罗兰·米勒对亲密关系的综述研究，本文搬运书籍片段；亲密关系除开狭义的男女朋友关系，这里还指人与人的各种链接；无论你是搞学术、做心理学、英语爱好者，或者对关系有迷茫的小伙伴，我都非常推荐读这本书
RowlandS.Miller
习读专用，不作商用，如侵联删。
CHAPTER2
TheShortHistoryofRelationshipScience◆DevelopingaQuestion
◆ObtainingParticipants◆ChoosingaDesign◆SelectingaSetting◆TheNatureofOurData◆TheEthicsofSuch
Endeavors◆InterpretingandIntegratingResults◆AFinalNote
◆ForYourConsideration◆ChapterSummary
1
Ibetyoudreadachapteronresearchmethods.Youprobablyregarditasadistractiontobeenduredbeforegettingto“thegoodstuff.”Love,sex,andjeal-ousyprobablyappealtoyou,forinstance,butresearchdesignsandproceduresarenotatthetopofyourlist.
2
3
Thisisn’talwayseasy.Aswe’llseeinthischapter,theremaybevariouswaystoaddressaspecificresearchquestion,andeachmayhaveitsownpar-ticularadvantagesanddisadvantages.Reputablescientistsgatherandevaluateinformationsystematicallyandcarefully,butnosingletechniquemayprovidetheindisputableanswerstheyseek.Athoughtfulunderstandingofrelation-shipsoftenrequiresustocombineinformationfrommanystudies,evaluatingdiversefactswithjudiciousdiscernment.Thischapterprovidestheoverviewofthetechniquesofrelationshipsciencethatyouneedtomakesuchjudgments.
4
Onlybasicprinciplesaredescribedhere—thisisoneoftheshortestchap-tersinthebook—buttheyshouldhelpyoudecidewhatevidencetoacceptandwhattoquestion.Andtrustme.There’salotherethat’sworththinkingaboutevenifyou’vereadaMethodschapterbefore.Hopefully,whenwe’refinishedyou’llbebetterequippedtodistinguishusefulresearchevidencefromuselessanecdotesormerespeculation.Forevenmoreinformation,don’thesitatetoconsultothersourcessuchasMehlandConner(2012)andLeary(2012).
5
IsaacNewtonidentifiedsomeofthebasiclawsofphysicsmorethan400yearsago(backin1687).Biologyandchemistryhavebeenaroundforjustaslong.Thesystematicstudyofhumanrelationships,ontheotherhand,isarecentinventionthatissonewandsorecentthatyoucanactuallytalk,ifyouwant,withmostofthescientistswhohaveeverstudiedhumanintimacy!Thisisnosmallmatter.Becauserelationshipsciencehasashorthistory,itislesswellknownthanmostothersciences,andforthatreason,itislesswellunderstood.Veryfewpeopleoutsideofcollegesanduniversitiesappreciatetheextraordi-narystridesthisnewdisciplinehasmadeinthelast50years.
6
7
Thatbegantochange,thankgoodness,whenanexplosionofstudiesputthefieldonthescientificmapinthe1960sand1970s.PioneeringscientistsEllenBerscheidandElaineHatfieldbegansystematicstudiesofattractionandlovethatwerefueledbyanewemphasisonlaboratoryexperimentsinsocialpsychology(Reisetal.,2013).Inaquestforprecisionthatyieldedunambigu-ousresults,researchersbeganstudyingspecificinfluencesonrelationshipsthattheywereabletocontrolandmanipulate.Forinstance,inaprominentlineofresearchontheroleofattitudesimilarityinliking,DonnByrneandhiscolleagues(e.g.,Byrne&Nelson,1965)askedpeopletoinspectanatti-tudesurveythathadsupposedlybeencompletedbyastrangerinanotherroom.Then,theyaskedtheparticipantshowmuchtheylikedthestranger.Whattheparticipantsdidn’tknowwasthattheresearchershadpreparedthesurveyeithertoagreeordisagreewiththeparticipants’ownattitudes(whichhadbeenassessedearlier).Thismanipulationofattitudesimilarityhadcleareffects:Apparentagreementcausedpeopletolikethestrangermorethandisagreementdid.
8
Themethodologicalrigorofprocedureslikethesesatisfiedresearchers’desiresforclarityandconcision.Theylegitimizedandpopularizedthestudyofinterpersonalattraction,makingitanindispensablepartofpsychologytext-booksforthefirsttime.Inretrospect,however,theseinvestigationsoftendidapoorjobofrepresentingthenaturalcomplexityofrealrelationships.Thepar-ticipantsinmanyofByrne’sexperimentsneveractuallymetthatotherpersonorinteractedwithhimorherinanyway.Indeed,intheprocedureI’vebeendescribing,ameetingcouldn’toccurbecausethestrangerdidn’tactuallyexist!Inthis“phantomstranger”technique,peopleweremerelyreactingtocheckmarksonapieceofpaperandweretheonlyrealparticipantsinthestudy.Theresearchersweremeasuringattractiontosomeonewhowasn’teventhere.Byrneandhiscolleagueschosethismethod,limitingtheirinvestigationtoonecarefullycontrolledaspectofrelationshipdevelopment,tostudyitconclusively.However,theyalsocreatedarathersterilesituationthatlackedtheimmediacyanddramaofchattingwithsomeoneface-to-faceonafirstdate.
9
Butdon’tunderestimatetheimportanceofstudieslikethese.Theydem-onstratedthatrelationshipscouldbestudiedscientificallyandthatsuchinves-tigationshadenormouspromise,andtheybroughtrelationshipsciencetotheattentionoffellowscholarsforthefirsttime(Reis,2012).Andinthedecadessince,throughthecombinedeffortsoffamilyscholars,psychologists,sociolo-gists,communicationresearchers,andneuroscientists,relationshipsciencehasgrownandevolvedtoencompassnewmethodsofconsiderablecomplexityandsophistication.Today,relationshipscience:
oftenusesdiversesamplesofpeopledrawnfromallwalksoflifeandfromaroundtheworld,
examinesvariedtypesoffamily,friendship,andromanticrelationships,
frequentlystudiesthoserelationshipsoverlongperiodsoftime,
studiesboththepleasantandunpleasantaspectsofrelationships,
oftenfollowsrelationshipsintheirnaturalsettings,and
usessophisticatedtechnology.
