaseriesofconstructivereplications,supportingthepredictivevalidityofthereportedﬁndings.Yetanothercaveatconcernsthemagnitudeofthecross-laggedcorrelationsobtained,mostlyaround.20.Thesameorderofmagnitudeofcross-laggedcorrelationswasobtainedbyotherinvestigatorsofstressorandstrainassociations(Bateman&Strasser,1983;Frese,1985).Ameta-analyticstudyof1,861correlationsfromlargelycross-sectionalstudiesofstressor–strainrelationships(Crampton&Wagner,1994)foundanaveragecorrelationof.07(adjustedforsamplingerror),andareviewofstudiesusingobjectivestressorstopredictstrainsfoundthatthetruecorrelationsweretypicallybetween.20and.30(Semmer,Zapf,&Greif,1996).AsFrese(1985)noted,however,evencorrelationsof.20maybehighlyimportantwhenonelooksatthesample’sextremes.Rosenthal,Rosnow,andRubin(2000)havedescribedseveralexamplesofkeyclinicaldecisionstakeninmedicinebasedoncorrelationssmallerthan.05.Manyadditionalvariablescouldpossiblyinﬂuencethechangeinteachers’stressorsandstrainsovertime,includingphysicalconstitution,healthbehaviors,environmentalfactors,andone’savailableandutilizedcopingresources.

Whatcouldbesomeofthepossiblereasonsforthelackofsupportfoundforourthirdhypothesis,concerningburnout?Burnoutcapturesthedissipa-tionofpersonalenergeticresources.Hobfoll’s(19)conservationofre-sourcestheorysuggeststhatlossofresourcesmaybecentraltostressor

StressandStraininTeachers329

responses.Itcouldbethatforsomestressors,likephysicalconditionsandextracurricularduties,thelossofrelevantresourcesimpingeuponateacher’scopingresourcesattheverystartoftheschoolyear,thusmakingT1burnoutaneffectivesummarymeasureofthecontemporaneouseffectsofenviron-mentalandcontextualstressors.Therefore,T1burnoutwasfoundtobeasigniﬁcantpredictoroftheT2levelsofthesestressors.Forothertypesofstressors,likeheterogeneousclasses,theeffectofthestressoronateacher’spersonalcopingresourcestakestimetoaccumulateandonlybecomesapparentattheT2levelsofburnout.Weofferthisinterpretationofthedisconﬁrmationofthethirdhypothesistofutureresearchersinthisarea.

Wesuggestthattheﬁndingsofourstudyhavesomedirectimplicationsforfutureresearch.ThesubstantialsupportthatourﬁndingsprovideforthehypothesisthatT1stressorspredictT2somaticcomplaintsisofimportanceinadvancingourunderstandingoftheetiologyofcertaindiseases.Thepremiselongheldamongresearchersthatthereisadeﬁniteandgeneralcausalorderleadingfromcognitionsofjobstoaffectiveresponsestojobs(cf.James&Tetrick,1986)needstobereexamined.Futureresearchersintheareaofstressor–strainlinkagesshouldexaminethedirectionalityoftheirrelationshipbasedontheintrinsicpropertiesofthetypeofstrainsinvesti-gated.Followingourmajorﬁndings,pastcross-sectionalstudiesthatfol-lowedatheoreticalmodelpostulatingthatstressorsleadtostrainshouldbereassessed.Forexample,wesuggestthatjobsatisfactionshouldbeconsid-eredasaprecursorofstressors,ratherthanthereverse.

Wearguethatourﬁndingsmayhavesomepracticalimplications.Forexample,iftheobjectiveofaninterventionistoreducecurrentlevelsofemployeeburnout,thenacomprehensiveapproachwhichfocusesbothonreducingthestressorstowhichemployeesareexposedandonfactorsamelioratingburnout,suchasgoodhealthandsleeppractices,wouldprob-ablybemoreeffectivethan,say,onlystresscounseling.Moregenerally,strategiesforthepromotionofteachers’qualityofworkinglife,includingthoseaimedatreducingburnoutlevelsandsupportingpositivementalandphysicalhealth,areoftendiscussedbyteachers’organizations.Aprecondi-tionfortheeffectivenessofsuchpoliciesisasolidunderstandingofthecausalmaplinkingstressorsandstrains.Thisstudyisanattempttofurtherourunderstandingofthismap.

REFERENCES

Bacharach,S.B.,&Bamberger,P.(1992).Causalmodelsofrolestressorsantecedentsand

consequences:Theimportanceofoccupationaldifferences.JournalofVocationalBehav-ior,41,13–34.

