Double Standards in the Evaluation of Men and Women Author(s): Martha Foschi Source: Social Psychology Quarterly, Vol. 59, No. 3, Special Issue: Gender and Social Interaction (Sep., 1996), pp. 237-254 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2787021 . Accessed: 18/03/2011 12:26 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=asa. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to Social Psychology Quarterly. http://www.jstor.org Social Psychology Quarterly 1996,Vol. 59, No. 3, 237-254 Double Standardsin the Evaluationof Men and Women* MARTHA FOSCHI University ofBritishColumbia studieson genderand double Thisarticlepresentstheresultsfromtwoexpectation-states standardsfor task competence.The emergenceof such standards under several conditions is investigated. In bothstudies,menand women,participating in experimental and thenas a teamin solvinga perceptual opposite-sex dyads,workedfirstindividually task.As predicted,resultsfromExperiment I show thatalthoughsubjectsof bothsexes womenwereheldtoa stricter achievedequal levelsofperformance, standardofcompetence was morepronouncedwhenthereferent thanmen.Thisdifference ofthestandardwas the 2 investigates theextentto whichthedoublestandard partnerratherthanself.Experiment is affectedby level of accountability for one's assessments.Resultsshow a significant was low, butnotwhenit was difference bysex ofreferent ofstandardwhenaccountability inselfand inpartnerreflected increased.In bothstudies,measuresofperceivedcompetence reportedstandards,as predicted.Theoreticaland practicalimplications of thesefindings are discussed. liking,controlmotives,or groupprejudices) can affectwhatstandardswill be used. The In task-orientedgroups, the processes researchpresentedhere links standardsto wherebyindividualsassign competenceto groupprejudices.Specifically,the objective eachotherhavecrucialconsequencesfortheir is to studywhether,under certaincondifutureinteraction.The standardsused to forcompetenceare tions,different standards judge thatcompetence,in turn,play a key used to evaluatemembersof different social role in these processes:because such stancategories-evenwhen theyperformat the thelevel and type dardsare normsspecifying same level. The workfocuseson genderas of outcome requiredto inferability,the the basis of sucha doublestandard.In other inferencevaries dependingon the standard words, giventhatwomenoftenare assigned used. For example, a score of 70% is lower levels of abilitythan men, to what sufficient evidenceforabilityif thestandard extent is this assignmentthe resultof the is 60% or higher,butthatscorebecomesan if the standardis applicationof a double standard,which is unconvincing performance forthefemaleperformers? Whatare at least 80%. A similarexample may be stricter some of the variables that affect when this constructed forlack of ability.Because the occurs? The states research expectation assignmentof task competenceis directly its branchon status relatedto achievingstatusand influencein a program,particularly thetheoretical characteristics, provides backto understandhow group, it is important ground. are set. standards the Expectationstatestheoryinvestigates In manytasks, standardsare not clearly ofpowerandprestigehierachies defined beforehand.As a result, factors development unrelatedto the task (e.g., interpersonalin task groups(Bergeret al. 1977; Berger, Wagner, and Zelditch 1985; Wagner and Berger1993; Websterand Foschi 1988). The * The studiesreported in thisarticlewerecarriedout tradition,and undera researchgrantfromthe Social Sciences and theoryhas a long-standing providestrongsupportfor HumanitiesResearch Council of Canada (# 482-88- empiricalfindings 0015, StrategicGrantsDivision, Women and Work its predictions.(For assessments,see, for acknowledgethissupport.I would example, Deaux 1985; Wiley 1986.) A Theme). I gratefully also like to thankRicardo Foschi for his work in developingthe computerprogramforthe experimental centralconceptin thisprogramis thatof a any valued attribute task;LarissaLai, MarieLembesis,and KirstenSigerson "statuscharacteristic," for theirassistancein runningthe experiments;and implying taskcompetence.Such characterison an earlier ticsconsistof at leasttwo states(e.g., either SandraSchmidtforhereditorialcomments to the versionof this article. Direct correspondence or low levelof mechanicalability,either ofAnthropology andSociology, high authorat theDepartment limitedor extensiveformaleducation),one of Universityof British Columbia, Vancouver, B.C., Canada V6T IZI. whichis evaluatedmorepositivelythanthe THEORETICAL BACKGROUND 237 238 SOCIAL PSYCHOLOGY QUARTERLY are also definedas As an example,I assume genderto be the other.These attributes rangingfromspecificto diffuse,depending diffuse attribute. Thus,whena mansucceeds, on theirperceivedapplicability.A specific two consistentpieces of information (status is associatedwithwell-defined and level of performance) characteristic are availableand a character- definiteinferenceof competenceresults. a diffuse expectations; performance istic carries,in addition,predictionsabout Successby a woman,however,represents an in a wide, indeterminate variety inconsistent performance combination; therefore a weaker of tasks.In manysocieties,gender,'ethnic- inferenceof abilityensues. On the other dif- hand,failurewill be viewedas a consistent ity,and socioeconomiclevel constitute Thus women,for outcomefor a woman but not for a man. fuse statuscharacteristics. example,oftenare expectednot only to be Consequently thisoutcomewillbe interpreted to menin variousspecificskills,but as indicating inferior lack of abilitymorestrongly in also to be inferiorin generalcompetence. the female performerthan in her male (For discussionsand reviewsof genderas a counterpart. For a morefullydetailedpresensee Ridgeway1993; Wagner tationof these hypotheses,includingscope statusattribute, 1988.) conditions,see Bergeret al. (1977); experioftaskabilitymaybe made mentaltests(e.g., Pughand Wahrman1983; The assignment evaluations)or in- Wagner,Ford,and Ford 1986) provideclear directly (fromperformance (on thebasisof statuscharacteristics).empiricalsupport. directly This assignment,in turn,resultsin perfor- Foschi(1989) proposesbothan elaboration beliefsabouthow and an extensionof the above formulation. manceexpectations-stable willcarryouta giventask The proposal incorporatesideas from (1) wella groupmember aretheoretical attribution in thefuture. Suchexpectations workexamining howtheperceived between causes of successand failureare affected therelationship constructs mediating by and thepowerand presstatuscharacteristics the performer'smembershipin a social orderis definedin tigeorderofthegroup.This category (Deschamps 1983; Hansen and behaviors: termsof thefollowinginterrelated O'Leary 1985; Whitley, McHugh,and Frieze intheoffer andacceptance unequaldistribution 1986) and (2) expectationstatesresearchon thetypeofevalofperformance opportunities, theeffectsof standardson the interpretation exuationsreceived,andtheratesofinfluence of performance outcomes(Foschi and Freestatestheory focuseson exerted.Expectation man and Hart1985). 1991; Foschi, Warriner, plaining how this order originatesand is At the core of this is the notionthat theory basedonthethemaintained whileinterventions in differences inferred ability persistin spite theinequalities. waysofredressing oryidentify of because equal performances different with is concerned The presentresearch in whichexpectations are based on standardsfor competenceare applied to situations members.It is both status and performanceevaluation. higher-and to lower-status useful standards in to termsof their classify in thosecases in I am interested Specifically, (see Foschi and Foddy 1988). A in statusbutperform at the strictness whichactorsdiffer same level (either well or poorly). For strict standardfor ability requires more theoreticalas well as applied reasons, the evidence of competence (e.g., a larger case occurswhenthesetwo numberof correctresponses,attainedovera mostinteresting tasks) than constitute itemsof information verydifferentlargernumberof more difficult groundsforcompetence.Thus let us assume does a lenientstandard.Conversely,a strict evaluationsareobjective standardfor lack of ability toleratesless thattheperformance thandoes a lenient (i.e., the operationof biases has been evidenceof incompetence in standard.Morever, a situationinvolves a blocked)and thatthe statuscharacteristic questionis diffuse.Accordingto expectation "universal" standardif the same set of is appliedto all performers; if will requirements statestheory,the resultingexpectations that this is not the case, "double" or even of all the information be a combination obtain.In Foschi(1989) the actorconsidersrelevantto the situation. "multiple"standards I proposethat,undercertainconditions,a status characteristic that differentiates the lI use thetermsex to referto biologicaldifferences into two classes activates the use performers betweenmenandwomen,andgenderforculturalaspects forcompetenceand for of these differences.See Foschi, Lai, and Sigerson of double