10
Herearesomeexamplesofhowthefieldcurrentlyoperates:
AttheUniversityofTexasatArlington,WilliamIckesandhiscolleaguesstudyspontaneous,unscriptedinteractionsbetweenpeople(whohavesometimesjustmet)byleavingthemaloneonacomfortablecouchforafewminuteswhiletheirconversationiscovertlyvideotaped(Ickes&Hodges,2013).Thecameraisactuallyhiddeninanotherroomacrossthehallandcan’tbeseenevenifyou’relookingdirectlyatit,sothere’snocluethatanyoneiswatching(seeFigure2.1).Afterward,iftheparticipantsgivetheirpermissionfortheirrecordingstobeused,theycanreviewthetapesoftheirinteractioninprivatecubicleswheretheyareinvitedtoreportwhattheywerethinking—andwhattheythoughttheirpartnerswerethinking—ateachpointintheinteraction.Themethodthusprovidesanobjectivevideotapedrecordoftheinteraction,andparticipants’thoughtsandfeelingsandperceptionsofoneanothercanbeobtained,too.
11
12
13
14
15
16
17
HowdothesescholarsstudyrelationshipsThefirststepinanyscientificendeavoristoaskaquestion,andinafieldlikethisone,somequestionsemergefrompersonalexperience.Relationshipresearchershaveanadvantageovermanyotherscientistsbecausetheirownexperiencesincloserelationshipscanalertthemtoimportantprocesses.Indeed,theymaybehipdeepintheveryswampstheyaretryingtodrain(Miller,2008)!Broadersocialproblemsalsosuggestques-tionsforcarefulstudy.Forinstance,thehugeincreaseintheU.S.divorceratefrom1960to1980resultedinaconsiderableamountofresearchondivorceassocialscientiststooknoteoftheculture’schanges.
18
Questionsalsocomefrompreviousresearch:Studiesthatansweroneques-tionmayraisenewones.Andstillotherquestionsaresuggestedbytheoriesthatstrivetoofferexplanationsforrelationalevents.Usefultheoriesbothaccountforexistingfactsandmakenewpredictions,andstudiesoftenseektotestthosehypotheses.Relationshipscienceinvolvesquestionsthatspringfromallofthesesources;scientistswillputtogethertheirpersonalobservations,theirrecognitionofsocialproblems,theirknowledgeofpreviousresearch,andtheirtheoreticalperspectivestocreatethequestionstheyask(Fiske,2004).
19
20
So,whoserelationshipsarestudiedRelationshipresearchersusuallyrecruitparticipantsinoneoftwoways.Thefirstapproachistouseanyonewhoisreadilyavailableandwhoconsentstoparticipate;thisisaconveniencesamplebecauseitis(comparatively)convenientfortheresearchertoobtain.Universityprofessorsoftenworkwithcollegestudentswhoarerequiredtoberesearchparticipantsaspartoftheircoursework.Althoughsomespecificcharacteris-ticsmustsometimesbemet(sothatastudymayfocus,forinstance,onlyondatingpartnerswhohaveknowneachotherforlessthan2months),research-erswhouseconveniencesamplesareusuallygladtogetthehelpofeveryonetheycan.
21.
Incontrast,projectsthatusearepresentativesamplestrivetoensurethat,collectively,theirparticipantsresembletheentirepopulationofpeoplewhoareofinterest.Atrulyrepresentativestudyofmarriage,forexample,wouldneedtoincludemarriedpeopleofallsorts—allages,allnationalities,andallsocio-economiclevels.That’satallorderbecause,ifnothingelse,thepeoplewhovol-untarilyconsenttoparticipateinaresearchstudymaybesomewhatdifferentfromthosewhorefusetoparticipate(seetheboxonpage50).Still,somestud-ieshaveobtainedsamplesthatarerepresentativeof(volunteersin)theadultpopulationofindividualcountriesorotherdelimitedgroups.AndstudiesthatarestraightforwardenoughtobeconductedovertheInternetcanattractverylargesamplesthataremuchmorediversethanthosefoundonanyonecampusoreveninanyonecountry(Goslingetal.,2010).
22
23
Let’sconsideraspecificexample.Backin1978,RussellClarksentmenandwomenoutacrossthecampusofFloridaStateUniversitytopropositionmembersoftheothersex.Individually,theyapproachedunsuspectingpeopleandrandomlyassignedthemtooneofthreeinvitations(seeTable2.1);somepeopleweresimplyaskedoutonadate,whereasotherswereaskedtohavesex!Thenotableresultswerethatnowomanacceptedtheofferofsexfromastranger,but75percentofthemendid—andthatwasmorementhanacceptedthedate!
24
Thiswasastrikingresult,butsowhatThestudyinvolvedasmallcon-veniencesampleonjustonecampus.PerhapstheresultstoldusmoreaboutthemenatFSUthantheydidaboutmenandwomeningeneral.Infact,Clarkhadtroublegettingthestudypublishedbecauseofreviewers’con-cernsaboutthegeneralityoftheresults.So,in1982,heandElaineHatfieldtriedagain;theyrepeatedthestudyatFSUandgotthesameresults(Clark&Hatfield,1989).
25
Well,stillsowhatItwas4yearslater,buttheprocedurehadstillbeentriedonlyinTallahassee.Ifyougivethisexamplesomethought,you’llbeabletogenerateseveralreasonswhytheresultsmightapplyonlytooneparticulartimeandoneparticularplace.
26
I’dliketosuggestadifferentperspective.Let’snotfusstoomuchabouttheexactpercentageofcollegemeninFloridaorelsewherewhowouldcon-senttosexwithastranger.That’sthekindofspecificattitudethatyou’dexpecttovarysomefromonedemographicgrouptoanother.Insteadofendlesslycriticizing—or,evenworse,dismissing—theresultsoftheClarkandHatfield(1989)studies,let’srecognizetheirlimitationsbutnotmisstheirpoint:Menweregenerallymoreacceptingofcasualsexthanwomenwere.Whensome-bodyactuallyasked,menweremuchmorelikelytoacceptasexualinvitationfromastrangerthanwomenwere.Statedgenerally,that’sexactlytheconclu-sionthathasnowbeendrawnfromsubsequentinvestigationsinvolvingmorethan20,000participantsfromeverymajorregionoftheworld(Schmitt&theInternationalSexualityDescriptionProject,2003),andClarkandHatfieldwereamongtheveryfirsttodocumentthissexdifference.Theirmethodwassimple,andtheirsamplewaslimited,buttheywereontosomething,andtheirproce-duredetectedabasicpatternthatreallydoesseemtoexist.1
1Forinstance,inastudyinMay2006alongthewestcoastofFrance,57percentofthemenbutonly3percentofthewomenacceptedinvitationstohavesexwithanattractivestranger(Guéguen,2011).InJune2009,38percentofthemenbutonly2percentofthewomeninurbanareasofDenmarkdidso(Hald&Hgh-Olesen,2010).(Theseglaringdifferencesaresmaller,however,whenmenandwomenareaskedtoimagineoffersforsexfromcelebritiessuchasJenniferLopezandBradPitt[Conley,2011]!).
