the role of awareness: divergent automatic stereotype ...faculty.weber.edu/eamsel/research...

Le por e an d Bro wnDIVERGENT ST EREO TYPE ACTIVAT ION AND CO RREC TION

THE ROLE OF AWARENESS DIVERGENTAUTOMATIC STEREOTYPE ACTIVATION ANDIMPLICIT JUDGMENT CORRECTION

Lorella LeporeUniversity of Reading United Kingdom

Rupert BrownUniversity of Kent Canterbury UK

Two experiments investigated automatic stereotype activation and its correctionIn Experiment 1 (n = 57) the category ldquoBlacksrdquowas primed using a lexical decisiontask People high and low in prejudice subsequently formed divergent impressionsof the target person Replicating previous findings (Lepore amp Brown 1997)highndashprejudice participantsrsquo judgments were more negative and lowndashprejudicepeoplersquos ratings more positive Awareness of a connection between priming andimpression formation reversed this pattern Experiment 2 (n = 40) revealed thatawareness of a connection irrespectiveof priming recall promoted a correctionofthe judgmentUnaware highndash and lowndashprejudice participants again showed diver-gent automatic stereotype activation but aware respondents corrected their judg-ments in opposite directions Thus when automatic stereotype activation isdifferentiated implicit correction processes act upon different accessible knowl-edge resulting in divergent judgment corrections Implications for stereotypingand models of contrast effects are discussed

Research on the automaticity of stereotype activation has grown rapidlyin recent years At the basis of this line of investigation lie questions rele-

321

Social Cognition Vol 20 No 4 2002 pp 321-351

This research was supported by ESRC (Economic and Social Research Council of GreatBritain) Grant R000234415awarded to both authors The research reported in this articlewas conducted while Lorella Lepore was a research fellow at the University of Kent at Can-terbury UK

We thank Lisa Lawrence Angelika Schipper and Nils Wandersleben for their assistancewith the data collection We would also like to thank Judi Ellis and Clare Harries for theirhelpful feedback and suggestions and the anonymous reviewers who provided many in-sightful comments on a previous version of the manuscript

Address correspondence concerning this article to Lorella Lepore Department of Psy-chology University of Reading Earley Gate Reading RG6 6AL UK Endashmaillleporereadingacuk

vant to everyday social encounters What are the consequences of catego-rization What are peoplersquos automatic responses to members ofstereotyped groups Theories of modern (McConahay 1986) aversive(Gaertner amp Dovidio 1986) and subtle racism (Pettigrew amp Meertens1995)all statethatprejudice lingers despite its reduction in expressed atti-tudes and stereotype content arising from changed norms of social inter-actionBased on these conceptualizations it is often surmised that explicitresponses to category members will appear to be nonprejudiced whereasthe implicit and automatically accessed responses will still be prejudicedSuch incongruency between implicit and explicit attitudes is implied forexample in Devinersquos (1989)seminal work According to her model nega-tive stereotypes are activated automatically upon categorization regard-less of peoplersquos prejudice level Their use however can be consciouslycontrolled if motivation and cognitive capacity allow for the suppressionof the automatic prejudiced response Other studies support theincongruency of implicit and explicit attitudes Banaji and Greenwald(1995) for example found that implicit gender bias in judgments of fameand explicit gender measures were not correlated (see also Banaji ampHardin 1996 for a review see Blair 2001) Such findings suggest that ste-reotypes are activated automatically in an allndashorndashnone manner and re-gardless of peoplersquos beliefs about the target group Deriving from thepremise that stereotypes are activated automatically upon categorization(Allport 1954 Bargh 1999 Devine 1989Greenwald amp Banaji 1995)is re-search investigating the effects of stereotype suppression (eg Monteith1993 Wyer Sherman amp Stroessner 2000) The conscious suppression ofprejudiced responses elicited by the automatic and seemingly inevitableactivation of negative stereotypes should curb unwanted stereotyping injudgments and behaviors (but cf Macrae Bodenhausen Milne amp Jetten1994) in a process of selfndashregulation eventually leading to the automatiza-tion of such suppression itself (see Monteith Sherman amp Devine 1998andMonteith amp Voils 2001 for reviews)

At present evidence regarding the congruency or incongruency of im-plicit and explicit attitudes is contradictory and not yet reconciled (seeBlair 2001) In contrastwith the findings reported above several studieshave observed automatic facilitation effects to be correlated with scoreson a prejudice measure (Lepore amp Brown 1997 Locke MacLeod ampWalker 1994 Wittenbrink Judd amp Park 1997 See also Fazio JacksonDunton amp Williams 1995 Kawakami Dion amp Dovidio 1998) Investi-gating the relation between categorizationand stereotyping Lepore and

322 LEPORE AND BROWN

Brown (1997) found that automatic stereotype activation is consistentwith beliefs about the group and happens differentially for highndash andlowndashprejudice people negative stereotypes were activated automati-cally only by highly prejudiced individuals

The present experiments center on this interplay of automatic andcontrolled processes Their purpose was twofold to assess thereplicability of the differentiated automatic activation of stereotypes ob-served in people high and low in prejudice and to examine whether itobtains when responses to the prime are relatively more controlled

GROUP REPRESENTATIONS STEREOTYPE ENDORSEMENTAND STEREOTYPE ACTIVATION

Lepore and Brown (1997) proposed a model of representations of socialgroups according to which peoplersquos beliefs about the group affect howeasily specific associations come to mind The authors argued thatwithin the complex knowledge base about a group in memory the ste-reotypical content is similarly available to people who are or are not prej-udiced against that group However as a result of repeated activationand use of endorsed stereotypic characteristics highndash and lowndashpreju-dice people may hold group representations in which different aspectsof the same stereotype are more strongly associatedwith the group nodeand thus more easily activated The findings supported this model De-spite their common stereotype knowledge highndash and lowndashprejudiceparticipants formed divergent impressions of a target person as a conse-quence of category activation (Lepore amp Brown 1997) Specificallyhighndashprejudice participants activated negative stereotypic aspects andlowndashprejudice respondents showed activation of the positive character-istics after subliminal priming of the category ldquoBlacksrdquoreflecting a dif-ferent pattern of stereotype endorsement Thus stereotype activation asa consequence of categorization does not happen in an allndashorndashnone fash-ion but differentially with lowndashprejudice people demonstrating no ac-tivation of the negative stereotype1

DIVERGENT STEREOTYPE ACTIVATION AND CORRECTION 323

1 In a vein consistent with our model Carlston (1992)hypothesized that for highndashpreju-dice people both the categorization process and the subsequent attribution of [negative]stereotypic traits are automatic whereas for lowndashprejudice individuals only categoriza-tion happens automatically As automaticity develops through frequency of use the acti-vation of knowledge repeatedly accessed can become automatic (eg Bargh 1996Carlston 1992)

CONSCIOUS PRIMES AND CONTRAST EFFECTS

Outside the laboratory we are usually aware of the presence of othersThe use of a conscious prime in the present experiments allows to repli-cate the findings obtained with a subliminal priming procedure byLepore and Brown (1997) while mimicking more realistically the initialreaction to encountering a member of a stereotyped group According toBargh (1996) the automatic processes instigated with either a subndash orsupraliminal prime can be considered preconscious because they re-quire neither an intention that the process occurs nor an awareness of theprocess occurring

It is widely accepted that the effects of suprandash and subliminal primesare the same and automaticHowever such effects are not always identi-cal (see Bornstein amp DrsquoAgostino 1992) Most importantly for the currentpurpose when the prime is conscious there is a possibility that con-trolled processes take over that is that people become aware of the po-tential influence of the stimuli on their judgment (eg Bargh 1992 1994)The attempt to counteract such influence could give rise to contrast ef-fects2 People may try to control and correct their judgments when in thepresence of a group member or in a laboratory context when they areaware of the categorical prime The outcome of a subndash or supraliminalprime may then differ

However awareness of the prime should not lead to judgment correc-tion when it simply means that the stimuli can be consciously perceivedIf people are aware of the stimuli but unaware of their consequences theprocess is still automatic (Bargh 1992 Strack 1992) Indeed some mod-els proposed to account for contrast effects are based on the idea thatawareness of the influence of the stimulus rather than awareness of thestimulus itself is necessary to instigate judgment corrections (see Martinamp Achee 1992 Strack 1992 Wegener amp Petty 1997 Wilson amp Brekke1994)

ASSIMILATIVE AND CONTRASTIVE JUDGMENTS

Judgment of a person can be influenced by the context in which the im-pression is formed Knowledge that is accessible and applicable can be