standards lackofcompetence, bothof whicharestricter (1994) fora discussionof thispoint. DOUBLE STANDARDS 239 No doublestandards The higherthe nevertheless. benefiting forlower-status performers.2 are predictedfor(2) or inconsistency betweenstatusand outcome, the male performer thestandard. thestricter (3). and Once a double standardis activated,it I propose the status characteristics multiplestandardstheoryfor a situationin affectsthe degreeof abilitythatis inferred. to whicha person(self) workson a joint task The applicationof a morelenientstandard are themanensuresthatmoreabilityis assigned witha partner(other).The propositions statedfromself's pointof view and applyto to him, regardlessof level of performance, or diffuse, thanto the woman withthe same record.3 any statuscharacteristic-specific with Thus double standardscontributeto the individually as well as in combination forthecase maintenanceof the initial,status-basedasothers.HereI presenta summary of gender,and assume the followingscope signmentof competenceand are another of thepowerandprestigeorderof conditions:(1) self values the task and is component bothto do it well and to arriveat the group. The practice is both subtle motivated correctassessmentsof the two performers' (because it does notinvolveeitherdevaluing theperformance competence;(2) selfis awarethatthepartner or overvaluing directly)and is not necessarilyconscious(because an actor is oftheoppositesex (i.e., sex ofperformer a salientfactorin thesituation); (3) selftreats does not have to formulatesuch standards (i.e., explicitlyin order to use them). For an genderas a diffusestatuscharacteristic an indicationof women's inferiorcompe- alternativebut compatibleformulation on also proposedwithinexpectence); (4) self knows the resultsof each doublestandards, (as- tationstatestheory,see Foddyand Smithson person's prior individualperformance sessedby a thirdparty)and believestheseto (1989); also see BiernatandManis(1994) and be unbiased, but no previouslyset and Biernat,Manis, and Nelson (1991) fortheir are availableby which workon stereotypes and shifting agreed-upon standards standards.4 Multiple standardsfor competencethat to inferability(or lack thereof)fromthose at thesame benefitthe higher-status results;(5) bothpersonsperform are comperformer level; and (6) self has no othergroundson mon in a varietyof everydaytask settings, whichto base assessmentsof task compe- rangingfrominformal groupsto formalwork contexts.The social psychologicalliterature tence. Foschi(1989) specifiesthatselfmaydefine containsseveraldescriptions of theiroperathe task in one of the followingways: (1) 3 Othertypesof gender-based (3) explicitlydissomasculine,(2) feminine, double standardsalso in codesofmorality andcriteria ciated from gender, or (4) not explicitly exist,suchas differences Similarly,double (or even definedin relationto gender.It is predicted for physicalattractiveness. multiple)standardsfor competencemay be based on thatin (1), selfwill tendto activatea stricter attributes otherthanstatuscharacteristics (e.g., level of forcompe- interpersonalliking, or personal qualities such as standard forthefemaleperformer, In general,a multiplestandard refersto the Thisalso friendliness). tenceas wellas lackofcompetence. criteriaforassessingthe same trait(or will occur in (4), where genderand task use of different or performance)in different categoriesof become relatedthrough"status generaliza- behavior people. The presentresearchis concernedonly with tion." In this process, a status attribute gender-based in inferring doublestandards ability.For a becomesrelevantto the task at hand unless reviewof varioustypesofmultiplestandards,see Foschi to thecontrary. (1992). thereis specificinformation 4 Also, in some conditions, thelower-status personis In such a case, thedoublestandardswill be treated with a more lenient (but not explicitly stated) less pronouncedthan when the task is standardthanthe higher-status counterpart, and is told themselves thathis or herperformance masculine,buttheywill manifest is a sufficient demonstration 2 doublestandardto refereitherto I use theexpression one personwhois assessingtheperformances byhimself in a particular orherself andbya specificpartner context, or to theaveragerequirement appliedto selfand to other by a numberof individuals.I use the plural double standards,however,ifI wishto emphasizethata double standard of different maybe activated,either magnitude by each of severalindividualsor by the same person whenmakingassessmentsacross variouscircumstances and/or partners. ofabilitywheninfactitis not.Suchstatements arerarely theresultof a genuineerrorin assessment butratherare madeforotherpurposes:forinstance,to avoidconveying a poorevaluationorto meetquotas.Forthisreason,such falloutsidethescopeofthisarticle. patronizing standards Theyare worthinvestigating, however,because depriving the lower-status actorof a truthful appraisalis yet anotherpracticethroughwhichthe statusquo can be maintained.For discussionsof thisand othertypesof morelenientstandards forthe lower-status person,see, for example, Blalock (1979: chap. 4) and Epstein (1970:978). 240 SOCIAL PSYCHOLOGY QUARTERLY gender,as well as These are sex linkageof task and level of regarding tion,particularly Thus the task was definedas a number of studies providing indirect performance. evidenceof thispractice(see Foschi 1992 for masculine,and bothactorsperformed at an a review).To myknowledge,however,only averageratherthanan extremelevel. These Biernatet al. (1991), Foddy and Smithson twodesigndecisionsmaximizethelikelihood (1989), and Foschi (1989) have proposed thatdistinctdouble standardswill be actiformaltheoriesabout such standards.(Also vated:shouldthispracticenot appearunder see Jassoand Webster1995 fora theoretical these conditions,it would be unlikelyto analysiscombiningideas fromthe multiple appear at other levels of these variables. and thedistributive justiceformula- Thus, regardingsex linkage,I consideredit standards hereis partof an crucialto be able to demonstrate firstthat tions.)The workpresented ongoingresearchprogramdesignedto test double standardsindeedemergedunderthe conditions of an explicitly Foschi's (1989) theory.Two studiesalready facilitating mascunamely linetask.The resultsthencouldbe used later havebeencompletedin thisprogram, Foschietal. (1994) andFoschi,Sigerson,and as a baselineforothersex-linkage conditions. Lembesis(1995); bothaddressdouble stan- Similarly,I chose an average level of on the assumption dardsin an extendedcontextwhereselfis an performance thatextreme a moredefinite evaluatorof othersbut not a performer.5levels,becausetheyconstitute Resultsfromthefirststudyshowthat,given indicationof eitherabilityor lack of ability, an averageoutcomeby a male and a female wouldbe morelikelyto escape statuseffects men but not womenfavoredthe thanwouldaverageoutputs.(See Foschiet al. performer, In thesecondstudy,subjectsof both 1995. For discussionsof thispointin relation former. an advantage to gendereffectsin particular,see Epstein sexes gave themale performer whenthetwohad 1970; Kalin and Hodgins 1984; Lott 1985; overhisfemalecounterpart achieved an average outcome, but this NievaandGutek1980;WallstonandO'Leary advantagewas reversedwhen theirperfor- 1981.) Let us thenassumeopposite-sex mances were outstanding.In both studies dyads,and double standardswere measuredindirectly. conider themfromeach person'spointof The two experiments reportedhere, on the view. Thus, eitherself is a man and his otherhand, assess double standardsdirectly partneris a woman,or selfis a womanand and investigate theiroccurrencein theorigi- herpartner is a man. The taskcontextmeets nal self-and-other context of expectation all theotherconditionsspecifiedso far,and statesresearch.The objectivesare (1) to test bothpersonsperformsuccessfully. The hykey hypotheseson the activationof double pothesestobe testedconcernsex ofperformer andtheresulting levelsofperceived and thatperson'srolein thedyad(eitherself standards theroleof or other). I propose the followingtwo and (2) to investigate competence, two additional selected factors in those hypotheses: thestandard is setforself 1. (Gender).Whena mananda processes:whether Hypothesis at thesame level of success or forother,and the level of accountabilitywomanperform as theirrespectiveopposite-sexpartners, forself's assessments. the woman's performance will tend to be assessedwitha stricter standard forabilitythan EXPERIMENT1 the man's. As a result,the level of compeOBJECTIVES tenceinferred aboutthewomanwillbe lower than that inferred abouttheman. This studyconcernsa situationin which A of expectationstates major assumption twofactors,assumedto affectthemagnitude of competenceare of the double standard,are held constant. theoryis thatassignments relativeto theactorsin a givensituation; that a result from is, they comparison between 5Although Foschi (1989) largelyoverlapswiththe original status characteristics theory,the two differ these actors (Bergeret al. 1977: chap. 4; Bergeret al. 1985). Hence, in the dyads slightlyin scope. The inclusionof