Bateman,T.S.,&Strasser,S.(1983).Across-laggedregressiontestoftherelationships

330Shirom,Oliver,andStein

betweenjobtensionandemployeesatisfaction.JournalofAppliedPsychology,68,439–445.

Bliese,P.D.,&Halverson,R.R.(1996).Individualandnomotheticmodelsofjobstress:An

examinationofworkhours,cohesion,andwell-being.JournalofAppliedSocialPsychol-ogy,26,1171–11.

Borg,M.G.(1990).OccupationalstressinBritisheducationalsetting:Areview.Educational

Psychology,10,103–126.

Boyle,G.J.,Borg,M.G.,Falzon,J.M.,&Baglioni,A.J.(1995).Astructuralmodelofthe

dimensionsofteacherstress.BritishJournalofEducationalPsychology,65,49–67.Brenner,S.O.,So¨rbom,D.,&Wallius,E.(1985).Thestresschain:Alongitudinalconﬁrmatory

studyofteacherstress,coping,andsocialsupport.JournalofOccupationalPsychology,58,1–13.

Cohen,A.(1983).Comparingcoefﬁcientsacrosssubsamples.SociologicalMethodsand

Research,12,77–94.

Conley,J.J.(1984).Thehierarchyofconsistency:Areviewandmodeloflongitudinalﬁndings

onadultindividualdifferencesinintelligence,personalityandself-opinion.PersonalityandIndividualDifferences,5,11–25.

Cox,T.,Boot,N.,Cox,S.,&Harrison,S.(1988).Stressinschools:Anorganizational

perspective.WorkandStress,2,353–362.

Crampton,S.M.,&Wagner,J.S.(1994).Percept-perceptinﬂationinmicro-organizational

research:Aninvestigationofprevalenceandeffect.JournalofAppliedPsychology,79,67–76.

deJonge,J.,Dormann,C.,Janssen,P.P.M.,Dollard,M.F.,Landeweerd,J.A.,&Nijhuis,

F.J.N.(2001).Testingreciprocalrelationshipsbetweenjobcharacteristicsandpsycho-logicalwell-being:Across-laggedstructuralequationmodel.JournalofOccupationalandOrganizationalPsychology,74,29–46.

deLange,A.H.,Taris,T.,Kompier,M.,Houtman,I.L.D.,&Bongers,P.M.(2004).The

relationshipsbetweenworkcharacteristicsandmentalhealth:Examiningnormal,re-versed,andreciprocalrelationshipsina4-wavestudy.WorkandStress,18,149–166.deLange,A.H.,Taris,T.W.,Houtman,I.L.D.,&Bongers,P.M.(2003).“Theverybestof

themillennium”:Longitudinalresearchandthedemand-control-(support)model.JournalofOccupationalHealthPsychology,8,282–305.

Demerouti,E.,Bakker,A.B.,Nachreiner,F.,&Schaufeli,W.B.(2001).TheJobDemands–

ResourcesModelofburnout.JournalofAppliedPsychology,86,499–512.

Dormann,C.,&Zapf,D.(1999).Socialsupport,socialstressorsatwork,anddepressive

symptoms:Testingformainandmoderatingeffectswithstructuralequationsinathree-wavelongitudinalstudy.JournalofAppliedPsychology,84,874–884.

Doty,D.H.,&Glick,W.H.(1998).Commonmethodsbias:Doescommonmethodsvariance

biasresults?OrganizationalResearchMethods,1,374–406.

Edwards,J.R.,Cable,D.M.,Williamson,I.O.,Lambert,L.S.,&Shipp,A.J.(2006).The

phenomenologyofﬁt:Linkingthepersonandenvironmenttothesubjectiveexperienceofperson–environmentﬁt.JournalofAppliedPsychology,91,802–827.

Edwards,J.R.,Caplan,R.D.,&VanHarrison,R.(1998).Person-EnvironmentFitTheory:

Conceptualfoundations,empiricalevidence,anddirectionsforfutureresearch.InC.L.Cooper(Ed.),Theoriesoforganizationalstress(pp.28–68).NewYork:OxfordUniver-sityPress.

Faragher,B.,Cass,M.,&Cooper,C.L.(2005).Therelationshipbetweenjobsatisfactionand

health:Ameta-analysis.OccupationalandEnvironmentalMedicine,62,105–112.Finkel,S.E.(1995).Causalanalysiswithpaneldata.ThousandOaks,CA:Sage.

Fletcher,B.C.,&Jones,F.(1993).ArefutationofKarasek’sdemand-discretionmodelwitha

rangeofdependentmeasures.JournalofOrganizationalBehavior,14,219–330.

StressandStraininTeachers331

French,J.R.P.,Caplan,R.D.,&Harrison,R.V.(1982).Themechanismsofjobstressand

strain.Chichester,UK:Wiley.