27
So,it’sabsolutelytruethattheClarkandHatfield(1989)studieswerenotperfect.That’sajudgmentwithwhichClarkandHatfield(2003)themselvesagree!Butaslongastheirresultsareconsideredthoughtfullyandjudiciously,evensmallstudiesusingconveniencesampleslikethesecanmakeimportantcontributionstorelationshipscience.Ourconfidenceinourcollectiveunder-standingofrelationshipsreliesonknowledgeobtainedwithdiversemeth-ods(Reis,2002).Anysinglestudymayhavesomeimperfections,butthoseweaknessesmaybeansweredbyanotherstudy’sstrengths.Withaseriesofinvestigations,eachapproachingaproblemfromadifferentangle,wegraduallydelineatethetruth.Tobeathoughtfulconsumerofrelationshipscience,youshouldthinkthewaythescientistsdo:Noonestudyisperfect.Becautious.Variousmethodsarevaluable.Wisdomtakestime.Butthetruthisoutthere,andwe’regettingcloserallthetime.
28
Okay,we’veformulatedaresearchquestionandobtainedsomeparticipants.Now,weneedtoarrangeourobservationsinawaythatwillanswerourques-tion.Howdowedothat
29
Correlationsdescribepatternsinwhichchangeinoneeventisaccompaniedtosomedegreebychangeinanother.Thepatternscanbeoftwotypes.Ifthetwoeventsarepositivelycorrelated,theygoupanddowntogether—thatis,asonegoesup,sodoestheother,andastheothergoesdown,sodoesthefirstone.Inspeed-datingstudies,forinstance,themoretwostrangersthinktheyhaveincommonafterabriefinteraction,themoretheytendtolikeeachother(Tidwelletal.,2013).Higherlevelsofperceivedsimilarityareassociatedwithgreaterliking.
30
Incontrast,iftwoeventsarenegativelycorrelated,theychangeinoppo-sitedirections:asonegoesup,theothergoesdown,andasonegoesdown,theothergoesup.Forexample,peoplewhohavehighneuroticism2tendtobelesssatisfiedwiththeirmarriagesthanothersare;higherneuroticismisassociatedwithlowermaritalsatisfaction(Malouffetal.,2010).PositiveandnegativecorrelationsareportrayedinFigure2.2,whichalsoincludesanexampleofwhatweseewhentwoeventsareuncorrelated:Ifeventsareunre-lated,oneofthemdoesn’tchangeinanypredictablewaywhentheothergoesupordown.
31
Patternsliketheseareoftenintriguing,andtheycanbeveryimportant,buttheyareroutinelymisunderstoodbyunsophisticatedconsumers.Please,alwaysrememberthatcorrelationstellusthattwoeventschangetogetherinsomerecognizableway,but,allbythemselves,theydonottelluswhythatoccurs.Correlationaldesignstypicallystudynaturallyoccurringbehaviorwithouttryingtoinfluenceorcontrolthesituationsinwhichitunfolds—andthecorrelationsthatareobserveddonottellusaboutthecausalconnectionsbetweenevents.Becarefulnottoassumetoomuchwhenyouencounteracor-relation;manydifferentplausiblecausalconnectionsmayallbepossiblewhenacorrelationexists.Considerthefactthatperceivedsimilarityispositivelyrelatedtoliking;herearethreestraightforwardpossibilities:
2Takealookbackatpage27ifyou’dliketorefreshyourmemoryofwhatneuroticismis.
32
oneofthesetwomaycausetheother—perceivedsimilaritymightleadtogreaterliking.Or,
theotherofthesetwocouldcausetheone—sothatlikingothersleadsustoassumethatwehavealotincommonwiththem.Or,
somethingelse,athirdvariable,mayexplainwhysimilarityandlikingarerelated.Similaritymaynotleadtoliking,andlikingmaynotleadtoperceivedsimilarity;instead,somethingelse,likereallygoodlooks,maycauseustolikeothersandtoassume(orhope)thatwe’recompatiblewiththem.
33
Anyofthesethree,alongwithmanyothermorecomplexchainsofevents,maybepossiblewhentwoeventsarecorrelated.Ifallwehaveisacorrelation,allweknowisthatapredictablepatternexists.Wedon’tknowwhatcausalconnec-tionsareinvolved.3
3Ishouldnote,however,thatifwehavelotsofcorrelationsinvolvinganumberofvariables,orifwehavetakenourmeasurementsonseveraloccasionsoveraspanoftime,sophisticatedstatisti-calanalysescanusuallyruleoutsomeofthepossiblecausalconnectionsthatmakecorrelationalfindingsambiguous.Weshouldbecarefulnottoassumethatsimplecorrelationsinvolvecausalconnections,butadvancedstatisticaltechniquesoftenmakeitpossibletodrawsomedefensibleconclusionsaboutcauseandeffectwithincorrelationaldesigns.
34
Whenit’spossible,thewaytoinvestigatecausalconnectionsistouseanexperimentaldesign.Experimentsprovidestraightforwardinformationaboutcausesandtheireffectsbecauseexperimenterscreateandcontrolthecondi-tionstheystudy.Inatrueexperiment,researchersintentionallymanipulateoneormorevariablesandrandomlyassignparticipantstothedifferentconditionstheyhavecreatedtoseehowthosechangesaffectpeople.Thus,insteadofjustasking“Dotwothingschangetogether”experimentersask“Ifwechangeone,whathappenstotheother”
35
Let’sillustratethedifferencebetweenanexperimentandacorrelationalstudybyreconsideringDonnByrne’sclassicworkonattitudesimilarityandattraction(e.g.,Byrne&Nelson,1965).HadByrnesimplymeasuredpartners’perceptionsofeachother’sattitudesandtheirlikingforeachother,hewouldhaveobtainedapositivecorrelationbetweenperceivedsimilarityandliking,buthewouldnothavebeensurewhytheywererelated.