scope conditions(4) and (5) makesthe formermorelimitedthanthe latter, studiedhere,the man shouldexperiencethe althoughrelaxingthese in futurework should be a same competenceadvantageover his female relatively straightforward matter.On theotherhand,by whether he occupiestheroleof selfor includingsituations whereselfis an evaluatorof others partner differences but not a performer, Foschi (1989) extends status of other.In otherwords,although characteristics theory. due to role may exist (stemming,for DOUBLE STANDARDS 241 forthisstudy.The motives),sta- researchor werepretested example,fromself-enhancing were seated individually at factorunder two participants tusis assumedto be thestronger the specified scope conditions(also see adjacent stations equipped with personal computers said tobe linkedto eachother.The Foschiet al. 1994). Thus: Hypothesis2. (Role). The process de- stationswere separatedby a partitionand scribedin Hypothesis1 will occur whether subjectswereprecludedfrombothseeingand occupytherole talkingwith each other.The experimenter theactorsunderconsideration dyads. statedthatthe purposeof the studywas to of selfor of otherin theirrespective will investigate on a "contrast performance sensiIn otherwords,thefemaleperformer be assessedby herselfas well as by hermale tivity"task in two simulatedworkenvironabilitystandardthan ments: an individualand a team setting. partnerwitha stricter Conversely,the Subjectswere informed thatonly the memwill her male counterpart. willbe assessedbyhimselfas bers of the researchstaffwould see their maleperformer well as by his femalepartnerwitha more individualresponses,and that theirnames wouldbe keptconfidential. Instructions were lenientabilitystandard. wordedso as to motivateparticipants to do well (i.e., to be "task-oriented"). Each team METHOD thatits two memberswere of was informed at theuniversity thesameyearandfaculty but Subjectsand Experimenters of different sex. Contrastsensitivitywas discoveredabilityand Subjectswere72 men and 72 women,all describedas a recently fromthefaculties(schools)of of highpotentialvalue to a varietyof tasks. undergraduates of British Reliableresearchwas said to haveshownitto Artsand Science at theUniversity Columbia. Average ages (with standard be mainlyintuitiveand relativelyspecific. deviationsin brackets)were 18.58 [0.96] for Thus the subjectsheard that "althoughno the men and 18.80 [1.11] for the women. significantrelationshiphas so far been and each person establishedbetweenit and attributes such as was voluntary Participation was paid $8 for the session. A pool of mathematicalskill or artisticability,men in large have been foundto be generallyfar more subjectswas obtainedby recruiting classes; thosestudents accurate than women at solving contrast first-and second-year beyond sensitivity problems." who had takencoursesin psychology task consistsof an introductory level and/orwho had partici- The contrastsensitivity were severaltrials.On each trial,subjectsview a experiments patedin social psychology area coveredto about the same rectangular excludedas prospective subjects. was teamedwithanother extentby smallerrectanglesof two different Each participant of theoppositesex, and teamswereassigned colors.Subjectsmustdecidewhichofthetwo in the overallpattern. at randomto one of threeconditions:(1) colors is predominant higherscoreforselfthanforother;(2) higher The taskis actuallyambiguousto allow for of acceptance/rejection of scoreforotherthanforself;or (3) no score the measurement thathas foreitherperson.The studythuswas a 2 (sex influence.It is a reliableinstrument of subject and partner)x 3 (feedback been used extensivelyin expectationstates condition)design,with24 subjectsper cell. studies. A computerized versiondevelopedexplicEach sessionwas conductedby one of two female researchassistantsof similar age. itlyforthepresentworkwas used here.Brief forthetaskas well as thevisual Special attentionwas paid to maintaining instructions of appearanceand deliveryof stimulithemselveswere presentedon the uniformity acrosssessions. screen.The stimuliwerewhiteand red on a instructions black background.The computerprogram gave the subject10 secondsto look at each Proceduresand Materials himor herfora response, pattern, prompted For comparability,procedures were a and, afterfive seconds, showed the next variant of the standardizedexperimentalpattern.Subjectswereassuredthat,although therewas alwaysa situationdeveloped for expectationstates thetaskappeareddifficult, of theconresearch(Bergeret al. 1977:43-48). Instruc- correctanswer.Computerization taskoffersmanyadvantages, eitherwere adapted trastsensitivity tionsand questionnaires used in previous including(1) eliminatingthe possibilityof fromreliable instruments 242 SOCIAL PSYCHOLOGY QUARTERLY in communicatingquestionaboutit wouldfollowlogically(and errorby the experimenter fromthepreviousquesfeedbackto thesubjects,record- quiteunobtrusively) prearranged influence tionon scores.6 ing theirresponses,and computing Duringthesecondseriesof trials,thetwo rates, and (2) more controlby the experito workas a team time for each subjectswere instructed menterover the presentation patternas well as over the time elapsed and to tryto arriveat a correctchoicein each It also enhancesthecredi- trial. The intentionof thus creating a betweenpatterns. in eachpersonwas to bilityofthecoverstoryandresultsin a highly "collectiveorientation" make him or her assess the relativecompeengagingtask. appearsin tencesof self and other.Subjectswere told An overviewof theexperiment Table 1. Duringthe firstpartof the study, thata team would be awardedtwo points participants in the experimental conditions wheneverbothpersonswere correct.As an forteamwork, each of thesix on the addedincentive (4)) workedindividually ((1) through task and made decisionson 20 patterns.At teamswiththemostpointswouldwin a $20 theend of thisseries,thescoresobtainedby prize. For consistencywith other expectation the two persons (trial by trial as well as states witha similardesign,the experiments overall)appearedon bothcomputerscreens. taskvariedslightly sensitivity during They showed either11 correctanswersby contrast It this now involved two phase. patterns per self and 13 by the partner,or the reverse. had 10 trial; subjects seconds to decide which Next, each person received a printoutof areas containedmore thesescores.Theneach subjectcompleteda of the two rectangular writtenquestionnairethat included several white.The same abilityas in thefirstphase checksand a measureof his or was said to be involved.Aftera subjectmade manipulation a the partner's "choice" was her own standardsfor the higher-scoring decision, relayed.The feedbackwas manipulatedto have contrastsensitivperson"to definitely resultin 20 disagreements and five agreeityability." mentson theinitialchoices.Each subjectthen I chose thescoresof 11 and 13 to indicate made his or her own finalselection.In the an equallyaveragelevel of success by both disagreement trials,this entailedeitherrepersons,for the reasons presentedearlier. mainingwithself's initialchoiceor changing Such a level also has the advantage of it to agree with the partner's.The former givingsubjectsa widerrangeof choicesfor decisionis referred to as an s-response.The settingtheirown abilitystandardthanwould proportion of s-responsesoperationalizes ina more definitesuccess. The latter, in fluencerejection,a variablewhich,in this by the setting,is associatedreliablywithperceived addition,could have been interpreted subjects as implicitlyinformingthem of competencein selfand in other. whatthatstandardshouldbe. On the other At the conclusionof this series,subjects hand, because reportingthat both persons completeda second questionnairethat inreceived exactly the same scores would cluded further checksand one manipulation likelyhave createdsuspicion,thetwo scores additionalmeasureof relativecompetence. weremade to differslightly.This difference This instrument also served to assess any also focuses attentionon the betterper- misunderstandings and/orsuspicionsregardthe factthat,forthis ing the procedures.Next, subjects were formerand highlights eithera interviewed person,outcomeand sex represent to check further on individually combination. theseissues,and thenweredebriefed. consistentor an inconsistent As discussedearlier,degreeof consistency The controlconditions ((5) and(6) inTable is assumed to contributeto the use of a 1) excludedtheindividual performance phase doublestandard. andthecorresponding scores.Becausedouble One item in the questionnaire prompted subjectsto recordbothscores; the immedi- 6 If subjectshad been askedto set an abilitystandard atelyfollowingitemaskedthemto statetheir foreach of two personswithcloselymatchingscores, own abilitystandard forthebetterperformer,theyprobablywouldhave producedmechanicalanswers thesamestandard forboth.On theotherhand, of correctanswersrequired.I indicating as a percentage thesmalldegreeof difference betweenthesescoresdid assumedthat,giventhe scoresreceived,the not justifyasking for standards for ability in one subjectswould most likelyactivateand