2 Contrast effects are more likely to occur when the prime can be consciously perceivedbut the subliminal presentation of exemplars can also trigger them (Winkielman ampSchwarz 1996)

used to interpret the target personrsquos behavior (assimilation effects) par-ticularly if its influence is subtle Alternatively if accessible knowledgeis deemed unrepresentative inappropriate or biasing the impression ismoved away from it (contrast effects) (Higgins 1996 Strack 1992Wegener amp Petty 1997) The use of priming procedures to investigate ac-cessibility effects in person perception typically involves activating aconstruct (or category exemplar) in the first task which directly affectsthe impression subsequently formed (eg Higgins Rholes amp Jones1977) For example participants apply the primed construct (egldquokindrdquo) to categorize the ambiguous behavior of a target person (egLombardi Higgins amp Bargh 1987 Srull amp Wyer 1979 1980) Suchassimilative effects of the context on person perception can be reducedor reversed if for example participants are motivated to be accurate(eg Ford amp Kruglanski 1995 Thompson Roman Moskowitz Chaikenamp Bargh 1994) are reminded of or recall the priming events (egLombardi et al 1987 Newman amp Uleman 1990 Strack Schwarz BlessKuumlbler amp Waumlnke 1993) if primes are blatant (eg Martin 1986Skowronski Carlston amp Isham 1993) or if the primed exemplars are ex-treme (eg Herr 1986)

Contrast effects may be grouped in two main categories comparisonand correction (see Stapel Koomen amp Zeelenberg 1998) Comparisoncontrast effects arise when exemplars (eg Herr 1986) or personalizedtraitndashimplying sentences (eg Stapel Koomen amp van der Pligt 1996)constitute a comparison standard against which the target is perceivedThe accessible information in such cases can be more extreme or distinc-tive than the usually ambiguous target information The comparison be-tween prime and target does not necessarily require awareness(Moskowitz amp Skurnik 1999 Stapel et al 1996) However correctioncontrasteffects arise when people become aware that the primed and ac-cessible information could influence their judgment (eg Ford ampThompson 2000 Stapel et al 1998) Such information is then dis-counted as unrepresentative or deemed inappropriate and subtractedfrom the reaction to the target person (Martin 1986 Martin amp Achee1992 Schwarz amp Bless 1992 Strack 1992) In sum accessible informa-tion functions as a comparison standard in comparison contrasts and asan unwanted and contaminating influence in correction contrasts (seeStapel et al 1998p 880) Awareness of stimulus influence appears to bea key element to distinguish the two kinds of contrast effects The extentand meaning of such awareness however varies considerably in cur-


rent correction models as will be discussed later Furthermore some ev-idence suggests that correction and not just comparison contrasteffectscan be obtained through automatic processes (eg Glaser amp Banaji1999)

ASSIMILATION AND CONTRAST EFFECTS IN THIS RESEARCH

The present experiments depart from procedures often employed to in-vestigate assimilationand contrasteffects These usually involve a directlink between what is being primed (eg friendly) and the person judg-ment subsequently made (how friendly the target person will be per-ceived to be) In the present case however the connection betweenpriming and impression formation is not direct The category (ie thesocial group) is activated in the priming phase by means of category la-bels and neutral associates (eg blacks dreadlocks) Stereotypical asso-ciates (eg aggressive athletic) should become activated throughspreading activation (eg Collins amp Loftus 1975) as a consequence ofcategorization These in turn should affect ratings of the target personin the impression formation phase Awareness of the prime per se thenshould not trigger judgment corrections unless participants perceive aconnection between the priming and the impression formation phaseAs long as people are not aware of the effects that previously presentedstimuli can have on subsequent judgments they will not try to controlsuch effects and the processing will go on automatically as if the stimulihad been presented subliminally (Bargh 1992 see also Jacoby amp Kelley1987)

Because of the unusual priming procedure used here (which does notinvolve exemplars or traits) both assimilation and contrast effects haveto be defined According to Lepore and Brownrsquos (1997) model peoplehigh and low in prejudice hold different group representations and thuspriming of the category Blacks may trigger a different pattern of stereo-type activation The resulting negative (in highndashprejudice people) orpositive (in lowndashprejudice people) impression of the target person is anassimilation effect of sorts divergent impressions are formed becausedifferent content is rendered accessible by the category prime ldquoCon-trastrdquo effects are defined in relation to such differentiated stereotype ac-tivation highndashprejudice people could correct for the activated andaccessible negative content and lowndashprejudice people could correct forthe activated and accessible positive content Divergent judgment cor-


rections would result from this process However if stereotype suppres-sion takes place simply as a result of heightened prime awareness bothhighndash and lowndashprejudice people should be more positive in their judg-ments of the target

EXPERIMENT 1 PRIMING WITH A LEXICAL DECISION TASK

Experiment 1 was designed primarily to replicate previous findings ofdifferentiated stereotype activation (Lepore amp Brown 1997) The samestimulus words previously presented by means of a parafoveal sublimi-nal priming procedure (see Lepore amp Brown 1997) were embedded in alexical decision task (Meyer amp Schvanevelt 1971) together with neutralwords and nonwords Primed with category labels and neutral associ-ates of the category highndashprejudice participants were expected to evalu-ate the target person more negatively and lowndashprejudice participantsmore positively Because of the conscious nature of the prime howevercontrast effects were also possible

METHODParticipants and DesignFiftyndashseven university students white British nationals agreed to par-ticipate when approached by the experimenter on campus They werepaid pound2 The design was 2 (high vs low Prejudice) acute 2 (prime vs no primeCondition) between participants Participants were randomly allocatedto the experimental or control conditions

Materials and ProcedureThe experiment was run entirely on two Macintosh Quadra 650 comput-ers placed in separate cubicles Participants were informed that theywould take part in two unrelated experiments and that the second ex-periment was being conducted for another researcher Both the experi-menter and the instructions on the screen explained the lexical decisiontask Participants had to judge whether the briefly appearing stimuluswas a word or not Speed and accuracy were emphasized and partici-pants were told that reaction times would be recorded

Lexical Decision Task The priming phase consisted of 52 trials In theprime condition 13 words evocative of the category ldquoBlack peoplerdquo wereused (ldquoBlacksrdquo ldquoWestndashIndiansrdquo ldquoAfrondashCaribbeanrdquo ldquorastafarjanrdquoldquocul-turerdquo ldquoraprdquo ldquoafrordquo ldquoNotting Hillrdquo ldquoBrixtonrdquo ldquocoloredrdquo ldquoreggaerdquo


ldquoethnicrdquo ldquodreadlocksrdquo see Lepore amp Brown 1997) Thirteen wordswere neutral concrete and matched in length to the target words ldquoavail-abilityrdquo ldquotreerdquo ldquosuccessrdquo ldquogaprdquo ldquoorangerdquo ldquoexceptionsrdquo ldquoreorganisa-tionrdquo ldquothingsrdquo ldquoinformationrdquo ldquonumberrdquo ldquoaccommodationrdquoldquoexamplerdquo ldquosentencerdquo Twentyndashsix stimuli were pronounceablenonwords matched in length with the real words In the no prime condi-tion there were 13 neutral concrete words (the same ones used in the ex-perimental condition) 13 neutral abstract words matched in length tothe concrete ones (ldquoalwaysrdquo ldquohoweverrdquo ldquonothing elserdquo ldquoin factrdquo ldquodur-ingrdquo ldquoeverything elserdquo ldquobutrdquo ldquowhateverrdquo ldquofurthermorerdquo ldquocom-pletelyrdquo ldquothenrdquo ldquoneverthelessrdquo ldquocalledrdquo) and 26 nonwords the sameones used in the prime condition

Each word appeared on the screen for 250 ms Three neutral words un-related to the category ldquoblacksrdquo (ldquofireplacerdquo ldquomethodologyrdquo and ldquodan-gerrdquo) and three nonwords were used for the 6 practice trials A ldquoreadyrdquosignal a central dot appeared on the screen for 700 ms immediately pre-ceding each trial

Impression Formation Task The dependent measures were identicalto the ones previously used (see Lepore amp Brown 1997) In the contextof a supposedly unrelated experiment on person perception the tar-get person whose ethnicity was not specified was presented bymeans of eight behavioral sentences (eg ldquoHe plays football regu-larlyrdquo) These were descriptive of two positive and two negative ste-reotypic constructs (athletic and fun loving aggressive and unreliable)Each construct was represented by two sentences Then participantsrated the person on twentyndashone rating scales descriptive of the fourstereotypic constructs All scales ranged from 1 (not at all) to 9 (ex-tremely)

A computerndashbased administration of the Lepore and Brownrsquos(1997) prejudice scale concluded the study This 15ndashitem scale was de-veloped for use in a British context It comprises new statements aswell as statements adapted from preexisting modern and subtle rac-ism scales (eg Jacobson 1985 McConahay 1986 Pettigrew ampMeertens 1995) The scale has good internal reliability (Cronbachrsquos a= 85) In the extensive individual debriefing the experimenter askedif participants thought the tasks were related and if so how Anycomments particularly any suspicions of a link between the parts ofthe experiment were noted