36
WhatByrnedidinsteadwasanexperiment.Oncehisparticipantsarrivedathislab,heflippedacointodeterminerandomlywhowouldencounterasimilarstrangerandwhowouldencounteronewhodidn’tagreewiththematall.Hecontrolledthatapparentagreementordisagreement,anditwastheonlydifferencebetweenthetwosituationsinwhichparticipantsfoundthem-selves.Withthisprocedure,whenByrneobservedhigherlikingforthesimilarstranger,hecouldreasonablyconcludethatthegreateragreementhadcausedthehigherliking.HowBecausetheparticipantswererandomlyassignedtothetwosituations,thedifferentdegreesoflikingcouldnotbeduetodifferencesinthepeoplewhoencounteredeachsituation;onaverage,thetwogroupsofparticipantswereidentical.Moreover,theyallhadidenticalexperiencesintheexperimentexceptfortheapparentsimilarityofthestranger.Theonlyreason-ableexplanationforthedifferentbehaviorByrneobservedwasthatsimilarityleadstoliking.Hisexperimentclearlyshowedthatthemanipulatedcause,atti-tudesimilarity,hadanoticeableeffect,higherliking.
37
Experimentsprovideclearer,moredefinitivetestsofcausalconnectionsthanotherdesignsdo.Donewell,theyclearlydelineatecauseandeffect.Why,then,doresearcherseverdoanythingelseThekeyisthatexperimentershavetobeabletocontrolandmanipulatetheeventstheywishtostudy.Byrnecouldcontroltheinformationthathisparticipantsreceivedaboutsomeonetheyhadnevermet,buthecouldn’tmanipulateotherimportantinfluencesonintimaterelationships.Westillcan’t.(Howdoyoucreatefull-fledgedexperiencesofromanticloveinalaboratory)Youcan’tdoexperimentsoneventsyoucannotcontrol.
38
So,correlationalandexperimentaldesignseachhavetheirownadvan-tages.Withcorrelationaldesigns,wecanstudycompellingeventsintherealworld—commitmenttoarelationship,passionatelove,unsafesex—andexam-inethelinksamongthem.Butcorrelationaldesignsarelimitedinwhattheycantellusaboutthecausalrelationshipsamongevents.Withexperimentaldesigns,wecanexaminecausalconnections,butwearelimitedinwhatwecanstudy.Hopefully,youcanseewhydifferentresearchersmaystudythesametopicindifferentways,withdifferentresearchdesigns—andwhythat’sagoodthing.
39
We’remakingprogress.We’vedevelopedourresearchquestion,recruitedourparticipants,andchosenourdesign.Now,wehavetoselectasettinginwhichtoconductourinvestigation.Theusualchoicesinclude(a)alaboratoryor(b)anatural,everydayenvironment,suchasacouple’shome.Eitherchoicehasadvantagesanddisadvantages.(You’regettingusedtothatnow,aren’tyou)Thelabofferstheadvantageofgreatercontroloverextraneous,unwantedinflu-ences,butitmayelicitartificialbehaviorthatdiffersfromwhatpeopleusuallydo(Callaghanetal.,2013).Naturalsettingsoffertheadvantageofobtainingmoretypicalbehavior,buttheycanbefullofdistractionsandhardtomanage.Whereverastudytakesplace,somebehaviorsaredifficulttostudybecausetheyarerare,orunpleasant,orveryintimate(orallthree).
40
Onewaytoover-comethesedifficultiesistohavesubjectsrole-playthebehaviorwe’retryingtounderstand—toact“asif”theywerejealous,forinstance,orwerehavinganargument,orweretryingtoenticesomeoneintobed.Role-playstudiesvaryinhowrealistictheyare.Atoneextreme,participantsmaybeaskedtoreadastoryinvolvingtherelevantbehaviorandtoimaginethoseeventshappeningtothem.Suchscenariosarealwayslessvividthantherealeventswouldbe,andtheyallowpeopletorespondinacool,collectedfashionthatmaybequitedifferentfromtheimpulsiveandemotionalreactionstheydisplaywhensucheventsreallytakeplace.
41
Attheotherextreme,studiesknownassimulationsaskpeopletoactoutaparticularroleinahypotheticalsituation.Forexample,aninvestigatormightaskacoupletopretendthattheyareangrywitheachotherandthenobservehowtheybehave.Thisstrategyismoreengrossing,butparticipantsstillknowthattheyareonlypretending.Role-playstudiesareanethicallydefensiblewaytostudyemotionallychargedtopics,butpeoplemaydowhattheythinktheyshoulddointhesesituationsratherthanwhattheyreallywoulddoiftheeventsactuallyoccurred.Onceagain,therearebothadvantagesanddisadvantagestoconsider.
42
Now,justwhattypeofinformationwillwebeactuallycollectingArewerecordingothers’judgmentsandperceptionsofarelationship,orareweinspectingspecificinteractionsourselvesTwomajortypesofresearchmea-suresaredescribedhere:(a)people’sownreportsabouttheirthoughts,feel-ings,andbehaviorsand(b)carefulobservationsofothers’behavior.We’llalsoexaminesomevariationsonthesethemes.Nomatterwhatdataweuse,ourmeasuresshouldhavepsychometricvalidityandreliability.Thatis,weshouldbeaccuratelymeasuringtheeventswe’retryingtomeasure(that’svalidity),and,ifthoseeventsaren’tchanging,weshouldgetthesamescorestimeaftertime(that’sreliability).
43
Themostcommonmeansofstudyingintimaterelationshipsistoaskpeo-pleabouttheirexperiences.Suchresponsesareself-reports,andtheycanbeobtainedinavarietyofformats:throughwrittenquestionnaires,verbalinter-views,orevenunstructureddiariesinwhichparticipantswriteaboutwhatevercomestomind.Thecommonthemelinkingsuchtechniquesisthatpeoplearetellingusabouttheirexperiences—we’renotwatchingthemourselves.