Frese,M.(1985).Stressatworkandpsychosomaticcomplaints:Acausalinterpretation.

JournalofAppliedPsychology,70,314–328.

Ganster,D.C.,&Schaubroeck,J.(1991).Workstressandemployeehealth.Journalof

Management,17,235–271.

Gelsema,T.I.,VanDerDoef,M.,Maes,S.,Janssen,M.,Akerboom,S.,&Verhoeven,C.

(2006).Alongitudinalstudyofjobstressinthenursingprofession:Causesandconse-quences.JournalofNursingManagement,14,2–299.

Hackman,J.R.,&Oldham,G.R.(1980).Workredesign.Reading,MA:AddisonWesley.Halbesleben,J.R.B.,&Buckley,M.R.(2004).Burnoutinorganizationallife.Journalof

Management,30,859–879.

Hobfoll,S.E.(19).Conservationofresources:Anewattemptatconceptualizingstress.

AmericanPsychologist,44,513–524.

Hobfoll,S.E.,&Shirom,A.(2000).Conservationofresourcestheory:Applicationstostress

andmanagementintheworkplace.InR.T.Golembiewski(Ed.),Handbookoforgani-zationbehavior(2ndRevisedEd.,pp.57–81).NewYork:Dekker.

Houkes,I.,Janssen,P.P.M.,deJonge,J.,&Nijhuis,F.J.N.(2001).Workandindividual

determinantsofintrinsicworkmotivation,emotionalexhaustion,andturnoverintention:Amulti-sampleanalyses.InternationalJournalofStressManagement,8,257s–283s.Hulin,C.L.,&Judge,T.A.(2003).Jobattitudes.InW.C.Borman,D.R.Ilgen,&R.J.

Klimoski(Eds.),Handbookofpsychology(Vol.12,pp.255–277).Hoboken,NJ:Wiley.Jackson,S.E.,&Schuler,R.S.(1985).Ameta-analysisandconceptualcritiqueofresearchon

roleambiguityandroleconﬂictinworksettings.OrganizationalBehaviorandHumanDecisionProcesses,36,16–78.

James,L.R.,&Tetrick,L.E.(1986).Conﬁrmatoryanalytictestofthreecausalmodelsrelating

jobperceptionstojobsatisfaction.JournalofAppliedPsychology,71,77–82.

Jex,S.M.,Beehr,T.A.,&Roberts,C.K.(1992).Themeaningofoccupationalstressitemsto

surveyrespondents.JournalofAppliedPsychology,77,623–628.

Judge,T.A.,Bono,J.E.,Thoresen,C.J.,&Patton,G.K.(2001).Thejobsatisfaction-job

performancerelationship:Aqualitativeandquantitativereview.PsychologicalBulletin,127,376–407.

Judge,T.A.,&Larsen,R.J.(2001).Dispositionalaffectandjobsatisfaction:Areviewand

theoreticalextension.OrganizationalBehaviorandHumanDecisionProcesses,86,67–98.

Kahn,R.L.,&Byosiere,P.(1992).Stressinorganizations.InM.D.Dunnette&L.M.Hough

(Eds.),Handbookofindustrialandorganizationalpsychology(2Ed.,Vol.3,pp.571–651).PaloAlto,CA:ConsultingPsychologistsPress.

Karasek,R.(1979).Jobdemands,jobdecisionlatitude,andmentalstrain:Implicationsforjob

redesign.AdministrativeScienceQuarterly,24,285–308.

Karasek,R.,&Theorell,T.(1990).Healthywork:Stress,productivity,andthereconstruction

ofworkinglife.NewYork:BasicBooks.

Kasl,S.V.(1996).Theinﬂuenceoftheworkenvironmentoncardiovascularhealth:A

historical,conceptual,andmethodologicalperspective.JournalofOccupationalHealthPsychology,1,42–56.

Kyriacou,C.(2001).Teacherstress:Directionsforfutureresearch.EducationalReview,53,

27–35.

Lazarus,R.S.(1999).Stressandemotion.NewYork:Springer.

Lazarus,R.S.(2001).Relationalmeaninganddiscreteemotions.InK.R.Scherer,A.A.

Schorr,&T.Johnston(Eds.),Appraisalprocessesinemotions:Theory,research,methods(pp.37–67).NewYork:OxfordUniversityPress.

Lazarus,R.S.,&Folkman,S.(1984).Stress,appraisal,andcoping.NewYork:Springer.

332Shirom,Oliver,andStein

McCarthy,C.J.,Lambert,R.G.,O’Donnell,M.,&Melendres,L.T.(2009).Therelationof

elementaryteachers’experience,stress,andcopingresourcestoburnoutsymptoms.TheElementarySchoolJournal,109,282–300.