RESULTS

Participants were defined as highndash and lowndashprejudice on the basis of amedian split of their scores on the prejudice scale (median = 69 SD =1237 Mhigh = 61 SD = 73 n = 29 Mlow = 82 SD = 67 n = 28) Since thescale is coded in the nonprejudiced direction higher scores indicatemore tolerant attitudes

Eleven participants reported being suspicious of a connection be-tween the lexical decision and the impression formation tasks These re-ports emerged during the funneled debriefing procedure (Bargh ampChartrand 2000) Participants were asked what they thought the studywas about whether they thought that the two ldquoseparate experimentsrdquo(priming and impression formation) were related (if so how) andwhether they thought they had been influenced by the primes Partici-pants who thought that the tasks were related were classified as ldquosuspi-ciousrdquo Naturally these were all in the prime condition ie had beenexposed to the category and categoryndashrelated primes3 A preliminaryanalysis revealed that suspicious participants responded to the impres-sion formation task quite differently from their nonsuspicious counter-parts Analyses were therefore conducted separately for these twogroups of participants As ldquosuspicionrdquo here refers to the perception of aconnection between priming and impression formationmdashand thus toawareness of a potential influence of the prime on target judg-mentmdashsuspicious participants are labeled ldquoawarerdquo and nonsuspiciousparticipants ldquounawarerdquo henceforth

Unaware and Control ParticipantsA 2 (prime vs no prime Condition) acute 2 (highndash vs lowndash Prejudice) acute2(positive negative Valence) mixed ANOVA was used to analyze theimpression formation data from the 46 participants The analysis re-vealed the predicted Prejudice acute Condition acute Valence interaction (F(142) = 410 p lt 05) As can be seen from Figure 1 lowndashprejudice partici-pants rated the target more positively on the positive scales when ex-


3 A few of the ldquosuspiciousrdquo participants raised the possibility that the person portrayedin the judgment task might be ldquoblackrdquo Participants at this stage had answered the preju-dice scale It is difficult to know when they became aware of a relation between primingand judgment and how clearly Consistent with observations often made in previous re-search (eg Martin et al 1990Moskowitz amp Skurnik 1999)participants did not think thatthe primes had influenced their judgment

posed to the categorical primes but less negatively on the negativescales The highndashprejudice participants showed a less consistent ten-dency in the opposite direction

Analysis of the simple effects revealed that the Prejudice acute Valence in-teraction was significant in the prime (F (1 43) = 910 p lt 005) but not inthe no prime Condition (F (1 43) = 19 ns) Thus highndash and lowndashpreju-dice people only differed when the category was primed In particularlowndashprejudice participants tended to higher positive ratings thanhighndashprejudice respondents (Ms on positive scales= 665 vs 713 t(42)=143 p lt 1) In contrast highndashprejudice participants evaluated the tar-get person more negatively than did lowndashprejudice participants (Ms onnegative scales= 636 vs 537 t (42) = 291 p lt 005) These results repli-cate previous findings (Lepore amp Brown 1997)

Aware vs Unaware ParticipantsTo test whether suspecting a relation between the prime and the impres-sion formation phase affected the judgment a 2(high vs low Prejudice) acute


4

45

5

55

6

65

7

75

Prime No Prime Prime No Prime

POSITIVE SCALES NEGATIVE SCALES

High-Prejudice

Low-Prejudice

FIGURE 1 Impression of the target person as a function of Prime and Va-lence of rating dimensions

2(Aware vs Unaware) acute 2(positive negative Valence) ANOVA wasconducted on the prime condition participants The analysis yielded asignificant 3ndashway Prejudice acute Awareness acute Valence interaction (F(129)=1439 plt001) which showed that ldquoawarerdquo lowndashprejudice partici-pants tended to evaluate the target person less favorably than ldquoun-awarerdquo lowndashprejudice participants (ie they gave higher ratings on thenegative scales and lower ratings on the positive scales) Amonghighndashprejudice participants awareness only reduced the negative rat-ings (see Figure 2)

Simple effects analysis revealed that the Prejudice acute Valence interactionwas significant for both the aware and unaware participants (F (1 30)=795 p lt 01 F (1 30) = 936 p lt 01 respectively)4 Thus the pattern ob-tained for the unaware participants in the previous analysis reversedwhen the respondents perceived a link between priming and impressionformation Pairwise comparisons within prejudice level were performedto investigate whether highndash and lowndashprejudice people reacted similarlywhen aware of a connection Highndashprejudice participants significantlydecreased their negative ratings when aware (as compared to unaware)(Ms = 548 vs 636 t (29)= 193p lt 05) although they were not affected byawareness on the positive ratings (Ms = 678 vs 665t(29)= 3ns) In con-trast aware compared to unaware lowndashprejudice participants signifi-cantly decreased their ratings on the positive scales (Ms = 597 vs 713t(29) = 306plt 005)and increased the ratings on the negative scales (Ms =639 vs 537 t (29) = 269 p lt01) A onendashway ANOVA was conducted onthe prejudice scores to check whether awareness was related to answerson the prejudice measure Results show that suspecting a connection be-tween priming and impression formation was not associated with preju-dice level Aware and unaware participants did not differ in their averageprejudice score (Munaware=71 Maware=76 F (1 31)=105 plt4)

DISCUSSION

Experiment 1 showed that highndash and lowndashprejudice people were differ-entially affected by a supraliminal activation of the category This repli-


4 As in Experiment 1 suspicion emerged only at debriefing (after participants had an-swered the prejudice scale) and the number of participants aware of a connection betweentasks was small the results are not completely overlapping those of Experiment 2 whichwas directly aimed at investigating awareness

cates previous findings with subliminal category activation (Lepore ampBrown 1997) Provided that participants are not aware of a relation be-tween prime and impression formation the subndash or supraliminal activa-tion of the category seems immaterial (see Bargh 19921996) As in otherstudies which used conscious primes (eg Higgins et al 1977 Srull ampWyer 1979 1980) assimilation effects were obtained but of a specialkind Highndash and lowndashprejudice people responded differently and ac-cording to their representation of the group that is the stronger associ-ates to the group node were activated Specifically highndashprejudicepeople formed a more negative impression of the target person than didlowndashprejudice people This is consistent with the model of automatic ac-tivation proposed by Lepore and Brown (1997) The links between cate-gory labels and stereotypical associates are stronger and more easilyactivated when the characteristics are endorsed The findings in this andearlier research (Lepore amp Brown 1997) are inconsistent with a view ofstereotype activation as inevitable and allndashorndashnone (eg Bargh 1999Devine 1989 Greenwald amp Banaji 1995 Monteith amp Voils 2001)


4

4 5

5

5 5

6

6 5

7

7 5

A w a re U n a w a re A w a r e U n a w a re

P O S IT IV E S CA L ES NEG A T IV E S CA L ES

H ig h -P r e ju d ic e

L o w -P re ju d ic e

FIGURE 2 Impression of the target person as a function of Awareness and Va-lence of rating dimensions

Awareness of a connection between priming and impression forma-tion generated a reversal of the judgments however The secondaryanalysis revealed that aware highndashprejudice participants formed a lessnegative impression but lowndashprejudice respondents reacted to theprime by downgrading the target person Although the number of awareparticipants was small and some caution must be exercised at this stagesuch a pattern too appears to be problematic for models of automaticallndashorndashnone stereotype activation Based on this idea and related re-search (see Monteith amp Voils 2001 for a review) when stereotypes areactivated and people become aware that their judgment may be biasedby them lowndashprejudice people should see stereotype use as inappropri-ate whereas highndashprejudice individuals might not regard such use as in-appropriate Only lowndashprejudice people would then be expected tocorrect their judgment This was not the case here Alternatively the sa-lience of norms against stereotyping and social desirability concernsmay prompt even highndashprejudice people to avoid stereotyping (egMonteith Spicer amp Tooman 1998 Exp1) In such cases both highndash andlowndashprejudice individuals should correct in the same direction ie theyshould curb the negative ratings and form a more positive and less nega-tive impression of the target Instead in the current experiment suspect-ing a relation between priming and impression formation resulted incontrast effects that were different for highndash and lowndashprejudice peopleThis constitutes preliminary support for the notion that the relative easewith which people high and low in prejudice activate positive and nega-tive stereotypic features could result not only in divergent automaticste-reotype activation but also in divergent judgment correction Peoplemay correct relative to their (different) activated knowledge