44
Self-reportdatahaveimportantbenefits.Foronething,theyallowusto“getinsidepeople’sheads”andunderstandpersonalpointsofviewthatmaynotbeapparenttooutsideobservers.Self-reportdataarealsoinexpensiveandeasytoobtain.Consider,forinstance,theshortself-reportmeasureprovidedinTable2.2:Thosesevenquestionsdoaremarkablygoodjobofassessingpeople’ssatisfactionwiththeircloserelationships.Formostpurposes,there’snoreasontoaskmoreelaboratequestionsoruseothermeanstodistinguishsatisfiedlov-ersfromthosewhoarelesscontentbecausethishandfulofstraightforwardquestionsworksjustfine(Renshawetal.,2011).Self-reportmeasurescanbebothveryefficientandveryinformative.Still(andbynow,thisprobablyisn’tasurprise!),self-reportsmayalsopresentpotentialproblems.Herearethreethingstoworryabout.
45
Self-reportsalwaysoccurinresponsetoaresearcher’sinstructionsorques-tions.Iftheparticipantsmisinterpretwhattheresearchermeansorintends,theirsubsequentself-reportscanbemisleading.Forinstance:“Withhowmanypeoplehaveyouhadsex”Whenmenanswerthatquestion,theytendtoincludepartnerswithwhomtheyhavehadoralsexbutnointercourse,whereaswomentendtocountonlythosepartnerswithwhomtheyhavehadintercourse(Guteetal.,2008).Thisisonereasonwhymenroutinelyreportthattheyhavehadsexwithmoremembersoftheothersexthanwomendo.4Infact,undetectedproblemswithpeople’scomprehensionoftermsdescribingsexualbehavior—including“orgasm”(Chabotetal.,2013)—maybeamajorprobleminsexualityresearch(Wiederman,2004).
46
Evenwhenpeopleunderstandourquestions,theymaynotbeabletoanswerthemcorrectly.Foronething,theymaylackinsightintotheiractions,sothatwhattheythinkisgoingonisn’tentirelyaccurate.Forinstance,womensaythephysicalattractivenessofamateislessimportanttothemthanmendo.However,whentheyencounterandevaluateseveralpotentialpartnersatonceinspeed-datingstudies,looksdomatterjustasmuchtowomenastheydotomen(Eastwick&Finkel,2008),andlooksarethemostimportantinfluenceonwholikeswhomforbothsexes(Luo&Zhang,2009).Onoccasion,whatpeoplecantellusabouttheirpreferencesandbehaviordoesn’taccuratelyreflectwhattheyactuallysayanddo.
47
Faultymemoriescanalsobeaproblem.Self-reportsaremostaccuratewhenpeopledescribespecific,objectiveeventsthathaveoccurredrecently.Theyaremorelikelytobeinaccuratewhenweaskthemaboutthingsthathappenedlongago(Aickenetal.,2013).Specificdetailsmaybeforgotten—inonestudy(Mitchell,2010),50percentofalargesampleofdivorcedpeopledidnotcorrectlyreportinwhichmonththeyweredivorced—andpastfeel-ingsareespeciallylikelytobemisremembered.Inparticular,ifapassionateromanceendsinpainanddiscontent,thedisappointedloversarelikelytohaveaveryhardtimerememberinghowhappyandenthusiastictheyfeltmonthsearlierwhentheyhadjustfalleninlove(Grote&Frieze,1998).
48
49
Anotherwaytocollectinformationaboutrelationshipsistoobservebehaviordirectly.Scientificobservationsarerarelycasualundertakings.Researcherseithermeasurebehaviorwithsophisticatedtoolsorcarefullytraintheircolleaguestomakeobservationsthatareaccurate,reliable,andoftenquitedetailed.
50
Somestudiesinvolvedirectobservationsofongoingbehaviorwhereasothersuserecordingsfromwhichobservationsaremadeatalatertime.Event-samplingisamethodthatusesintermittent,shortperiodsofobservationtocapturesamplesofbehaviorthatactuallyoccuroverlongerperiodsoftime;investigatorsmayrandomlysampleshortspansoftimewhenatargetbehaviorislikelytooccur,scatteringperiodsofobservationthroughdifferenttimesondifferentdays.TheworkbeingdonebyMatthiasMehl(Mehl&Robbins,2011)withsmallrecordersthatfitinapocketisafineexampleofthistechnique.Thedevicesarecalledelectronicallyactivatedrecorders,orEARs.(Getit)Theyswitchonforbriefperiodsatregularintervalsduringthedayandcapturethesoundsofwhateverinteractionsparticipantsarehavingatthetime.Youwon’tbesurprisedtolearnthatwhenEARscapturealotofsurlyemotionssuchasangerandcontemptincouples’conversations,thepartnersarelesscontentandtheirrelationshipismorefragile6monthslater(Slatcher&Ranson,2011).
51
Theobservationsthatresultfromaproceduresuchasthiscantakeseveralforms.Researcherssometimesmakeratingsthatcharacterizetheeventstheywitnessinrelativelyglobalterms.Forexample,anargumentmightberatedwithregardtotheextenttowhichitis”constructiveandproblemsolving”or”argumentativeandhostile.”Alternatively,observersmayemploycodingpro-ceduresthatfocusonveryspecificbehaviorssuchastheamountoftimepeoplespeakduringaninteraction,thenumberofsmilestheydisplay,orthenumberoftimestheytoucheachother(Humbadetal.,2011).Theseperceptionsaretyp-icallymoreobjectivethanratingsare,andtheycansometimesbemechanizedtobeevenmoreimpartial.Forinstance,JamesPennebakerhasdevelopedsoft-warethatcodesthewordspeopleuse,anditallowsanautomaticanalysisofthecontentofpeople’sconversations.(Andit’sbadnewswhenpartnersusetheword“you”toofrequently;suchpeopletendtobelesssatisfiedwiththeirrela-tionshipsthanthosewhouse“you”lessoften[Tausczik&Pennebaker,2010].)
52
Othertechnologiesprovideadditionalmeasuresofbehavior.Inaneye-trackingstudy,forinstance,participantsdonheadgearthatfocusestinyvideocamerasontheireyes.Then,whentheyinspectvariousimages,theireyemovementsindicatewhatthey’relookingat,andforhowlong(Gervaisetal.,2013).We’dbeabletotell,forinstance,whetheryoupreferblondesorbrunettesbypresentingtwoimagesdifferingonlyinhaircolorside-by-side:You’dspendmoretimescrutinizingtheimageyoufindmorealluring.