Mechanic,D.(1980).Theexperienceandreportingofcommonphysicalcomplaints.Journal

ofHealthandSocialBehavior,21,146–155.

Melamed,S.,Shirom,A.,Toker,S.,Berliner,S.,&Shapira,I.(2006).Burnoutandriskof

cardiovasculardisease:Evidence,possiblecausalpaths,andpromisingresearchdirec-tions.PsychologicalBulletin,132,327–353.

Miller,G.E.,Chen,E.,&Zhou,E.S.(2007).Ifitgoesup,Mustitcomedown?Chronicstress

andthehypothalamic-pituitary-adrenocorticalaxisinhumans.PsychologicalBulletin,133,25–45.

Mitchell,T.R.,&James,L.R.(2001).Buildingbettertheory:Timeandthespeciﬁcationof

whenthingshappen.AcademyofManagementReview,26,530–547.

Ortqvist,D.,&Wincent,J.(2006).Prominentconsequencesofrolestress:Ameta-analytic

review.InternationalJournalofStressManagement,13,399–422.

Rosenthal,R.,Rosnow,R.L.,&Rubin,D.B.(2000).Contrastsandeffectsizesinbehavioral

research.Cambridge,UK:CambridgeUniversityPress.

Semmer,N.,Zapf,D.,&Greif,S.(1996).‘Sharedjobstrain’:Anewapproachforassessing

thevalidityofjobstressmeasurements.JournalofOccupationalandOrganizationalPsychology,69,293–310.

Shirom,A.,&Mazeh,T.(1988).Periodicityinseniority–jobsatisfactionrelationship.Journal

ofVocationalBehavior,33,38–49.

Shirom,A.,&Melamed,S.(2006).Acomparisonoftheconstructvalidityoftwoburnout

measuresintwogroupsofprofessionals.InternationalJournalofStressManagement,13,176–200.

Siegrist,J.,&Rodel,A.(2006).Workstressandhealthriskbehavior.ScandinavianJournalof

WorkEnvironmentandHealth,32,473–481.

Sijtsma,K.(2009).Ontheuse,themisuse,andtheverylimitedusefulnessofCronbach’salpha.

Psychometrika,74,107–120.

Spector,P.E.(2006).Methodvarianceinorganizationalresearch:Truthorurbanlegend?

OrganizationalResearchMethods,9,221–232.

Staw,B.M.,&Cohen-Charash,Y.(2005).Thedispositionalapproachtojobsatisfaction:More

thanamirage,butnotyetanoasis.JournalofOrganizationalBehavior,26,59–78.Suls,J.,&Bunde,J.(2005).Anger,anxiety,anddepressionasriskfactorsforcardiovascular

disease:Theproblemsandimplicationsofoverlappingaffectivedispositions.Psycholog-icalBulletin,131,260–300.

Taris,T.W.(2006).Istherearelationshipbetweenburnoutandobjectiveperformance?A

criticalreviewof16studies.WorkandStress,20,316–334.

terDoest,L.,&deJonge,J.(2006).Testingcausalmodelsofjobcharacteristicsandemployee

well-being:Areplicationstudyusingcross-laggedstructuralequationmodeling.JournalofOccupationalandOrganizationalPsychology,79,499–507.

Twisk,J.W.R.(2003).Appliedlongitudinaldataanalysisofepidemiology.Cambridge,UK:

CambridgeUniversityPress.

Ursin,H.,&Eriksen,H.R.(2004).Thecognitiveactivationtheoryofstress.Psychoneuroen-docrinology,29,567–592.

Watson,D.,&Pennebaker,J.W.(19).Healthcomplaints,stress,anddistress:Exploringthe

centralroleofnegativeaffectivity.PsychologyReview,96,234–254.

Zapf,D.,Dormann,C.,&Frese,M.(1996).Longitudinalstudiesinorganizationalstress

research:Areviewoftheliteraturewithreferencetomethodologicalissues.JournalofOccupationalHealthPsychology,1,145–169.

CorrectiontoShirom,Oliver,&Stein(2009)

Inthearticle,“Teachers’StressorsandStrains:ALongitudinalStudyofTheirRelationships”byArieShirom,AmalyaOliver,andEstherStein(InternationalJournalofStressManagement,2009,Vol.16,No.4,pp.312-332),thetwoco-authors’afﬁliationswereincorrectlylisted.Theco-authors’correctafﬁlia-tionsareasfollows:AmalyaOliver,HebrewUniversityofJerusalemandEstherStein,BeitBerlEducationalCollege.

因篇幅问题不能全部显示，请点此查看更多更全内容

查看全文