EXPERIMENT 2 THE ROLE OF AWARENESS

As Experiment 1 left open the question of why awareness of a connectionbetween priming and impression formation promoted these contrast ef-fects Experiment 2 was designed to investigate the role of awarenessmore systematically It has been argued that for judgment corrections tooccur people should be aware of the potential contaminating influenceof the context on the judgment know the direction of stimulus influenceknow how to correct and have the motivation to correct (Strack 1992Strack amp Hannover 1996 Tesser amp Martin 1996 Wegener amp Petty1997) However such full conscious awareness of the potential contami-


nating influence of the context on the judgment may not be necessaryThere is consensus that correction processes may not be open to intro-spection (Martin amp Achee 1992 Martin et al 1990 Newman DuffHedberg amp Blitstein 1996 Wegener amp Petty 1997 Wilson amp Brekke1994) Current models incorporate in different ways implicit processesin judgment correction

According to the Flexible Correction Model (Wegener amp Petty 19951997) correction processes arise when people become aware of a poten-tial bias that is they are set in motion by perceived bias Naiumlve theories(which can be verbalized) of how the context can affect the judgmentdrive corrections and determine their direction Full conscious aware-ness of stimulus influence is not always necessary as repeated experi-ences with a biasing factor may decrease the amount of consciousawareness necessary to initiate judgment correction (Wegener Petty ampDunn 1998) Besides the correction process itself may not be fully con-scious People can report a perception of bias but are not necessarilyaware of using a certain theory or able to say how they used it (WegenerDunn amp Tokusato 2001) However even when operating implicitlynaiumlve theories in this model act on the assumed direction of stimulus in-fluencemdashthey carry assumptions about how the judgment is influencedthat go beyond a mere feeling of bias (see also Moskowitz amp Skurnik1999)

The SetndashReset model of contrast effects (Martin 1986 Martin amp Achee1992 see also Martin amp Stapel 1998) also posits that correction attemptsarise when people perceive that the context (eg the primes)may contam-inate the judgment But cues in the judgment situation (eg blatantprimes lack of fit) which people may not be able to verbalize can directlyprompt a correction process if they render the default use of accessible in-formation inappropriate to interpret the targetrsquos behavior If people havethe goal of forming an impression (which involves assessing onersquos reac-tion to the target person) an implicit filtering process automatically trig-gers a correction through ldquoresettingrdquo when it detects irrelevant aspects inthe reactionto the target (eg not target related not judgment related) theirrelevant response is deleted and a more relevant one is accessed (Martinamp Achee 1992 p 205 Martin amp Stapel 1998 p 234) In subtracting the re-action to the context from the reaction to the target people may also sub-tract some of their true reactions to the target resulting in contrast effectsCorrection through resetting involves some cognitive effort to be per-formed (Martin et al 1990Moskowitz amp Skurnik 1999) but activationof


the resetting process and its effect on the judgment lie outside of aware-ness (Martin amp Achee 1992 p 205) Thus corrections in this model can beimplicit not open to introspection and not possible to verbalize The out-come of corrections however are conscious thoughts and feelings fromwhich people construct explanations for the presumed influence of thestimuli upon judgment (Martin amp Stapel 1998) Contrast effects throughcorrection processes were recently obtained in automatic evaluation withparameters that rule out the possibility of deliberate responses (Glaser ampBanaji 1999) Not just comparison then but also correction contrast ef-fects can be obtained automatically

A clear distinction has been made between awareness of the stimulusand awareness of stimulus influence (eg Bargh 1992 Strack 1992) andit is the latter that according to many authors instigates judgment correc-tion (eg Martin amp Achee 1992Strack 1992Wegener amp Petty 19951997Wilson amp Brekke 1994) However the distinction is blurred when recall ofthe priming stimuli is used as a measure of stimulus influence or to facili-tate such awareness In fact the evidence that recall of the priming epi-sodes affects the judgment by promoting contrast effects is contradictoryPriming recall affected impression of the target in some studies (egLombardi et al 1987 Newman amp Uleman 1990 Thompson et al 1994lowndashaccuracy participants) whereas it was unrelated to the person judg-ment in others (eg Banaji Hardin amp Rothman 1993 Experiment 2 FordStangor amp Duan 1994 Martin 1986 Martin Seta amp Crelia 1990 Stapel etal 1996 1997) Besides recall of the priming events (ie awareness of thestimulus) may not necessarily imply awareness of stimulus influenceLombardi Higgins and Bargh (1987) found that even remembering a sin-gle prime prompted contrast effects This suggests that participants werenot aware of how the prime could affect their judgment (awareness ofstimulus influence) (see also Wegner amp Bargh 1998 p 476) StrackSchwarz Bless Kuumlbler and Waumlnke (1993) had no direct measure of recallParticipants subtly reminded of the prime rendered contrastive judg-ments but were probably not clearly aware of stimulus influence Ac-cording to the setndashreset model awareness of the priming episode per sedoes not lead to contrastivejudgments (eg Martin amp Achee 1992) If par-ticipants recall the primesmdashand thus are awareof the priming eventsndash butdo not attribute to those events the increased accessibility of the conceptsin mind they will use the primed concepts in their judgments becausethoseconcepts will feel like their true reactionsto the targetThe judgmentwill then be assimilative not contrastive


Experiment 2 was designed to test directly the impact of awareness (of arelation between prime and impression formation) and recall (of thepriming words) on the judgment of the target person On the basis ofLepore and Brownrsquos (1997) model and findings activation of stereotypesfollowing priming of the category was expected to be different for peoplehigh or low in prejudice When unaware of a connection between primingand impression formation and thus unaware of stimulus influence highndashand lowndashprejudice people were expected to render divergent assimilativejudgments to the category prime as in Experiment 1 Suspecting or per-ceiving a connection between priming and impression formation shouldtrigger a correction process as it did in Experiment 1 This process couldbe initiated and carried out implicitly and result in partialling out of thetarget judgment content that is accessible but deemed irrelevant to thetarget person and the judgment at hand (eg Martin amp Achee 1992 Mar-tin amp Stapel 1998) According to Lepore and Brownrsquos (1997) modelknowledge that becomes accessible after category activation is differentfor highndash and lowndashprejudice people The correction process should thenimpact different activated content in these two groups who should sub-tract different features from their person judgments Recall of the primingstimuli is not expected to make a difference if a connection between prim-ing and judgment is not perceived consistent with the setndashreset model(Martin et al 1990 Martin amp Achee 1992)

METHODParticipants and DesignForty university students white British nationals agreed to participateThey were paid pound2 The design was a 2 (highndash lowndash Prejudice) acute 2(Aware Unaware) between participants

ProcedureApparatus and materials were the same used for Experiment 1 Partici-pants were run individually In one condition participants were told thatthey would take part in two different experiments the second one beingrun for another researcher In the other condition participants were toldthat the experiment consisted of two parts which looked different butwere in fact related To make the instructions more credible after the lex-ical decision task the computer stopped and the experimenter used acode to make it continue The screen instructions then informed the re-spondent either that the other experiment or that the other part of the ex-


periment would soon start Participants completed the impressionformation task and ratings

To measure awareness a double step procedure was used A questionappeared on the screen asking participants if they thought the two parts experiments were related They could press a key from 1 to 5 to indi-cate how much they thought the tasks were related from 1 (not at all re-lated) to 5 (very much related) If participants pressed 4 or 5 thecomputer went on asking them to write down how they thought thetasks were related A surprise recall task followed participants wereasked to write down all the words they could remember from the lexicaldecision task The prejudice scale was then completed In the extensiveindividual debriefing the experimenter further probed participants forsuspicion and recorded any explanations they gave for taskrelatedness

RESULTS

As usual participants were divided into highndash and lowndashprejudice groupsat the median (= 73 SD = 1448 Mhigh = 60 SD = 86 n = 19 Mlow = 83 SD =84n = 21) The sets of instructionsdid not produce reliable differences par-ticipants were or were not aware of a connection between priming and im-pression formation regardless of the instructions they had been givenParticipants were therefore classified as ldquoawarerdquo (n = 26)or ldquounawarerdquo (n =14)on the basis of the double step check of awareness Classificationon thismeasured variable was used in the analyses reported below In additionldquoawarerdquo participants were further classified as ldquorightrdquo (n = 12) orldquowrongrdquo(n = 14) on the basis of the judged correctness of their answers5

Two judges unaware of participantsrsquo prejudice level coded respondentsrsquoexplanationsof the link they saw(if they perceived anyconnection betweenprime and impression formation)asaccurateor notAgreement was100


5 Answers were considered right when participants mentioned the possibility of beinginfluenced by the words in the lexical decision task or when they mentioned that stereo-types might have been involved (eg ldquoThe first part seemed to suggest different peopleand places I was still thinking about these during the second part I attached meanings tosome of the words and thought about them in relation to the person describedrdquo ldquoStereo-type black person in second exercise to some people First exercise contained many wordsconnected to black people in Britainrdquo) Wrong explanations of the relation between prim-ing and impression formation included those asserting that both tasks were fast or had todo with memory etc (eg ldquoBoth parts of the experiment test a reaction to a visual stimu-lus and how you form a judgment based on that stimulusrdquo ldquoTesting memory and how fastone can transfer thoughts (ie in brain) to actions ie writing on computerrdquo)