53
Observationssuchasthesegenerallyavoidthedisadvantagesofself-reports.Ontheotherhand,weneedself-reportsifwe’retounderstandpeople’spersonalperceptionsoftheirexperiences.Observationalstudiescanalsobeexpensive,sometimesrequiringcostlyequipmentandconsuminghoursandhoursofobservers’time.Oneremarkablestudyfilmedeverywakingmomentoftheinteractionsof32differentfamiliesoverthecourseof1week,andthe
1,540hoursofresultingvideorequiredthousandsofhoursofcarefulinspectiontocodeandcategorize(Carey,2010).
54
Observationalresearchcanalsosufferfromtheproblemofreactivity:Peoplemaychangetheirbehaviorwhentheyknowtheyarebeingobserved.(Acamerainyourlivingroomwouldprobablychangesomeofyourbehavior—atleastuntilyougotusedtoit.)Forthatreason,researchersarealwaysgladtoconductobservationsthatcannotpossiblyalterthebehaviorsthey’restudying—andinonesuchinvestigation,relationshipscientistsmoni-toredtheFacebookprofilesof1,640people—almosttheentirefreshmanclassataparticularuniversity—astheircollegeyearswentby(Wimmer&Lewis,2010).Theytrackedthepublicinformationintheprofilestodeterminehowtheusers’tastesandvaluesinfluencedthefriendshipstheyformed.Theresearch-ershadspecific,seriousaims—thiswasnotinformalbrowsing—andtheycouldn’thaveunwantedinfluenceonthebehaviortheywerestudyingbecausetheparticipantsdidnotknowthattheywerebeingwatched!Wewillundoubt-edlybeseeingmanymorestudiesofsocialnetworkingontheWebintheyearstocome.(DoyoufindthistroublingWhy)
55
Wecanalsoavoidanyproblemswithreactivityifweobservebehaviorthatpeoplecannotconsciouslycontrol,andphysiologicalmeasuresofpeople’sautonomicandbiochemicalreactionsoftendojustthat.Physiologicalmeasuresassesssuchresponsesasheartrate,muscletension,genitalarousal,brainactiv-ity,andhormonelevelstodeterminehowourphysicalstatesareassociatedwithoursocialbehavior.
56
Someinvestigationsexaminethemannerinwhichphysiologyshapesourinteractionswithothers.Forinstance,theleveloftheneuropeptideoxytocininyourbloodhelpstodeterminehowempathicandtrustingyouare(Campbell,2010);remarkably,ifyou’regivenadoseofoxytocin,you’llbetemporarilylesssuspiciousofothers(MacDonald&MacDonald,2010).Therearealsophysi-ologicalfoundationstoourattachmentstyles.Strongerautonomicreactionstosocialstressors,includingthereleaseofhormonessuchasadrenalin,arefoundininsecurepeoplethaninsecureones;socialthreatsthatcreateuneasyarousalinanxiousandavoidantpeopleoftenleavesecurepeoplecoolandcalm(Diamond&Fagundes,2010).
57
Otherstudiesseektomapthephysiologicalfoundationsofsocialbehavior.Forexample,fMRIhasidentifiedthestructuresinourbrainsthatseemtoregu-lateloveandlust(Tomlinson&Aron,2012).fMRIimagesshowwhichpartsofthebrainareconsumingmoreoxygenandarethereforemoreactivethanotherswhencertainstatesoccur—andasitturnsout,warmromanticaffectionandyearningsexualdesireappeartobecontrolledbydifferentpartsofourbrains.(Areyousurprised)
58
Physiologicalmeasuresareoftenexpensive,buttheiruseisincreasingbecausetheyallowresearcherstoexplorethephysicalfoundationsofourrela-tionships.Theyareagoodexampleofthemannerinwhichrelationshipscienceisbecomingmorecomplexandsophisticatedallthetime.
59
Historicalarchivesalsoavoidtheproblemofreactivity.Personaldocumentssuchasphotographsanddiaries,publicmediasuchasnewspapersandmaga-zines,andgovernmentalrecordssuchasmarriagelicensesandbirthrecordscanallbevaluablesourcesofdataaboutrelationships,andwhenthesearedated,theybecome“archival”information.Inonestudy,researchersexam-inedolduniversityyearbookphotostodetermineifpeople’sexpressionsasyoungadultscouldpredicttheirchancesofafuturedivorce(Hertensteinetal.,2009).(WhatdidtheyfindSeechapter5!)Archivalmaterialsare“nonreac-tive”becauseinspectionofarchivaldatadoesnotchangethebehaviorsbeingstudied.Theycanbelimited,however,becausetheymaynotcontainalltheinformationaresearcherwouldreallyliketohave.
60
Studiesusingarchivalmaterialsoftenrunnoriskatallofembarrassingany-one,butresearchonrelationshipsdoesoccasionallyrequireinvestigatorstoaskquestionsaboutsensitivetopicsortoobserveprivatebehavior.Shouldwepryintopeople’spersonalaffairs
61
ThisisnotanissueIposelightly.Althoughit’senormouslyvaluableandsorelyneeded,relationshipsciencepresentsimportantethicaldilemmas.Justaskingpeopletofilloutquestionnairesdescribingtheirrelationshipsmayhaveunintendedeffectsonthosepartnerships.Whenweaskpeopletospecifywhattheygetoutofarelationshiportoratetheirlovefortheirpartners,forinstance,wefocustheirattentionondelicatematterstheymaynothavethoughtmuchabout.Weencouragethemtoevaluatetheirrelationships,andstimulatetheirthinking.Moreover,wearousetheirnaturalcuriosityaboutwhattheirpartnersmaybesayinginresponsetothesamequestions.Researchers’innocentinqui-riesmayalertpeopletorelationshipproblemsorfrustrationstheydidn’tknowtheyhad.