be performer and for standardsfor lack of abilityin the concerned with this standard. Also, the other. DOUBLE STANDARDS 243 Table 1. Overviewof Experiment 1 Part2 Dependent VariablesMeasured Part1 Manipulation Phase Scores Receivedby Condition (1) (2) (3) (4) (5) (6) Sex of Subject M F M F M F Sex of Partner F M F M F M Subject Partner (Maximum:20) 11 11 13 13 Standards forAbility in HigherScoring Person Perceived Competence in Subject and in Partner - 13 13 11 11 - Yes Yes Yes Yes Yes Yes Yes Yes - - No No Yes Yes standardsare assumedto be appliedto such articulatedsuspicionsabout the conformity scores,thequestionaboutstandardswas not aspects of the study, six misunderstood askedeither,and subjectsworkeddirectlyin crucialsectionsof theinstructions and/orthe questions,and eightshoweda clear the team setting. In all other respects, written procedureswere the same as in the experi- lack of task orientationand/orcollective mentalconditions. The purposeof thecontrol orientation. (Several subjectsfell into more groupswas to providebaselinemeasuresof thanone of thesecategories.)The analysis competenceinferred fromonlytwo itemsof thatfollowsincludesonly the 129 retained information, namelysex of selfand of other, subjects.7 and sex linkageof task. In sum,I expectedtheresultsto showthat ManipulationChecks thefemalebetterperformer wouldbe judged The postexperimental questionnaireconby a stricter abilitystandardthanher ma4e counterpart, regardlessof theirroles in the sistedof a varietyof items,some of which dyad.The same setsof scoresthuswouldbe werefillersadded to maintaintherealismof interpreted differently, dependingon sex of the cover story.Includedin this instrument self and of other.As a result,measuresof were six five-pointbipolar scales used to of the task. wouldreflecta higher assess the subjects'perceptions perceivedcompetence means and The standard devia(in brackets) thanfor level forthe male betterperformer tions to items corresponding these were: his femalecounterpart, of regardless again would be largerin creative(1)-routine (5): 2.79 [1.01]; imporrole. This sex difference the controlgroupsthanin the experimentaltant(1)-unimportant(5): 2.96 [0.88]; easy (1)becausein thelattertheeffects of (1)-difficult(5): 3.93 [0.89]; intuitive conditions, would be moder- learned (5): 2.04 [0.80]; and masculine information gender-related ated by the equalityin the two persons' (1)-feminine (5): 2.97 [0.49]. Analysisof varianceforeach attribute showedno statistilevels. performance effectsfromeithertypeof cally significant dyad8or feedbackcondition,as anticipated. The thefirstfourattributes figuresregarding RESULTS are withinexpectedrangesand indicatethat On the basis of the information obtained questionnaire 7 In line withcommonlyaccepteduse in statistical throughthe postexperimental ifp ' .05, ofmarginal and interview,15 subjects(eightmen and analysis,I call a resultsignificant if p > .05 and ' .10, and seven women) were excluded from the or borderlinesignificance otherwise.In the lattercase, I omitthe nonsignificant 10.4% of the presentationof details fromthe statisticaltests. The analysis.This figurerepresents totalnumberof participants. Rejectionrules expressionp = .000 resultsfromroundingoffto three be- decimalplaces. were conservativeand were formulated 8 I use sex composition of dyad,typeof dyad,sex of and thispercentage compareswell forehand, and other, and sex of subject and partner with the exclusionrates of similarexperi- self interchangeably. Because thepartnerwas alwaysof the ments.The excludedsubjectsmaybe classi- oppositesex, forbrevitysometimesI simplyreferto sex fied as follows: seven volunteeredwell- ofsubject. 244 SOCIAL PSYCHOLOGY QUARTERLY of the conditions,by sex of referent) had createdperceptions was signifitheprocedures taskas intended.The data on thefifthitem, cant,as expected. to the instruc- Perceivedcompetence however,show that,contrary in selfandotherwas tions, subjects on average consideredthe measuredprimarily through level of rejection I examinethis of influencefromthe partner,calculatedas abilityto be gender-neutral. pointin thediscussionsectionbelow. the proportionof times a subject did not The questionnairealso included similar changehis or herinitialanswer.The higher scales to assess the subjects' dispositions this figure, the higher self's perceived toward their task. Averages and standard competence advantageoverother.In addition (1)-uninterested to this behavioralmeasure,I obtainedone deviationswere: interested (1)-unmotivated auxiliaryindicator:perceivedtask abilityin (5): 2.35 [0.91]; motivated (5): 2.43 [0.88]; and involved (1)-unin- self relativeto thepartner.I used a bipolar volved (5): 2.29 [0.88]. As can be seen, scale, with responsesrangingfrom "self values for all threeitemsagain fell within muchworse" (1) to "self muchbetter"(5). neither Resultsfrombothmeasuresappearin Table expectedranges.Also as anticipated, independentvariable had any statistically3. effects.Finally,responsesto two significant Analysis of variance on rejection of otherquestionnaireitems indicatethat all influenceshows significantresults as exsubjectsrecalledexactlythe scoresreceived pected:fromtypeof dyad(M (male subject) by themselvesand by their partner,and = .597, M (femalesubject) = .486, F (1, identified thesex of thelatter. correctly 123) = 31.93, p = .000), fromfeedback condition(M (subjectworsethanpartner)= .476, M (subjectbetterthanpartner)= .592, DependentVariables M (no scores) = .562, F (2, 123) = 12.95,p I begin by examiningthe results on = .000), andfromtheinteraction (F (2, 123) person, = 3.59, p = .031). Simplecontrasts on the standards for the higher-scoring ANOVA on comparisonsof directrelevanceto the hyshownin Table 2. As predicted, main effect p6thesesrevealthat,foreach feedbacklevel, these data shows a significant of womenacceptedmore influencefromtheir (or referent fromsex of betterperformer thequestionon standards) (M (male referent) partnerthan did men fromtheirs.Results = 68.05, M (female referent)= 73.45, F (1, fromthesetestswereas follows:Conditions 82) = 8.49, p = .005), while neitherthis (1) and (2): F (1, 123) = 5.84, p = .017; was signifi- Conditions(3) and (4): F (1, 123) = 3.75, p person'srole northeinteraction cant. Simplecontrasts reveala strongeffect = .055; Conditions (5) and(6): F (1, 123) = in the"other"conditions of sex ofreferent (F 29.68, p = .000. Findingsfromthe two (1, 82) = 5.75, p = .019) but a marginally controlgroupsindicatethat,in theabsenceof effectin the "self" groups(F (1, performance therewas a clear significant information, 82) = 3.02, p = .086). Neitherof the other sex differencein expectations:men felt two contrasts(between"self" and "other" superiorto womenin task ability,whereas Person(in Percentages) 1: StandardsforAbilityin theHigher-Scoring Table 2. Experiment Condition (as perTable 1) Sex and Role of Referent of Standards N M SD Other 72.50 7.03 (1) Male subject scoringworsethan femalepartner 22 F (2) Femalesubject scoringworsethan male partner 23 M Other 66.39 11.24 (3) Male subject scoringbetterthan femalepartner 21 M Self 69.86 6.81 (4) Femalesubject Scoringbetterthan male partner 20 F Self 74.50 8.09 245 DOUBLE STANDARDS 1: PerceivedCompetencein Self and in Partner Table 3. Experiment Condition (as perTable 1) Self's Ability Relativeto Partner's Rejectionof Influence fromPartner N M SD M SD (1) Male subject Scoringworsethan femalepartner 22 .516 .103 2.57 .791 (2) Femalesubject scoringworsethan malepartner 23 .437 .109 2.13 .548 (3) Male subject scoringbetterthan femalepartner 21 .624 .129 3.10 .301 (4) Femalesubject scoringbetterthan male partner 20 .558 .122 3.00 .324 (5) Male subject and femalepartner, no scores 21 .655 .089 3.17 .577 (6) Femalesubject and malepartner, no scores 22 .473 .102 2.18 .501 In sum, results from the manipulation womenheld the oppositeview. Because the subjectsperceivedthetaskas gender-neutralchecks show thatoverall, subjects'percepratherthan masculine,what these resdlts show is the singleeffectof sex composition ofdyad.Thustheyestablishthatgenderwas a givingthemthesame scores.Table 3, however,shows subjectsformeddifferent in Conditions expectations forthe subjects. that diffusestatuscharacteristic (1) and (3), and in Conditions(2) and (4). Thus it is Data fromthe othertwo comparisonsshow, possibleto view each of thesefourconditionsas either as anticipated,that this sex effect was confirming or disconfirming (albeit not definitely) the formedin thecorresponding controlgroups. byequalityin the expectations butnoteliminated tempered One thencan comparethesefindings withthepredictions levelsof performance. thatthe originalstatuscharacteristic formulation makes Resultson theauxiliarymeasureshowthe for such a situation,and with resultssuch as those same pattern as those on rejection of obtainedby Wagner et al. (1986). For the current whererelativesuccess and relativefailure influence,again as expected. On relative experiment, ANOVA indicates areambiguous,thattheorywouldpredictthatdifferences abilityof selfand partner, controland disconfirmation conditionswill be effects