The number of ldquohitsrdquo on the target and neutral words was calculatedfrom the recall data Participants were then divided into high or low re-call on the basis of a median split of their target words ldquohitrdquo score To as-sess the effects of recall and awareness of stimulus influence on theimpression formation task two separate sets of analyses were con-ducted

Effects of Awareness A 2 (high low Prejudice) acute 2 (Aware yes no) acute2 (positive negative Valence) mixed ANOVA with repeated mea-sures on the last factor was conducted The analysis revealed the usualmain effect for Valence (F (1 36) = 519 p lt 03) and the predictedAwareness acute Prejudice acute Valence interaction (F (1 36) = 834 p lt01)(see Figure 3)

Analysis of the simple effects revealed that the Prejudice acute Valence in-teraction was significant only when participants were unaware of a con-nection between prime and judgment (F (1 37) = 835 p lt 01) and notwhen a connection was suspected (F (1 37) = 74 ns) Replicating thepattern obtained for unaware participants in Experiment 1 highndashpreju-dice respondents rated the target person more negatively (M = 64) thanlowndashprejudice participants (M =51 t (36) = 237 p lt 025) Highndashpreju-dice individuals also rated the target person less positively (M = 562)than lowndashprejudice participants (M = 662 t (36) = ndash181 p lt 05) In addi-tion highndashprejudice participants rated the target person less positivelywhen they were not aware of a connection (M = 562) than when theywere (M = 663 t (36)= ndash196 p lt 05) Lowndashprejudice respondents ratedless negatively when they were not aware of a connection (M = 51) thanwhen they were (M = 621 t (36) = 257 p lt 01) Thus these comparisonstoo replicate the findings of Experiment 1

To explore further how awareness affected participantsrsquo responsesthe analysis was repeated splitting participants who had seen a connec-tion between lexical decision and impression formation into two groupsthose who had correctly identified the link and those who had furnishedthe wrong explanation The only significant effect was the main effect forValence (F (1 22) = 747 p lt02) Therefore it appears that it did not mat-ter whether participants accurately perceived how the stimuli could in-fluence their judgments What seemed to be important in producingassimilation or contrast effects was participantsrsquo mere suspicion of aconnection

Once again a onendashway ANOVA conducted on the prejudice scoresconfirmed that awareand unaware participants did not differ in their re-


sponses on the prejudice measure (Maware = 71 Munaware = 75 F(1 38)= 69p lt 5)

Effects of Recall To investigate the effect of recall of the priming wordson judgments of the target person a 2 (high low Prejudice) acute 2 (Recallhigh low) acute 2 (positive negative Valence) mixed ANOVA with re-peated measures on the last factor was conducted6 A marginally signifi-cant Recall acute Prejudice interaction (F (1 36) = 388 p lt06) indicated thatregardless of valence highndashprejudice participants tended to rate the tar-get person less extremely when they recalled less (Mlow recall = 59 Mhigh re-

call = 64) and lowndashprejudice participants tended to lower their ratingswhen they recalled more (Mlow recall = 64 Mhigh recall = 60) Because the in-teraction did not involve Valence this indicates that recall did not altersystematically the positivity of participantsrsquo social judgments in the im-pression formation task In fact the two interactions involving Recalland Valence were both nonsignificant (Recall acute Valence F (1 36)= 164p


4

45

5

55

6

65

7

75

A ware Unaware A ware Unaware

P OS ITIV E S CAL ES NEGA TIVE S CA LES

H ig h -p re ju d ic e

L o w -p reju d ic e

FIGURE 3 Rating of the target person as a function of Awareness and Valence of rating dimen-sions

6 Only 2 participants did not recall any priming words at all Thus we could not divideparticipants into ldquorecallrdquo and ldquonot recallrdquo groups as Lombardi et al (1987) did

lt 3 Recall acute Valence acute Prejudice F (1 36) = 00 ns) One would expectthat if participants correctly perceived the connection they would recallmore priming stimuli but in fact these participants did not recall signifi-cantly more words The ANOVA on recall scores by prejudice and 3 lev-els of awareness (unaware guessed ldquowrongrdquo guessed ldquorightrdquo) yieldedno main effects or interactions (all Fs lt1) Participantsrsquo mean overall re-call was 51 words (SD = 24) Treatment means were aware ldquorightrdquo 57aware ldquowrongrdquo 49 unaware 47 Thus the effects of recall of primingstimuli and awareness of stimulus influence were independent

DISCUSSION

Prompted by the pattern reversal due to suspicion of a connection be-tween priming and judgment in Experiment 1 Experiment 2 explicitlytested the role of awareness on impression formation Consistent withprevious research and the setndashreset model (eg Banaji et al 1993 Mar-tin et al 1990 Martin amp Achee 1992 Ford et al 1994 Stapel et al 1996)priming recall did not affect the impression formed In addition itproved to be unrelated to perception of bias Although in the present ex-periment the priming stimuli were conscious and recall of them was tobe expected the connection between priming the category and measur-ing stereotypic judgments was not direct The procedure did not implypriming a construct and relying on that same construct for the impres-sion formation task Thus recalling the primes did not in itself facilitatethe perception of a relation between primes and judgment and henceperception of a possible bias All but two participants remembered atleast some of the priming words further proof that priming recall is un-related to perception of bias If this were otherwise only two of our par-ticipants should have been unaware

In this experiment as expected judgment corrections were promotedby the perception of a connection between priming and impression for-mation7 We have postulated that highndash and lowndashprejudice people spon-


7 The experimental manipulation of ldquoawarenessrdquoperhaps too subtle was not effectiveand thus the findings rest on the measured variable a double step check of awarenessHowever procedures similar to the ones adopted in the current experiments have beenused in implicit memory research to classify participants as ldquoawarerdquo or ldquounawarerdquo (egBowers amp Schacter 1990 Schacter Bowers amp Booker 1989) This coupled with the consis-tent role of awareness in producing judgment correction in both experiments increasesconfidence in the measures and findings

taneously activate different knowledge in response to a category primeThis again was confirmed by the lack of negative stereotype activationin unaware lowndashprejudice participants Because accessible knowledge isdifferent judgment corrections triggered by perception of a relation be-tween prime and judgment were divergent in highndash and lowndashprejudicepeople Previous research has shown that resetting correction processesinvolve lowering ratings on dimensions rendered accessible by theprime and increasing ratings on ldquoprime inconsistentrdquo dimensions(Stapel et al 1998Experiment 4) Similarly here highndashprejudice partici-pants lowered negative ratings and increased positive ratings of the tar-get when partialling out activation of negative content andlowndashprejudice participants did just the opposite

Priming procedures are designed not to arouse suspicion of the relationbetween prime and judgment and people are usually unaware that thepriming stimuli can affect their judgment (see also Martin amp Stapel 1998)Thus it is not surprising that aware participantsrsquo reports of the connectionbetween the tasks were mostly vague and often wrong Yet accessibleknowledge in this study was subtracted from the judgment when partici-pants perceived a relation between prime and judgment no matter howdistant from the actual one This finding is consistent with the setndashresetmodel (Martin 1986Martin amp Stapel 1998)that allows for corrections to beinitiated automatically and for the process to be beyond introspection Inthis model corrections are performed on the content activated by the primeIt follows that within prejudice level such corrections would be unaffectedby the accuracy of the explanations articulated Accessible knowledge canbe perceived as biasing even implicitly (eg Carlston amp Smith 1996p 199Higgins 1996 Wegener amp Petty 1997 p 183) Both the absence of differ-ence in judgments rendered by ldquorightrdquo and ldquowrongrdquo participants and thefact that even ldquorightrdquo lowndashprejudice people corrected becoming more neg-ative seem to indicate that participants were unaware of the direction andmagnitude of stimulus influence8 The corrective contrast effects observedhere appear to have been the result of implicit processes


8 The measure used here may only indirectly reflect participantsrsquo perception of bias(eg participants could have perceived a stereotypical relation between tasks even whengiving the ldquowrongrdquo explanation) As the question was about the relation between tasksand not about admitting to be influenced (which participants usually deny see footnote 3)it seems reasonable to assume that what participants reported was what they were ablerather than willing to verbalize Besides the perception at the actual time of judgmentmay have been even less clear than when thinking about it afterwards It seems likely thatthe perception of bias could have been subtle and unspecified here