62
Simulationsandotherobservationalstudiesmayhaveevenmoreimpact.ConsiderJohnGottman’s(2011)methodofaskingspousestorevisittheissuethatcausedtheirlastargument:Hedoesn’tencouragepeopletoquarrelandbicker,butsomeofthemdo.Spousesthatdisagreesourlyandbitterlyareatmuchgreaterriskfordivorcethanarespouseswhodisagreewithgraceandhumor,andGottman’sworkhasilluminatedthespecificbehaviorsthatfore-casttroubleahead.Thisworkisextremelyimportant.ButdoesitdodamageShouldweactuallyinvitecouplestoreturntoadisagreementthatmayerodetheirsatisfactionevenfurther
63
Theanswertothatquestionisn’tsimple.Relationshipscientistsordinarilyareverycarefultosafeguardthewelfareoftheirparticipants(Kimmel,2004).Detailedinformationisprovidedtopotentialparticipantsbeforeastudybeginssothattheycanmakeaninformeddecisionaboutwhetherornottoparticipate.Theirconsenttoparticipateisvoluntaryandcanbewithdrawnatanytime.Afterthedataarecollected,theresearchersprovidepromptfeedbackthatexplainsanyexperimentalmanipulationsanddescribesthelargerpurposesoftheinvestigation.Finalreportsregardingtheoutcomesofthestudyareoftenmadeavailablewhenthestudyiscomplete.Inaddition,whenticklishmattersarebeinginvestigated,researchersmayprovideinformationaboutwherepar-ticipantscanobtaincouples’counselingshouldtheywishtodoso;psychologi-calservicesmayevenbeofferedforfree.
64
Asyoucansee,relationshipscienceisbasedoncompassionateconcernforthewell-beingofitsparticipants.Peoplearetreatedwithrespect,thankedwarmlyfortheirefforts,andmayevenbepaidfortheirtime.Theymayalsofindtheirexperiencestobeinterest-ingandenlightening.Peoplewhoparticipateinstudiesofsexualbehavior(Kuyperetal.,2014)anddatingviolence(Shoreyetal.,2011),forinstance,routinelyhavepositivereactionsandaredistressedveryrarely.That’sreassuring.Still,shouldwebetryingtostudysuchprivateandinti-matematters
65
66
Soit’sprettysillytoexpectthatvalueseducationwillchangeanything.Agovernmentprogramthatseekstoimproverelationshipswouldprobablydobettertofundeffectivetrainingforbetterjobsortoincreasetheminimumwagethantotrytoteachpeopletorespectmarriage.Andclearly,ifweseektopromotehumanwell-being,weneedgoodinformationaswellasgoodinten-tions.Inaculturethatoffersusbizarreexamplesof“love”onTVshowssuchasTheBachelorandTheBachelorette—andinwhichrealmarriagesaremorelikelytobefailuresthantobesuccesses(Cherlin,2009)—itwouldbeunethicalnottotrytounderstandhowrelationshipswork.Intimaterelationshipscanbeasourceofthegrandest,mostgloriouspleasurehumanbeingsexperience,buttheycanalsobeasourceofterriblesufferingandappallingdestructiveness.Itisinherentlyethical,relationshipscientistsassert,totrytolearnhowthejoymightbeincreasedandthemiseryreduced.
67
68
Withthosecautionsinplace,let’snotethatthedataobtainedinrelation-shipstudiescanalsopresentuniquechallengesandcomplexities.Herearetwoexamples:
69
Paired,interdependentdata.Moststatisticalproceduresassumethatthescoresofdifferentparticipantsareindependentofeachother—thatis,oneper-son’sresponsesarenotinfluencedbyanyoneelse’s—butthat’snottruewhenbothmembersofacoupleareinvolved.Wilma’ssatisfactionwithherrelation-shipwithFredisverylikelytobeinfluencedbywhetherornotFredishappytoo,sohersatisfactionisnotindependentofhis.Responsesobtainedfromrela-tionshippartnersareofteninterdependent,andspecialstatisticalproceduresareadvisableforanalyzingsuchdata(e.g.,West,2013).
70
Threesourcesofinfluence.Furthermore,relationshipsemergefromtheindi-vidualcontributionsoftheseparatepartnersandfromtheuniqueeffectsofhowtheycombineasapair.Forexample,imaginethatBettyandBarneyhaveahappymarriage.OnereasonforthismaybethefactthatBarneyisanespeciallypleasantfellowwhogetsalongwellwitheveryone,includingBetty.Alterna-tively(or,perhaps,inaddition),Bettymaybetheonewho’seasytolivewith.However,BettyandBarneymayalsohaveabetterrelationshipwitheachotherthantheycouldhavewithanyoneelsebecauseoftheuniquewaytheirindividualtraitscombine;thewholemaybemorethanthesumofitsparts.Relationshipresearchersoftenencounterphenomenathatresultfromthecombinationofallthreeoftheseinfluences,thetwoindividualpartnersandtheidiosyncraticpartnershiptheyshare.Sophisticatedstatisticalanalysesarerequiredtostudyallofthesecomponentsatonce(Kenny&Ledermann,2010),anotherindicationofthecomplexityofrelationshipscience.
71
Sowhat’smypointhereI’venotedthatstud-iesofcloserelationshipstackleintricatemattersandthatstatisticalsignificancetestinginvolvesprobabilities,notcertainties.ShouldyoutakeeverythingIsaywithagrainofsalt,doubtingmeateveryturnWell,yesandno.Iwantyoutobemorethoughtfulandlessgullible,andIwantyoutoappreciatethecomplexitiesunderlyingthethingsyou’reabouttolearn.Remembertothinklikeascientist:Nostudyisperfect,butthetruthisoutthere.Weputmorefaithinpatternsofresultsthatareobtainedbydifferentinvestigatorsworkingwithdifferentsamplesofparticipants.Wearealsomoreconfidentwhenresultsarereplicatedwithdiversemethods.
72
Forthesereasons,scientistsnowdofrequentmeta-analyses,whicharestudiesthatstatisticallycombinetheresultsfromseveralpriorstudies(e.g.,Roblesetal.,2014).Inameta-analysis,aninvestigatorcompilesallexistingstudiesofaparticularphenomenonandcombinestheirresultstoidentifythethemestheycontain.Ifthepriorstudiesallproducebasicallythesameresult,themeta-analysismakesthatplain;iftherearediscrepancies,themeta-analysismayrevealwhy.
73
Withtoolslikethisatitsdisposal,relationshipsciencehasmadeenormousstridesdespiteitsshorthistoryandthecomplexityofitssubjectmatter.Anddespitemyearliercautions,(nearlyallof)thethingsI’llsharewithyouinthistextaredependablefacts,reliableresultsyoucanseeforyourselfifyoudowhattheresearchersdid.Evenmoreimpressively,mostofthemarefactsthathadnotbeendiscoveredwhenyourparentswereborn.