fromtypeofdyad(M (male between significant whereasthosebetweencontroland confirmasignificant, subject) = 2.94, M (female subject) = 2.41, tionconditionswill not. The resultsfromcontrastson F (1, 123) = 28.54, p = .000), from rejectionof influencesupportall fourof thesepredicfeedbackcondition(M (subject worse than tions,as follows:Conditions(3) and (5) (male subjects, F (1, 123) = 0.84, p = .362; Conditions partner)= 2.35, M (subject betterthan confirmation): (2) and (6) (female)subjects,confirmation): F (1, 123) = 1.20, p = .276; Conditions(1) and (5) (male subjects,disconfirmation): F (1, 123) = 17.27, p = .000; Conditions(4) and (6) (femalesubjects,disconfiranalysesare as follows: mation):F (1, 123) = 6.28, p = .014. Contrastson Resultsfrominternal relativeabilityyield highlysimilarresults.In Conditions(1) and (2): F (1, 123) = 7.46, p reported addition, althoughon both measures of perceived = .007; Conditions (3) and(4): F (1, 123) = competenceone would have expectedthe resultsfrom 0.32, p = .572; Conditions(5) and (6): F (1, Condition(3) to be higherthanthosefromCondition(5), theformer arestillhighenoughto support all hypotheses .000).9 123) = 36.10,p= involving them.Perhaps,giventhedifficulty of thetask, ambiguousfailurerelativeto the partnercommunicates 9 This experiment was designedto equate the two inferiority muchmoredefinitely thanthecorresponding without actually ambiguoussuccesscommunicates superiority. actorsat an averagelevelofperformance partner)= 3.05,M(no scores) = 2.66,F(2, 123) = 18.11, p = .000), and from the interaction(F (2, 123) = 7.35, p = .001). 246 SOCIAL PSYCHOLOGY QUARTERLY in it conditionsmay be due to the following: tionsofthetaskandoftheirinvolvement had been createdas expected.Findingson althoughstatus processes may be equally standardsindicate that the predictedsex- strongforselfand forother,theycouldhave by additionalprocessesin the effectis more pronouncedfor been tempered of-referent otherthan for self, and data on perceived case of self. When subjects had to set forthepartner (Conditions(1) and withthisdifference standards areconsistent competence (2)), genderwas theonlyitemof information in everyrespect. the two referents.Thus the differentiating distinction betweenthemale and thefemale DISCUSSION partner was a sharpone. On theotherhand, when had to set standardsfor subjects In myview,twoaspectsoftheresultsmerit special attention,namelythose on (1) sex themselves(Conditions(3) and (4)), other in additionto sex of self may forselfand information linkageof taskand (2) standards have become salient.For example,subjects forother. in have recalled theirown performances may task identified the instructions the Although as masculine,subjects'responsesindicatethat the past on similartasks-somethingthey instead, on average, they viewed it as could not do regardingthe partner.As a shouldhave Clearly,the sex linkagema- result,theadditionalinformation gender-neutral. standards for self less differentiated made the In future enough. strong was not nipulation work it would be useful to reword the thanthoseforother. Althoughno data were collectedin the to place moreemphasison that instructions thatcould be used to test this experiment linkage,and perhapseven to display "evimake thefollowingarguments dence" of it on graphsand tables. Notice, interpretation, thusunintend- it highly plausible. In the first place, however,thatthe experiment checkshad revealedno further in the manipulation a testofthehypotheses edlyconstitutes differences between "self" and "other" in results burdenof proofcase, and therefore a strictertest of the double standards conditions.A thoroughexaminationof all items,includingfillers,yielded quwstionnaire 10 formulation. the same result. Second, workon attribution main Regardingstandards,the significant a to overestimate the theory reports tendency sex of referent Hypothefrom supports effect effectsfrom importanceof situationalfactorsin assesssis 1, while the nonsignificant overestirole of referentand from the interaction mentsof self,and a corresponding in factors mation of assessing dispositional supportHypothesis2. Contrastsadd more informationon how these variables are others(Jonesand Nisbett1972; Monsonand related;the resultsfromthreeof the four Snyder1977; L. Ross 1977; M. Ross and contrasts (Conditions(1) and (2), (1) and (4), Fletcher1985). Empiricalfindingssupport and(2) and(3)) areclearlyas expected,while the existenceof this tendency("the fundaerror")undera varietyof betweenConditions(3) and (4) is mentalattribution thecontrast All thingsconsid- conditions.Accordinglyit would be more of borderline significance. receivesubstantial likelythatother,ratherthanself, would be ered,then,thehypotheses termsin this study. judged in stereotypical support. in the supportreceivedby Moreover,here the tendencywould have The difference Hypothesis1 in the "self" and the "other" beenenhancedbythefactthatsubjectsindeed wereconstrained by theavailableinformation in those terms.Comassess to the partner 10Alternatively, one could argue that the subjects and oral) volunteered by but ments(bothwritten believed the task to be masculine,as instructed, becausethiswouldhave several subjectsrenderinformalsupportto it to be gender-neutral reported This this view. In futureresearch,it would be enhancedthemin the eyes of the experimenter. argumentis highlyspeculative,however,because in relativelysimple to add questionson the doingso, thesubjectsalso wouldhavebeencontradicting to impress. factorsthat subjectsconsideredin making ofthepersontheyweretrying theinstructions Moreover,I found no indicationsthat any subjects theirassessments. whether concerned theyaccepted believedtheexperiment Finally,the followingimplicationof the thestatedsex linkage.Finally,resultsfromtherestofthe resultsis worth noting.Findingson perceived manipulationchecks do not supportthe view that shown on the top fourrows of competence was paramountin the impressingthe experimenter 3 those Table condition reflect, percondition, to I reason not accept subjects'minds.Therefore see no obtainedon standards.When standardsare theirresponseson perceivedsex linkage. DOUBLE STANDARDS 247 (as in of theseprinciples-thatis, it shouldmake clearlyby sex of referent differentiated the "other" conditions),thereis a corre- themthinkmoreaboutwhattheysay and do. in the two measuresof There is some evidence fromexpectation spondingdifference competence.When the differencebetween states researchitself that such awareness standardsdecreases(as in the "self" condi- decreasesstatuseffects.ThusMartinand Sell tions), the perceivedcompetencemeasures (1983) createdeitherhigheror lowerperforforselfthanforotherin a closer. Thus both manceexpectations become correspondingly setsof findingssupportthedoublestandards settingthateitherincludedor did notinclude stimuli(mirrors, self-focusing camera,televiformulation. sion monitor).Resultsindicatethatsubjects in the formergroupsshowedthe effectsof statusinformation less thanthosein thelatter. 2 EXPERIMENT Outsidethestatuscharacteristics literature, OBJECTIVES severalstudieshave examinedtheeffectsof A large number of studies on status accountability on information processing. The and expectationstates have resultsmostdirectly relevantto theresearch characteristics as reported hereshowthatsubjectsgive greater usedthestandardized experimental setting, when they I do here. This has served to maintain thoughtto receivedinformation across studies and thus has expectto be accountablefortheirresponses comparability increasedthecumulativeness of theresearch. thanwhentheydo not(Tetlock1983, 1985; of mostof thosestudies Tetlockand Kim 1987). Thus,thesefindings One commonfeature has been that(as in Experiment1) subjects revealthataccountablesubjectsconsiderthe fromseeingeach other. different have beenprevented aspectsof an issue morecarefully of thando nonaccountable Thishas servedto controlfortheformation subjects,pay more to inconsistent itemsof information, expectationson the basis of characteristicsattention In and attempt to integrate manipulated. otherthanthoseintentionally them.Accountability addition,subjectseitherhave been led to also has been foundto increasethe use of thatthey strategies intended to createfavorableimpresbelieveor have been toldexplicitly with sionsin others.