GENERAL DISCUSSION

The experiments reported in this paper were concerned primarily withspontaneous and relatively more controlled responses to members of so-cial groups that is with the consequences of categorization The find-ings confirm the differentiated automatic activation of stereotypes(Lepore amp Brown 1997) and provide the first evidence of divergentjudgment correction In both Experiment 1 and 2 the pattern obtained byLepore and Brown (1997)with a subliminal prime was replicated using asupraliminal priming procedure Following activation of the categoryldquoblack peoplerdquo negative stereotypic associates were activated only inunaware highndashprejudice people positive stereotypic characteristicswereactivated in unaware lowndashprejudice people However in both experi-ments this pattern reversed if participants had become aware of a possi-ble connection between priming and judgment Highndashprejudicerespondents formed a more positive impression and lowndashprejudice par-ticipants a more negative impression thus generating contrast effects intheir attempts to correct for the perceived influence of the categoryprime Recall of the priming words employed in Experiment 2 as an-other measure of awareness of stimulus influence did not affect partici-pantsrsquo judgments

According to Lepore and Brownrsquos (1997) model categorization doesnot result in automaticallndashorndashnone stereotype activationbut in the acti-vation of the endorsed aspects Divergent assimilative judgments hereresulted from an automatic and effortless process of spreading activa-tion (eg Collins amp Loftus 1975) because the stronger and more easilyactivated stereotypical associates differ for highndash and lowndashprejudicepeople (Lepore amp Brown 1997) Contrast effects emerged when highndashand lowndashprejudice people amended their impressions of the target awayfrom their differentiated spontaneous reactions resulting in divergentldquocontrolledrdquo judgments This is consistent with the Lepore and Brown(1997)rsquos model Group representations diversified by the varying en-dorsement of stereotypic characteristics seem to determine not only thedirection of automatic stereotype activation but also the direction of itscorrection

STEREOTYPE ACTIVATION AND USE

The activation of stereotypes has long been considered automatic andinevitable upon categorization (eg Allport 1954 Bargh 1999 Devine


1989 Greenwald amp Banaji 1995) The evidence however is not univocalon this point Specifically some research has uncovered factors that canprevent or facilitate stereotype activation Evidence that cognitive busy-ness (Gilbert amp Hixon 1991) prejudice level (Lepore amp Brown 1997)temporary goals (Macrae et al 1997 Sinclair amp Kunda 1999) andchronic goals (Moskowitz Gollwitzer Wasel amp Schaal 1999) can im-pede automatic stereotyping implies that stereotype activation is condi-tional rather than unconditional (Macrae amp Bodenhausen 2000)Nevertheless a view of stereotype activation as uncontrollable is stillprevalent (see for example Bargh 1999 Blair 2001 Monteith amp Voils2001)

Whilst both assimilation and contrast effects in the present studies areconsistent withmdashand predictable bymdashLepore and Brownrsquos (1997)model both patterns are difficult to predict if the activation of stereo-types is considered to be inevitable and allndashorndashnone both highndash andlowndashprejudice people should have spontaneously activated the nega-tive stereotype and there should have been no difference between themin automatic processes But there is no evidence of activation of negativestereotypes in lowndashprejudice people when unaware

As argued (see discussion of Experiment1) if stereotypes were inevi-tably activated and conscious control was needed to respond in anonprejudiced way one would expect only lowndashprejudice people tosuppress or correct their response as they see stereotype use as inappro-priate (see Devine amp Monteith 1999) This is not what happened in thepresent studies But this inconsistency could be due to the fact that re-search investigating suppression discrepancy and selfndashregulation pro-cesses typically employs procedures that invite processes morecontrolled than the ones seemingly at work here In research whereselfndashregulation was triggered by a less deliberate process (eg using amirror to elicit selfndashfocus and in turn making selfndashrelevant standards sa-lient) Macrae Bodenhausen and Milne (1998 Study 5) found that peo-ple who deemed stereotype use as appropriate stereotyped more andpeople who saw stereotype use as inappropriate stereotyped less Thusif selfndashregulation had been subtly instigated here we would have ex-pected a comparable pattern highndashprejudice people should have evi-denced even more negative stereotyping and lowndashprejudice peopleshould have used negative stereotypes even less when aware The oppo-site in fact occurred in the present studies Suppression or selfndashregula-


tion processes are then unlikely to explain the divergent correctionsobtained

What better explains the pattern found here is an implicit correctionprocess based on different spontaneously activated knowledge and re-sulting in divergent corrections

AWARENESS AND CONTRAST EFFECTS

It is assumed that comparison contrasts can be obtained through auto-matic processes and there is mounting evidence to this effect ( egMoskowitz amp Skurnik 1999 Stapel et al 1996 Weary Tobin amp Reich2001) Correction processes instead are thought to require awareness ofbias knowledge of direction and magnitude of stimulus influenceknowledge of how to correct and motivation to correct (Strack 1992Strack amp Hannover 1996 Tesser amp Martin 1996 Wegener amp Petty1997) However awareness based correction models actually allow forthe operation of implicit processes (see Martin amp Achee 1992 Martin ampStapel 1998 Wegener et al 2001) And implicit corrections seem to ex-plain the current findings best

Perceiving a connection triggers a correction process as envisaged inthe setndashreset model (Martin 1986 Martin amp Achee 1992) The resettingprocess can start automatically and be beyond the perceiverrsquos introspec-tive capabilities In this case the mere feeling that there was a connectionbetween prime and judgment may have promoted corrections whichdid not differ on account of the explanations articulated Becausehighndashprejudice people automatically activate negative stereotypic as-pects and lowndashprejudice people may activate the positive features in re-sponse to a category prime different knowledge was accessible andsubtracted from the judgment when correcting the initial reaction Indiscussing the contrast effects obtained here the meaning and role ofawareness has to be addressed

Firstly awareness of the primesrsquo influence cannot simply be equatedto an ability to recall the priming stimuli As in other studies (eg Banajiet al 1993Martin et al 1990Ford et al 1994Stapel et al 19961997) inExperiment 2 there was no systematic relationship between a measure ofrecall and target judgment in the impression formation task Priming re-call was also unrelated to feeling of a relation between prime and judg-ment that is to perception of bias This suggests that awareness of


stimulus influence and recall of priming stimuli should be considered asconceptually independent

Secondly suspicion that there might be a relation between the taskswhether or not that suspicion was wellndashfounded proved to have a deci-sive effect in both experiments To correct participants needed to beaware that the prime could influence their judgment but seemingly theydid not need to know how it could (ie the direction of stimulus influ-ence) Besides aware participantsrsquo corrections were opposite their diver-gent spontaneous judgments even when their perception of the relationbetween priming and impression formation and thus of possible biaswas accurate This is difficult to explain assuming that the correctionprocess at work rests on full conscious awareness Fully aware highndashand lowndashprejudice people who had correctly understood the relation be-tween priming and judgment would have assumed that the direction inwhich the category could influence their judgment was stereotypicallynegative and thus both groups would have corrected in the same direc-tion becoming more positive It seems less than likely that lowndashpreju-dice people would have deliberately corrected the judgment renderingit more stereotypical and negative The correction processes at work inthis case must be more implicit

According to the FCM theories can also operate implicitly (egWegener et al 2001) This then could account for the fact that there wasno difference in the kind of corrections obtained when participants per-ceived the relation between priming and judgment rightly or wronglyOpposite corrections have been found when people had different theo-ries of bias (Wegener amp Petty 1995 Experiment 4) Different correctionscould be predicted in the present case if highndash and lowndashprejudice peoplehad different theories of how the stimuli could affect the judgmentHowever this would imply that lowndashprejudice people assume that theinfluence of stereotypes on their judgment is positive and therefore cor-rect in the negative direction Such a possibility open in principle be-cause theories of bias were not assessed directly at present appears to beunlikely The stereotype used here is mainly negative and lowndashpreju-dice people know it as well as highndashprejudice people do (Lepore ampBrown1997Study 1) In addition about 80 of lowndashprejudice people inUSndashbased research report being prone to responses more prejudicedthan they deem appropriate (Devine amp Monteith 1999) It seems im-probable that lowndashprejudice people would hold such a theory of a posi-


tive stereotypical influence leading them either implicitly or explicitlyto correct by rendering more stereotypically negative judgments

Although possible an interpretation of the present findings in termsof comparison contrasts seems less parsimonious The negative (forhighndashprejudice participants)or positive (for lowndashprejudice participants)stereotypic aspects activated by the category prime could have acted ascomparison standards against which the target person was contrastedbut this intermediate step is purely speculative Crucially comparisoncontrasts observed when exemplars personalized traitndashimplying sen-tences or expectancies are the priming stimuli suit less well the primingprocedure adopted here and it has been shown that different proce-dures instigate correction or comparison processes (see Moskowitz ampSkurnik 1999)