74
Inmydesiretohelpyoubemorediscerning,I’vespentalotofthischap-ternotingvariousprosandconsofdiverseprocedures,usuallyconclud-ingthatnosingleoptionisthebestoneinallcases.Ihopedtoencourageyoutobemorethoughtfulaboutthecomplexitiesofgoodresearch.Butinclosing,letmereassureyouthatrelationshipscienceisinbettershapethanalloftheseuncertaintiesmaymakeitseem.Whenrelationshipsciencebegan,thetypicalstudyobtainedself-reportsfromaconveniencesampleofcollegestudents(Cooper&Sheldon,2002),andmanystudiesarestillofthatsort.However,researchersarenowroutinelystudyingmorediversesampleswithsophisticateddesignsthatemploymorecomplexmeasures,andthevarietyofmethodswithwhichresearchersnowstudyrelationshipsisastrength,notaweakness(Ickes,2000).Furthermore,thefield’sjudiciousabilitytodifferentiatewhatitdoesanddoesnotyetknowisamarkofitshonestyanditsdevelopingmaturityandwisdom.
75
Peoplelikeeasyanswers.Theyliketheirinformationcut-and-dried.Manypeopleactuallyprefersimplenonsense—suchastheideathatmencomefromMarsandwomencomefromVenus—tothescientifictruth,ifthetruthishardertograsp.However,asanewconsumerofthescienceofrelationships,youhaveanobligationtopreferfactstogossip,evenifyouhavetoworkalittlehardertomakesenseoftheircomplexities.Don’tmistakescientificcautionforalackofquality.Tothecontrary,Iwanttoleaveyouwiththethoughtthatitdemon-stratesscientificrespectabilitytobeforthrightaboutthestrengthsandweak-nessesofone’sdiscipline.It’smoreoftenthefraudsandimposterswhoclaimtheyarealwayscorrectthanthecautiousscientists,whoarereallytryingtogetitright.
76
ChrisandKelseyhadtoparticipateinresearchstudiesiftheywantedtopasstheIntroductoryPsychologycoursetheyweretakingtogether,sotheysignedupforastudyof“RelationshipProcesses.”Theyhadbeendatingfor2months,andthestudywasseeking“premaritalromanticcouples,”andtheylikedthefactthattheywouldbepaid$5iftheybothparticipated.So,theyattendedasessionwithadozenothercouplesinwhichtheywereseparatedandseatedonoppositesidesofalargeroom.Theyreadandsignedapermissionformthatnotedtheycouldquitanytimetheywantedandthenstartedtoworkonalongquestionnaire.
77
Someofthequestionswereprovocative.Theywereaskedhowmanydif-ferentpeopletheyhadhadsexwithinthelastyearandhowmanypeopletheywantedtohavesexwithinthenext5years.Then,theywereaskedtoanswerthesamequestionsagain,thistimeastheybelievedtheotherwould.Chrishadneverponderedsuchquestionsbefore,andherealized,oncehethoughtaboutit,thatheactuallyknewverylittleaboutKelsey’ssexualhistoryandfutureintentions.Thatnight,hewasalittleanxious,wonderingandworryingaboutKelsey’sanswerstothosequestions.Havingreadthischapter,doyouthinkthisresearchprocedurewasethicalWhy
Thescientificstudyofrelationshipsisarecentendeavorthathascomeofageonlyinthelast35years.Thefieldhasnowgrowntoincludethestudyofalltypesofrelationshipsintheirnaturalsettingsaroundtheworld.
Researchquestionscomefromanumberofsources,includingpersonalexperience,recognitionofsocialproblems,theresultsofpriorresearch,andtheoreticalpredictions.Thequestionsusuallyseekeithertodescribeeventsortodelineatecausalconnectionsamongvariables.
Conveniencesamplesarecomposedofparticipantswhoareeasilyavailable.Representativesamplesaremorecostly,buttheybetterreflectthepopulationofinterest.Bothtypesofsamplescansufferfromvolunteerbias.
CorrelationalDesigns.Acorrelationdescribesthestrengthanddirectionofanassociationbetweentwovariables.Correlationsareinherentlyambiguousbecauseeventscanberelatedforavarietyofreasons.
ExperimentalDesigns.Experimentscontrolandmanipulatesituationstodelineatecauseandeffect.Experimentsareveryinformative,butsomeeventscannotbestudiedexperimentallyforpracticalorethicalreasons.
Researchcanbeconductedinlaboratoriesorinreal-worldsettingssuchasacouple’shome.Controloverextraneousvariablesisoftenreducedout-sidethelab.Role-playstudiesallowresearcherstoexamineemotionaleventsinanethicalmannerbutmaynotindicatewhatpeoplereallydoinsuchsituations.
Self-Reports.Withself-reports,participantsdescribetheirownthoughts,feelings,andbehavior,buttheymaymisunderstandtheresearchers’questions,havefaultymemories,andbesubjecttosocialdesirabilitybiases.
Observations.Inevent-sampling,briefobservationsaremadeintermit-tently.Observationsavoidtheproblemsofself-reports,buttheyareexpensivetoconduct,andreactivitycanbeaproblem.
PhysiologicalMeasures.Measurementsofpeople’sbiologicalchangesindicatehowourphysicalstatesareassociatedwithoursocialinteractions.
ArchivalMaterials.Historicalrecordsarenonreactiveandallowresearch-erstocomparethepresentwiththepast.
TheEthicsofSuchEndeavors
Participationinrelationshipresearchmaychangepeople’srelationshipsbyencouragingthemtothinkcarefullyaboutthesituationstheyface.Asaresult,researcherstakepainstoprotectthewelfareoftheirparticipants.
InterpretingandIntegratingResults
Statisticalanalysisdeterminesthelikelihoodthatresultscouldhaveoccurredbychance.Whenthislikelihoodisverylow,theresultsaresaidtobesignificant.Somesuchresultsmaystillbeduetochance,however,sothethoughtfulconsumerdoesnotputunduefaithinanyonestudy.Meta-analysislendsconfidencetoconclusionsbystatisticallycombiningresultsfromseveralstudies.
AFinalNote
Scientificcautionisappropriate,butitshouldnotbemistakenforweak-nessorimprecision.Relationshipscienceisingreatshape.