(See, forexample,Millerand would have no face-to-face interaction the partner.Such featuresof the situation Schlenker1985 for a study showingthat in makethe subjectless accountableforhis or participants' attributions areless egotistical of selfand other;thisfactor, publicthanin privateconditions.)In expectaherassessments in turn,should fosterstatusgeneralization. tion statesresearch,both of thesetypesof By "accountability"I mean the extentto processes(namelycognitiveand impression intoa decrease whichselfanticipates shouldtranslate havingto justifyhis or management) heractions.To myknowledge,variationsin in the use of stereotypicalresponses as increases. thisfactorhave notreceivedadequateatten- accountability The firstobjective of this study is to statesresearch.In mostof tionin expectation has beentreatedas a incorporate"low accountabilityfor one's thiswork,accountability variablethatis usefulto keep constant-a assessments"as a scope conditionof the of doublestandards. The decisionwhichis based on designratherthan theoryon activation ofthe Foschi et al. studythusservesas a partialreplication on theoreticalconsiderations. by retestingHypothesis1 (1994) and Foschi et al. (1995) discuss it firstexperiment as a factorin statusgeneralization, with the explicit addition of that scope explicitly but again, both studies keep the variable condition.I focus on those groups from a relatedtopic, is Experiment1 in which a double standard constant.Self-awareness, by Martinand Sell (1983). investigated emergedmost clearly,namelythe "other" In general,expectation statestheorymakes groups. In this way Experiment2 also of thatfinding. aboutself's awarenessof the assessesthereliability no assumptions principlesused in processingstatusinforma- The second objective is to explore the tion-that is, the theoryproposesthattheir limitsof thatscope condition by investigating operationmay or may not be conscious whetherit can be extendedto include a is not (Bergeret al. 1985:37). This positionalso mediumlevel. (Note thattheintention extendsto theuse ofdoublestandards (Foschi to studythe effectsof clearly contrasting on the values of this variable,such as "low" and 1989:63)). Increasedaccountability, otherhand,shouldmakepeople moreaware "high" levels,forexample.)The increasein SOCIAL PSYCHOLOGY QUARTERLY 248 shouldnarrowthegap between necessaryto replicatethe controlgroupsof accountability appliedto the Experiment1. An overviewof the design thestandard thetwostandards: shoulddecrease,whilethe appearsin Table 4. femaleperformer should standardset for the male performer of relativecompetence increase.Perceptions Proceduresand Materials shouldreflectthischangein standards. In this studyI thus examinewhethera The onlydifferences in theprocedures used the in thetwo experiments eliminates mediumlevelof accountability concerntheaccountor merelyreducesitsmagni- abilitymanipulation. doublestandard Thus, in Experiment 2, tude. Results also would serve as indirect some paragraphswere incorporated intothe evidenceof theextentto whichthe subjects instructionsas follows. Subjects in the as legitimate low-accountability treattheuse ofa doublestandard groups were instructed That is, medium "Forthesakeofconfidentiality, underthe circumstances. pleasedo not should not be sufficientto writeyourname on theseforms"and were accountability deterthe activationof double standardsif told "You can be assured that no one, these are deemed to be a proper (i.e., includingyourpartner,will be seeingyour culturallyacceptable) way of processing responsesto the questionnaires,with the (For analyses of the role of exceptionoftheresearchteam.In addition,at information. processes in the formationof no timewill you be requiredto justifyyour legitimation see RidgewayandBerger1986; responsesto these questionnairesor your expectations, Ridgewayand Walker1995.) choiceson thetasks,norwillyoube meeting withyourpartnerduringthe course of the study."In the medium-accountability condiMETHOD tions,theinstructions stated"At theconclusion of the studyyou will be meetingyour Subjectsand Experimenters to compareand discussyouranswers partner and on thetwo tasks Subjects were 48 men and 48 women. on thesequestionnaires And please rememAverageages and standarddeviationswere y?ouwill be performing. 18.65 [1.14] forthemenand 18.81 [1.23] for ber to writeyourfull name legiblyon the thewomen.The subjectpool was thesameas frontpage of each questionnaire."(The 1 impliedthat 2 was instructions thatusedforExperiment usedinExperiment 1; Experiment no interaction withthe partnerwould take aftertheearlierstudy. conductedshortly was teamedwithanother place, thusmakingall theconditionsof that Each participant of theoppositesex, and teamswereassigned studyhighlycomparableto thelow-accountNote at randomto one of two conditions,either abilitygroupsofthepresentexperiment. The study also that, as stated earlier, the medium low or mediumaccountability. conditionswere not designed was a 2 (sex of subjectand partner) accountability therefore x 2 (level of accountability) design,with24 to representa maximumincrease in this subjectspercell. Each sessionwas conducted factor.That could have been achieved,for in the same way as in the previousexperi- example, by tellingthe subjectsthat they ment.For thisreasonand because the same wereexpectedtojustifyall oftheiranswersto subjectpool was used, I did not considerit both the partnerand the researchteam.) 2 Table 4. Overviewof Experiment Part2 Dependent VariablesMeasured Part1 Phase Manipulation Scores Receivedby Condition (1) (2) (3) (4) Sex of Subject M F M F Sex of Partner F M F M Subject Partner (Maximum:20) 11 11 11 11 13 13 13 13 Standards forAbility in HigherLevel of Scoring Person Accountability Low Low Medium Medium Yes Yes Yes Yes Perceived Competence in Subject and in Partner Yes Yes Yes Yes DOUBLE STANDARDS 249 Manipulationchecks on this variablewere all subjectsrecalled exactlythe scores reand the ceived by themselvesand by theirpartner, includedin thesecondquestionnaire interview. thesex of thelatter. and correctly identified Finally,one 5-pointbipolarscale servedas a manipulationcheck of accountability, asRESULTS sessed in termsof perceivedlevel of privacy obtained of the situation.This itemread as follows: An analysisof the information instruments"Consideringwho will have access to your throughthe postexperimental resultedin theexclusionof 11 subjects(five responses(e.g., partner,researchassistant, men and six women),or 11.5% of thetotal otherresearchers),rate how private(1)numberof participants. The same rejection public (5) you perceivedtheseresponsesto rules as in Experiment1 were applied,and be." Means and standarddeviationsforthe theresulting exclusionrateis similarto that low-accountability groupswere as follows: oftheearlierstudy.The reasonsforexclusion Condition(1) (male subjects):2.17 [1.10]; were as follows:suspicion(threesubjects), Condition(2) (femalesubjects):1.98 [0.76]. (foursub- Values forthemedium-accountability misunderstanding the instructions groups and/or were: Condition(3) (male subjects): 2.88 jects), and lack of task orientation collectiveorientation(five subjects). (One [1.20]; Condition(4) (femalesubjects):3.13 person was classified in two of these [1.14]. As expected,ANOVA resultsshowa maineffectfromthismanipulation categories.)The followinganalysisincludes significant = 2.08, M (medium (M (low accountability) onlythe85 retainedsubjects. accountability)= 3.00, F (1, 81) = 24.46, p = .000), whereasneitherthe main effect Manipulation Checks fromsex of subject nor the interaction is As in Experiment 1, thepostexperimentalsignificant.Simple contrastsindicate that about the accountability level madea significant differquestionnaire yieldedinformation of thetask. Means and ence formen (F (1, 81) = 7.21, p = .009) as subjects'perceptions standarddeviationson these itemswere as well as forwomen(F (1,81) = 18.50, p = follows: creative (1)-routine (5): 2.85 .000). In other words, subjects in the [1.15]; important (1)-unimportant(5): 2.87 low-accountability groupsperceivedthecon[1.06]; easy (1)-difficult (5): 3.51 [1.18]; text of the studyto be significantly more intuitive(1)-learned (5): 2.31 [1.13]; and private than did those in the mediummasculine(1)-feminine (5): 2.85 [0.76]. accountability groups. Subjects' dispositionstowardthe task were also assessedas in theearlierstudy.Averages DependentVariables and standarddeviationson these measureAs in Study1, I beginby examiningthe mentswere:interested (1)-uninterested(5): 2.40 [1.10]; motivated (1)-unmotivated(5): resultson standardsfor the higher-scoring 2.45 [1.06]; and involved(1)-uninvolved person, shown in Table 5. As expected, main effect (5): 2.23 [1.00]. As expected,ANOVA for ANOVA indicatesa significant of standards(M (male each of these eight variables showed no fromsex of referent statisticallysignificanteffectsfrom either referent)= 66.77, M (femalereferent)= In 71.98, F (1, 81) = 7.19, p = .009), whereas typeof dyad or level of accountability. 1 in neitheraccountability nor the interaction is addition,a comparisonwithExperiment Sex of referent theserespectsshows a close overall corre- significant. had a significant was low (F spondencebetweenthetwo sets of findings. effectonly when accountability (The only noticeabledifferenceoccurs on (1, 81) = 5.16, p = .026). in selfandinpartner Perceivedcompetence perceiveddifficulty; subjectsin this study thanthosein was measuredprimarily assessedthetaskas less difficult through rejectionof Experiment1. I do not attachany special influence.As Table 6 shows, men rejected to thisdifference, however,given more influencefromthe partnerthan did importance betweenthe women.ANOVA indicates,as expected,that (1) the highlevel of similarity two studiesin all otherrespectsand (2) the the main effectfromsex of subject was factthatin bothexperiments, subjectsnone- significant(M (male subject) = .543, M thelessperceivedthetaskto be difficult.) (female subject) = .452, F (1,81)) = 9.68, p resultsalso showthat = .003), whereasthe effectsfromeither The postexperimental SOCIAL PSYCHOLOGY QUARTERLY 250 Person(in Percentages) 2: StandardsforAbilityin theHigher-Scoring Table 5. Experiment Sex of Referent of Standards Level of Accountability Condition (as perTable 4) N (1) Male subject 21 F Low 72.14 9.02 21 M Low 65.86 10.58 22 F Medium 71.82 6.08 21 M Medium 67.67 9.67 M SD scoring worse than femalepartner (2) Femalesubject scoring worse than malepartner (3) Male subject scoring worse than femalepartner (4) Femalesubject scoring worse than malepartner were not. = 4.40, p =. 