The current discussion implies a broad meaning of ldquoawarenessrdquo injudgment correction Admittedly the measures used in these studieswere indirect and further evidence non correlational in nature and per-haps including an assessment of theories of bias may be necessary toconfirm these conclusions but the present findings suggest that themere feeling of a contextual influence may trigger a correction processSimilarly implicit are corrections instigated by mood (see Ottati amp Isbell1996 Lambert Khan Lickel amp Fricke 1997) However direct empiricalevidence of automatic corrections is scarce Glaser and Banaji (1999)have documented automatic correction contrasts but in the context ofan automatic evaluation paradigm Stapel Martin and Schwarz (1998)found that corrections triggered by blatant warnings do not require in-sight into the source of bias To our knowledge the present studies offerthe first evidence of implicit corrections in a person perception para-digm and the first evidence of divergent corrections in stereotyping

ON BEING (UN)PREJUDICED

When considering the divergent corrections obtained in these studies atfirst the attention is caught by lowndashprejudice people whose judgmentbecomes more negative when implicit correction processes act upontheir positive activated knowledge In fact that corrections have the pos-itive sidendasheffect of reducing negative stereotyping in highndashprejudicepeople may be more unusual and have potentially much more impor-tant implications After all lowndashprejudice peoplersquos judgments only be-come as negative as the corrected highndashprejudice peoplersquos judgments


And research has unveiled lowndashprejudice peoplersquos potential for egalitar-ian responses

Lowndashprejudice people have been found capable of nonprejudiced andbeliefndashcongruent responses at both the automatic and the controlledlevel of processing they do not automatically activate negative stereo-types (Kawakami et al 1998 Lepore amp Brown 1997) and their re-sponses can be nonprejudiced even on the most uncontrollable type ofmeasure a physiological one (Vanman Paul Ito amp Miller 1997)Lowndashprejudice people have also been found not to be vulnerable to re-bound effects after instructions to suppress stereotypes (Monteith et al1998) to be capable of successful selfndashregulation (Macrae et al 1998Monteith 1993) to feel a moral obligation to respond without prejudiceor else experience guilt (Devine et al 1991 Monteith amp Walters 1998)and to be internally motivated to avoid prejudiced responses (Plant ampDevine 1998) Highndashprejudice peoplersquos reactions are also consistentwith their beliefs and thus their responses tend to be stereotypical andprejudiced at the automatic and controlled level of processing they acti-vate negative stereotypes (Kawakami et al 1998Lepore amp Brown 1997Wittenbrink et al 1997) and generally use them as they do not experi-ence guilt for prejudiced responses (Devine et al 1991 Monteith ampWalters 1998) They are however externally motivated to avoid stereo-typing (Plant amp Devine 1998) and thus will avoid prejudiced responsesif a social norm is salient or under outcome dependency (Monteith ampVoils 2001)

To be sure the judgment situation that the present studies might mirroris not one where full conscious control is possible as for example whenevaluating job applicants Full awareness that a bias may taint the judg-ment would lead to correction and even overcorrection for the perceivedbias (Wegener amp Petty 1997 Wilson amp Brekke 1994) According toWegener and Petty (1997) lowndashprejudice people having the motivationto correct would be more likely to do so Highndashprejudice people couldcorrect too if a norm against discrimination was salient But situations arenot always so clearndashcut Devine and Monteith (1999)have argued for ex-ample that in some cases it may not be clear what response is egalitarianThe present data suggest that in more ambiguous judgment situationsand in the absence of external motivation implicit corrections may lead toless stereotyping on the part of highndashprejudice people acting as an inter-nal trigger for people who usually are only motivated externally Re-search has concentrated more on lowndashprejudice people than on


highndashprejudice people in regard to stereotype activation and use(Monteith amp Voils 2001)But how highndashprejudice people react in differentsituations should be considered thoroughly as these may be the peoplefor whom effective prejudice reduction strategies are more necessary Thepresent research provides the first evidence that implicit corrections canreduce prejudiced responses in prejudiced individuals Although furtherresearch is surely needed to elucidate this point and fully draw its impli-cations the first glimpses of such spontaneous reductions of stereotypingin prejudiced people could be a promising beginning

REFERENCESAllport GW (1954) The nature of prejudice Reading Ma AddisonndashWesleyBanaji MR amp Greenwald AG (1995) Implicit gender stereotyping in judgments of fame Jour-

nal of Personality and Social Psychology 68 181ndash198Banaji MR amp Hardin C (1996) Automatic stereotyping Psychological Science 7 136ndash141Banaji MR Hardin C amp Rothman AJ (1993) Implicit stereotyping in person judgment Jour-

nal of Personality and Social Psychology 65 272ndash281Bargh JA (1992) Does subliminality matter to social psychology Awareness of the stimulus

versus awareness of its influence In RF Bornstein amp TS Pittman (Eds) Perception with-out awareness Cognitive clinical and social perspectives (pp 236ndash255) New York GuilfordPress

Bargh JA (1994) The four horseman of automaticity Awareness intention Efficiency and con-trol in social cognition In RS Wyer Jramp TK Srull (Eds) The Handbook of Social CognitionVol 2 Basic Processes (pp 1ndash40) Hillsdale NJ Erlbaum

Bargh JA (1996) Automaticity in Social Psychology In ET Higgins amp A W Kruglanski (Eds)Social Psychology Handbook of Basic Principles NY Guilford Press

Bargh JA (1999) The cognitive monster The case against the controllability of automatic stereo-type effects In S Chaiken amp Y Trope (Eds) Dualndashprocess theories in social psychology (pp361ndash382) New York Guilford Press

Bargh JA amp Chartrand TL (2000) The mind in the middle A practical guide to priming andautomaticity research In HT Reis amp CM Judd (Eds) Handbook of research methods in so-cial and personality psychology (pp 253ndash285) Cambridge UK Cambridge UniversityPress

Blair IV (2001) Implicit stereotypes and prejudice In GBMoskowitz (Ed) Cognitive social psy-chology The Princeton Symposium on the Legacy and Future of Social Cognition (pp 359ndash374)Mahwah NJ Lawrence Erlbaum

Bornstein RF amp DrsquoAgostino PR (1992) Stimulus recognition and the mere exposure effectJournal of Personality and Social Psychology 63 545ndash552

Bowers JS amp Schacter DL (1990) Implicit memory and test awareness Journal of ExperimentalPsychology Learning Memory and Cognition 16 404ndash416

Carlston D E (1992) Impression formation and the modular mind The associated systems the-ory In LL Martin amp A Tesser (Eds) The Construction of Social Judgments (pp 301ndash341)Hillsdale NJ Erlbaum

Carlston DE amp Smith ER (1996) Principles of mental representation In ET Higgins amp AWKruglanski (Eds) Social Psychology Handbook of Basic Principles (pp 184ndash210) New YorkGuilford Press

Collins AM amp Loftus EF (1975) A spreadingndashactivation theory of semantic processing Psy-chological Review 82 407ndash428

Devine P G (1989) Stereotypes and prejudice their automatic and controlled componentsJournal of Personality and Social Psychology 56 1 5ndash18

Devine PG amp Monteith MJ (1993) The role of discrepancyndashassociated affect in prejudice re-duction In DM Mackie amp DL Hamilton (Eds) Affect Cognition and Stereotyping Inter-active Processes in Group Perception (pp 317ndash344) San Diego Academic Press


Devine PG amp Monteith MJ (1999) Automaticity and control in stereotyping In S Chaiken ampY Trope (Eds) Dualndashprocess theories in social psychology (pp 339ndash360) New York GuilfordPress

Devine PG Monteith MJ Zuwerink JR amp Elliot AJ (1991) Prejudice with and withoutcompuction Journal of Personality and Social Psychology 60 817ndash830

Fazio R H Jackson JR Dunton BC amp Williams CJ (1995) Variability in automatic activa-tion as an unobtrusive measure of racial attitudes A bona fide pipeline Journal of Person-ality and Social Psychology 69 1013ndash1027

Ford TE amp Kruglanski AW (1995) Effects of epistemic motivations on the use of accessibleconstructs in social judgments Personality and Social Psychology Bulletin 21 950ndash962

Ford TE Stangor C amp Duan C (1994) Influence of social category accessibility and cate-goryndashassociated trait accessibility on judgments of individuals Social Cognition 12149ndash168

Ford TE amp Thompson EP (2000) Preconscious and postconscious processes underlying con-struct accessibility effects An extended search model Personality and Social Psychology Re-view 4 317ndash336

Gaertner SL amp Dovidio JF (1986) The aversive form of racism In Dovidio JF and GaertnerSL (Eds) Prejudice Discrimination and Racism NY Academic Press

Glaser J amp Banaji MR (1999) When fair is foul and foul is fair Reverse priming in automaticevaluation Journal of Personality and Social Psychology 77 669ndash687

Greenwald AG amp Banaji MR (1995) Implicit social cognition Attitudes selfndashesteem and ste-reotypes Psychological Review 102 4ndash27

Herr P M (1986) Consequences of priming Judgement and behavior Journal of Personality andSocial Psychology 51 1106ndash1115