039. Simple contrastsreveal or the interaction accountability Again, internalanalyses show that sex of borderlinesignificancefromsex of subject in (Conditions(1) and (2): F (1, 81) = 2.97, p difference subjectresultedin a significant only the low-accountabilitygroups (F = .089; Conditions(3) and (4): F (1, 81) = (1, 81) = 8.66, p = .004). Perceptions of 2.77, p = .10) and nonsignificant effects relativeabilityin self and in otherwere fromaccountability. In sum, all manipulation checks (except, assessedas well; as Table 6 also shows,these varied by sex of subject and level of again,forsex linkageoftask)areas expected. ANOVA indicatesa signifi- So aretheresultson standards andrejection accountability. of thewiderthegap in standards, the cantmaineffectfrombothof thesevariables, influence: ininfluence thedifference while the interactionwas not significant. laqger rejection. MoreResults for sex of subject were: M over,data fromthelow-accountability condi(malesubject)= 2.39, M (femalesubject)= tionscloselyreplicatethosefromConditions 2.12, F (1, 81) = 5.73, p = .019. For the (1) and(2) ofStudy1, as anticipated. Findings secondfactortheanalysisshowsthefollow- on perceivedrelativeabilityincludesomeuning: M (low accountability)= 2.38, M expectedresults,and theseare discussedbe= 2.14, F (1, 81) low. (mediumaccountability) 2: PerceivedCompetencein Self and in Partner Table 6. Experiment Condition (as perTable 4) Self's Ability Relativeto Partner's Rejectionof Influence fromPartner SD N M SD M (1) Male subject scoringworsethan low femalepartner, accountability 21 .552 .156 2.52 .602 (2) Femalesubject scoringworsethan malepartner, low accountability 21 .429 .146 2.24 .436 (3) Male subject scoringworsethan medium femalepartner, accountability 22 .534 .108 2.27 .550 (4) Femalesubject scoringworsethan malepartner, medium accountability 21 .474 .131 2.00 5.48 DOUBLE STANDARDS DISCUSSION 251 based double standardsin assessingcompefromthisexperiment are tence. The theoreticalideas both elaborate Two setsoffindings indicate thatthe and extendaspectsof the expectationstates First,results ofspecialinterest. whiletheresultsrepresent experdouble standard,althoughstill formulation, gender-based of the existenceand whenaccountabilityimentaldemonstration losesitsmagnitude present, a significant sex-of- directionof such double standardsand their is increased. Thus,although as well consequences. Thus, overall, subjects in subjectmaineffect is foundin standards standard in bothcases the opposite-sexdyadsreporta stricter as in rejectionof influence, for a femalethanfor a male partner,even is tracedto thelow-accountability significance at the same that thoughbothpersonsperformed thusindicate conditions only.Thesefindings level and the difference in scores between self areactivated and doublestandards gender-based was also The and other equal. gap between that accountabilat levelof usedmostdefinitely in is a the standardsis reflectedin differences ity.It also maybe thatlow accountability on perceived competence, and findings these of theother fortheoperation scope condition inexpectation studied statesthe- two variablesshow a high level of consisstatus processes in thegap in tency both within and across the two thereduction ory.Furthermore, inaccountabilityexperiments. an increase standards following Consideredtogether,the studiesidentify as an ofusinggender suggests thatthelegitimacy two conditionsunder which the double is somewhat limited. indication of competence inviewofthema- standardis most pronounced:(1) when the Thisshouldnotbe surprising is an "other"who differsfromself jor changesin thestatusvalueof genderdiffer- referent with only respectto gender,and (2) when intheUS andCanadain encesthathaveoccurred feel that accountabilityfor their subjects recentyears. is assessments low. The double standard Second, resultson relativeability,on the even the subjects report emerges though from other hand, are somewhatdifferent the that (against experimenter's instructions) thoseon rejectionof influence:whereasthe It lower whenaccount- theyconsiderthetaskto be gender-neutral. formeris significantly is also worth that noting the gender informaabilityincreases,a comparablefindingis jpot observedin rejectionof influence.In my tion was presentedwith minimal cues: between subjectsknew thatthe partnerwas of the view, this is relatedto differences this thetwomeasures.In linewithotherexpecta- oppositesex, but no itemsreinforcing as communication hobbies and first (such tion statesresearch,I considerrejectionof influenceto be the primaryindicatorof names)wereadded. The design of these studies included a perceivedcompetence.The measurehas the version of the experimental computerized advantagesof being both an unobtrusive used in expectationstates setting commonly behavioral response (and thereforemore As a research. result, findingsfrom the likelyto reflecta person'strueassessments) can easily be compared and highlyreliable(becauseit is theresultof presentexperiments made over severaltrials).Per- withotherworkon expectationstates,as I observations maineffectfromaccount- have done throughoutthis article. The haps a significant abilitywas observedin relativeabilitybutnot questionaboutthe standardsforcompetence in influence rejectionbecauseof thesubjects' that I used representsan additionto the awareness,in Conditions(3) and questions commonlyasked in that work. heightened by requestthe Standardsweremeasureddirectly (4), of whatwas beingassessedthrough to state the ofcorrect ing subjects percentages measure.As a result,theymayhave auxiliary a that for definite responses they required consideredit desirableto lower the values was inference of The underability. question in theyreported.In view of thisdifference, also stood the and readily by participants thenextsectionI base myconclusionsabout to be an not unobtrusive instrument: proved on the results perceivedcompetencemainly of who even those were any subsequently forrejectionof influence. rejectedbecauseof suspicionwereawarethat thisquestionwas centralto thestudy. GENERAL DISCUSSION AND These two experiments investigateoppoCONCLUSIONS site-sexdyads. Subjectswere informed spewas of theopposite In thisarticleI presenttheresultsfromtwo cificallythatthepartner theyeitherwere studieson the activationand use of gender- sex, whereason otherfactors 252 SOCIAL PSYCHOLOGY QUARTERLY intheamountandintensity told thattheywere equal to the partneror genderdifferences These features of of work-related stressthey experience.In weregivenno information. the situationcreateoptimumconditionsfor futureresearch, it would be useful to linksbetweeninterpersonal double genderto become a salientvariable.In the investigate of sex standardsand processes conceptualizedat investigation future,a morethorough effectscould be achievedby studyingcon- otherlevels of analysis,fromindividualto and macrostructural. of organizational textsin whichselfis pairedwitha partner double standardsis also eitherthe same or the oppositesex. Useful Understanding information also could be obtained from important forreasonsotherthantheirtheoretinwhichsex ofpartner is unknown ical significance.For example, suppose it conditions thatmembers ofa because these conditionswould serve to becomeswidelyrecognized address questionsabout the source of the social categoryare, or have been,commonly double standard:do men set strictstandards disadvantagedin settingsinvolvingevaluathrough theuse of a stricter forwomen,do womenset lenientstandards tion,particularly abilitystandard.How can this situationbe formen,or do bothpracticesoccur? on reversed?This is exactlythe questionthat These studiesprovidekey information actionprogramsaddress.Simply the operationof double standards.Thereis, affirmative an explicitlymore lenientstanhowever,stillmuchworkto do to understand introducing this practice. For example, it would be dardforone categoryin orderto makeup for worthwhileto study its occurrenceunder past wrongs is not the answer. Such an of lesser othervalues of sex linkageof task(particu- approachwouldlead to an inference larlythe"feminine"task)and otherlevelsof abilityamongmembers of thatcategory;this, It would also be of interest to in turn, would perpetuatethe inequality. performance. action programs test whethergender-baseddouble standards Well-designedaffirmative forlack of abilityconstitute anotherstatus- arenotbasedon theimplementation ofsucha maintenance practice.Finally,I shouldem- "reverse double standard." Instead they phasize that the theorybehind this work includedemonstrations ofabilitythatleaveno concernsnot only gender but any status doibts about the superiorquality of the to test chosenapplicants. attribute. Thus it wouldbe important thetheoryin othercases, involvinga single statuscharacteristic as well as several.After REFERENCES situations it is common all, in manyeveryday to be affectedby more than one status Berger,Joseph,M. HamitFisek,RobertZ. Norman,and and variable. Women, for example, experience MorrisZelditchJr. 1977. StatusCharacteristics lowerstatusand itsconsequencesto different Social Interaction.An ExpectationStatesApproach. New York:Elsevier. on thecombinedeffects of Berger, degreesdepending Joseph,David G. Wagner,and MorrisZelditch especiallyethnic- Jr.1985. "Introduction: genderand otherattributes, ExpectationStatesTheoryity, age, social class, and level of formal ReviewandAssessment."Pp.1-72 in Status,Rewards and Influence:How Expectations OrganizeBehavior, education. editedby JosephBergerand MorrisZelditchJr.San HereI haveexaminedaspectsof theuse of Francisco:Jossey-Bass. gender-baseddouble standardsfor compe- Biernat,Monica and Melvin Manis. 1994. "Shifting level. As discussed Standardsand Stereotype-Based tenceat theinterpersonal Judgments." Journal earlier,competencestandardsare basically ofPersonalityand Social Psychology66:5-20. imposedon a person'sperfor- Biernat,Monica,MelvinManis,andThomasE. Nelson. requirements 1991. "Stereotypesand Standardsof Judgment." mance.Whendifferent standardsare applied JournalofPersonalityand Social Psychology 60:485of to different categories people,thatpractice 99. to themaintenance of theinterac- Blalock, HubertM. 1979. Black-White contributes Relationsin the tional statusquo. It would be surprising, 1980's: Toward a Long-TermPolicy. New York: Praeger. however,iftheconsequencesof thispractice Deaux, Kay. 1985. "Sex andGender."AnnualReviewof were limitedto its directeffecton interper- Psychology36:49-81. sonal assignmentof competence.I suspect Deschamps,Jean-Claude.1983. "Social Attribution." Pp. 223-40 in Attribution Theoryand Research: thatdouble standardsalso are at therootof Conceptual,Developmentaland Social Dimensions, several other importantphenomena.Two editedby JosJaspars,FrankD. Fincham,and Miles in requirements Hewstone. ofdifferences possibleeffects London:AcademicPress. are the gap in earningsbetweenmen and Epstein, CynthiaF. 1970. "Encounteringthe Male women with equal qualifications,and the Establishment:Sex-Status Limits on Women's DOUBLE STANDARDS Careers in the Professions."AmericanJournalof Sociology75:965-82. Foddy,Margaretand Michael Smithson.1989. "Fuzzy Sets and Double Standards:Modelingthe Processof AbilityInference."Pp. 73-99 in SociologicalTheories in Progress: New Formulations,edited by Joseph Berger, Morris Zelditch Jr., and Bo Anderson. NewburyPark,CA: Sage. StanFoschi, Martha. 1989. "Status Characteristics, dards,and Attributions." Pp. 58-72 in Sociological Theoriesin Progress:New Formulations,editedby JosephBerger,MorrisZelditchJr.,and Bo Anderson. NewburyPark,CA: Sage. . 1992. "Gender and Double Standards for 253 Monson,ThomasC. and MarkSnyder.1977. "Actors, Observers,and the Attribution Process: Towards a Reconceptualization." JournalofExperimental Social Psychology13:89-111. Nieva, VeronicaF. and BarbaraA. Gutek.1980. "Sex Effectson Evaluation." Academyof Management Review5:267-76. Pugh,Meredith D. andRalphWahrman.1983. "NeutralizingSexismin Mixed-SexGroups:Do WomenHave to Be Better Than Men?" AmericanJournal of Sociology88:746-62. Ridgeway,Cecilia L. 1993. "Gender,Status,and the Social Psychologyof Expectations."Pp. 175-97 in Theoryon Gender/Feminism on Theory,edited by Paula England.New York:Aldine. Ridgeway,Cecilia L. and JosephBerger.1986. "Expectations,Legitimation,and Dominance Behavior in Task Groups."AmericanSociologicalReview51:60317. Ridgeway,Cecilia L. and Henry A. Walker. 1995. "Status Structures."Pp. 281-310 in Sociological on Social Psychology, editedby KarenS. Perspectives Cook, GaryA. Fine, and JamesS. House. Boston: Allynand Bacon. Ross, Lee. 1977. "The IntuitivePsychologistand His in theAttribution Distortions Process." Shortcomings: Pp. 173-220 in Advances in ExperimentalSocial Psychology,Vol. 10, editedby LeonardBerkowitz. New York:AcademicPress. Ross, Michaeland GarthJ.O. Fletcher.1985. "Attributionand Social Perception."Pp. 73-122 in Handbook of Social Psychology,3rd ed., Vol. 2, edited by GardnerLindzey and Elliot Aronson. New York: RandomHouse. andComplexity Tetlock,PhilipE. 1983. "Accountability of Thought." Journal of Personalityand Social 45:74-83. Psychology " Pp. 181-207 in Gender,Interaction, Competence. and Inequality,editedby Cecilia L. Ridgeway.New York:Springer-Verlag. Foschi,Marthaand MargaretFoddy. 1988. "Standards, Performances,and the Formationof Self-Other Expectations."Pp. 248-60 and 501-503 in Status New Theoryand Research,editedby Generalization: MurrayWebsterJr. and MarthaFoschi. Stanford: Stanford Press. University Foschi, Marthaand SabrinaFreeman.1991. "Inferior Performance, Standards,and Influencein Same-Sex Dyads." Canadian Journalof BehaviouralScience 23:99-113. Foschi,Martha,LarissaLai, and KirstenSigerson.1994. "Genderand Double Standardsin theAssessmentof JobApplicants."Social Psychology Quarterly 57:32639. Foschi,Martha,KirstenSigerson,and MarieLembesis. The RelativeEffects 1995. "AssessingJobApplicants: of Gender,AcademicRecord,and Decision Type.') SmallGroupResearch26:328-52. Foschi, Martha,G. Keith Warriner,and StephenD. Hart. 1985. "Standards,Expectations,and Interpersonal Influence."Social PsychologyQuarterly48: . 1985. "Accountability: A Social Check on the 108-17. FundamentalAttribution Error." Social Psychology Quarterly 48:227-36. Hansen, Ranald D. and VirginiaE. O'Leary. 1985. Attributions." "Sex-Determined Pp. 67-99 in Women, Tetlock,PhilipE. andJaeI. Kim. 1987. "Accountability editedby VirginiaE. and Judgment Processesin a PersonalityPrediction Gender,and Social Psychology, Task." Journalof Personality and Social Psychology O'Leary,Rhoda K. Unger,and BarbaraS. Wallston. 52:700-709. Hillsdale,NJ:Erlbaum. in Groups: Jasso, Guillerminaand Murray Webster Jr. 1995. Wagner,David G. 1988. "GenderInequalities A SituationalApproach."Pp. 55-68 and 480-84 in "Double Standardsin JustEarningsfor Male and FemaleWorkers."Presentedat the32ndCongressof Status Generalization:New Theoryand Research, theInternational Institute of Sociology,Trieste. edited by MurrayWebsterJr. and MarthaFoschi. Stanford: Stanford Press. Jones,EdwardE. and RichardE. Nisbett.1972. "The University Actorand theObserver:DivergentPerceptions of the Wagner,David G. and JosephBerger. 1993. "Status Causes of Behavior." Pp. 79-94 in Attribution: Characteristics Theory:The Growthof a Program." PerceivingtheCauses ofBehavior,editedby Edward Pp. 23-63 and 454-63 in TheoreticalResearch E. Jones, David E. Kanouse, Harold H. Kelley, Programs:Studiesin theGrowthof Theory,editedby Richard E. Nisbett, Stuart Valins, and Bernard Joseph Berger and Morris Zelditch Jr. Stanford: Weiner.Morristown, NJ:GeneralLearningPress. Stanford Press. University Kalin,RudolfandDavid C. Hodgins.1984. "Sex Bias in Wagner,David G., Rebecca S. Ford, and ThomasW. of OccupationalSuitability."Canadian Ford. 1986. "Can GenderInequalitiesBe Reduced?" Judgements JournalofBehaviouralScience 16: 311-25. AmericanSociologicalReview51:47-61. Lott, Bernice. 1985. "The Devaluationof Women's Wallston,BarbaraS. and VirginiaE. O'Leary. 1981. "Sex Makes a Difference: Differential of Competence."JournalofSocial Issues 41:43-60. Perceptions Martin,Michaeland JaneSell. 1983. "Self-Awareness, Womenand Men." Pp. 9-41 in ReviewofPersonality InformationUtilization,and Social Influence in and Social Psychology,Vol. 2, edited by Ladd Wheeler.BeverlyHills: Sage. CooperativeTask Settings."Sociological Focus 16: 147-53. Webster,MurrayJr.and MarthaFoschi. 1988. "Overview of StatusGeneralization." Miller, Rowland S. and Barry R. Schlenker.1985. Pp. 1-20 and 477-78 in StatusGeneralization: New Theoryand Research, "Egostismin Group Members:Public and Private of Responsibilityfor Group PerforAttributions edited by MurrayWebsterJr. and MarthaFoschi. mance." Social Psychology 48:85-89. Stanford: Stanford Press. University Quarterly 254 SOCIAL PSYCHOLOGY QUARTERLY by JanetS. Hyde and Marcia C. Linn. Baltimore: Whitley,BernardE., Jr., MaureenC. McHugh, and JohnsHopkinsUniversity Press. Irene H. Frieze. 1986. "Assessing the Theoretical of Wiley, Mary G. 1986. "How ExpectationStates in Causal Attributions Models forSex Differences OrganizeTheoryConstruction." SociContemporary Success and Failure." Pp. 102-35 in The Psychology edited ology15:338-41. of Gender:AdvancesthroughMeta-Analyses, ofBritishColumbia.Her researchinterests MarthaFoschi is Professorof Sociologyat the University as sourcesofbiases in abilityevaluation. on genderand ethnicity focuson statusprocesses,particularly she has recentlycarriedout on The two studiesreportedhere are part of a series of experiments gender-based doublestandards for competence.