Higgins TE (1989) Knowledge accessibility and activation Subjectivity and suffering from un-conscious sources In JS Uleman amp JA Bargh (Eds) Unintended Thought (pp 75ndash123)New York Guilford Press

Higgins TE (1996) Knowledge activation Accessibility applicability and salience In ET Hig-gins amp A W Kruglanski (Eds) Social Psychology Handbook of basic principles (pp133ndash168) New York Guilford Press

Higgins E T Rholes W S amp Jones C R (1977) Category accessibility and impression forma-tion Journal of Experimental Social Psychology 13 141ndash154

Jacobson CK (1985) Resistance to affirmative action Selfndashinterest or racism Journal of ConflictResolution 29 306ndash329

Jacoby LL amp Kelley CM (1987) Unconscious influences of memory for a prior event Personal-ity and Social Psychology Bulletin 14 314ndash336

Kawakami K Dion KL amp Dovidio JF (1998) Racial prejudice and stereotype activation Per-sonality and Social Psychology Bulletin 24 407ndash416

Lambert AJ Khan SR Lickel BA amp Fricke K (1997) Mood and the correction of positiveversus negative stereotypes Journal of Personality and Social Psychology 72 2 1002ndash1016

Lepore L amp Brown R (1997) Category and stereotype activation Is prejudice inevitable Jour-nal of Personality and Social Psychology 72 2 275ndash287

Locke V MacLeod C amp Walker I (1994) Automatic and controlled activation of stereotypesIndividual differences associated with prejudice British Journal of Social Psychology 3329ndash46

Lombardi W J Higgins E T amp Bargh J A (1987) The role of consciousness in priming effectson categorization Assimilation versus contrast as a function of awareness of primingtask Personality and Social Psychology Bulletin 13 411ndash429

Macrae CN amp Bodenhausen GV (2000) Social cognition Thinking categorically about oth-ers Annual Review of Psychology 51 93ndash120

Macrae CN Bodenhausen GV amp Milne AB (1998) Saying no to unwanted thoughtsSelfndashfocus and the regulation of mental life Journal of Personality and Social Psychology 74578ndash589

Macrae C N Bodenhausen G V Milne A B amp Jetten J (1994) Out of mind but back in sightStereotypes on the rebound Journal of Personality and Social Psychology 67 808ndash817

Macrae C N Bodenhausen G V Milne A B Thorn TMJ amp Castelli L (1997) On the activa-tion of social stereotypes The moderating role of processing objectives Journal of Experi-mental Social Psychology 33 471ndash489

Martin L L (1986) Setreset The use and disuse of concepts in impression formation Journal ofPersonality and Social Psychology 51 493ndash504

Martin L L amp Achee J W (1992) Beyond accessibility The role of processing objectives in


judgment In L L Martin amp A Tesser (Eds) The construction of social judgments (pp195ndash216) Hillsdale NJ Erlbaum

Martin L L Seta J J amp Crelia R A (1990) Assimilation and contrast as a function of peoplersquoswillingness and ability to expend effort in forming an impression Journal of Personalityand Social Psychology 57 27ndash37

Martin LL amp Stapel DA (1998) Correction and metacognition Are people naiumlve dogmatistsor naiumlve empiricists In V Yzerbyt G Lories amp B Dardenne (Eds) Metacognition Cogni-tive and social dimensions (pp 228ndash247) London UK Sage

McConahay JG (1986) Modern racism ambivalence and the Modern Racism Scale In JFDovidio amp SL Gaertner (Eds) Prejudice Discrimination and Racism (pp 91ndash125)NewYork Academic Press

Meyer DE amp Schvaneveldt RW (1971) Facilitation in recognizing pairs of words Evidence ofa dependence between retrieval operations Journal of Experimental Psychology 90227ndash234

Monteith MJ (1993) Selfndashregulation of prejudiced responses Implications for progress in prej-udicendashreduction efforts Journal of Personality and Social Psychology 65 469ndash485

Monteith MJ Sherman JW amp Devine PG (1998) Suppression as a stereotype control strat-egy Personality and Social Psychology Review 2 63ndash82

Monteith MJ Spicer CV amp Tooman GD (1998) Consequences of stereotype suppressionStereotypes on AND not on the rebound Journal of Experimental Social Psychology 34355ndash377

Monteith MJ amp Voils CI (2001) Exerting control over prejudiced responses InGBMoskowitz (Ed)Cognitive social psychology The Princeton Symposium on the Legacy andFuture of Social Cognition (pp 375ndash388) Mahwah NJ Lawrence Erlbaum

Monteith MJ amp Walters GL (1998) Egalitarianism moral obligation and prejudicendashrelatedstandards Personality and Social Psychology Bulletin 24 186ndash199

Moskowitz GB Gollwitzer PM Wasel W amp Schaal B (1999) Preconscious control of stereo-type activation through chronic egalitarian goals Journal of Personality and Social Psychol-ogy 77 167ndash184

Moskowitz GB amp Skurnik IW (1999) Contrast effects as determined by the type of primeTrait versus exemplar primes initiate processing strategies that differ in how accessibleconstructs are used Journal of Personality and Social Psychology 76 911ndash927

Newman LS Duff KJ Hedberg DA amp Blitstein J (1996) Rebound effects in impression for-mation Assimilation and contrast effects following thought suppression Journal of Ex-perimental Social Psychology 32 460ndash483

Newman L S amp Uleman J S (1990) Assimilation and contrast effects in spontaneous trait in-ference Personality and Social Psychology Bulletin 16 224ndash240

Ottati VC amp Isbell LM (1996) Effects of mood during exposure to target information on sub-sequently reported judgments An onndashline model of misattribution and correction Jour-nal of Personality and Social Psychology 71 39ndash53

Pettigrew TF amp Meertens RW (1995) Subtle and blatant prejudice in Western Europe Euro-pean Journal of Social Psychology 25 57ndash75

Petty R E amp Wegener D T (1993) Flexible correction processes in social judgment Correctingfor contextndashinduced contrast Journal of Experimental Social Psychology 29 137ndash165

Plant EA amp Devine PG (1998) Internal and external sources of motivation to respond with-out prejudice Journal of Personality and Social Psychology 75 811ndash832

Schacter DL Bowers J amp Booker J (1989) Intention awareness and implicit memory The re-trieval intentionality criterion In S Lewandowsky JC Dunn amp K Kirsner (Eds) Implicitmemory Theoretical issues (pp 47ndash65) Hillsdale NJ Erlbaum

Schwarz N amp Bless H (1992) Constructing reality and its alternatives An inclusionexclusionmodel of assimilation and contrast effects in social judgment In L Martin amp A Tesser(Eds) The construction of social judgment (pp 217ndash245) Hillsdale Erlbaum

Sinclair L amp Kunda Z (1999) Reactions to a black professional Motivated inhibition and acti-vation of conflicting stereotypes Journal of Personality and Social Psychology 77 885ndash904

Skowronski JJ Carlston DE amp Isham JT (1993) Implicit versus explicit impression forma-tion the differing effects of overt labelling and covert priming on memory and impres-sions Journal of Experimental Social Psychology 29 17ndash41

Srull TK amp Wyer RS Jr (1979) The role of category accessibility in the interpretation of infor-mation about persons Some determinants and implications Journal of Personality and So-cial Psychology 37 1660ndash1672

Srull TK amp Wyer RS Jr (1980) Category accessibility and social perception Some implica-


tions for the study of person memory and interpersonal judgments Journal of Personalityand Social Psychology 38 841ndash856

Stapel DA Koomen W amp Van der Pligt J (1996) The referents of trait inferences The impactof trait concepts versus actorndashtrait links on subsequent judgments Journal of Personalityand Social Psychology 70 437ndash450

Stapel DA Koomen W amp Van der Pligt J (1997) Categories of category accessibility The im-pact of trait versus exemplar priming on person judgments Journal of Experimental SocialPsychology 33 44ndash76

Stapel DA Koomen W amp Zeelenberg M (1998) The impact of accuracy motivation on inter-pretation comparison and correction processes Accuracy acute knowledge accessibility ef-fects Journal of Personality and Social Psychology 74 878ndash893

Stapel DA Martin LL amp Schwarz N (1998) The smell of bias What instigates correction pro-cesses in social judgments Personality and Social Psychology Bulletin 24 797ndash806

Strack F (1992) The different routes to social judgments Experiential versus informationalstrategies In L L Martin amp A Tesser (Eds) The construction of social judgments HillsdaleNJ Erlbaum

Strack F amp Hannover B (1996) Awareness of influence as a precondition for implementingcorrectional goals In PM Gollwitzer amp JA Bargh (Eds) The psychology of action Linkingcognition and motivation to behavior (pp 579ndash596) New York Guilford

Strack F Schwarz N Bless H Kubler A amp Wanke M (1993) Awareness of the influence as adeterminant of assimilation versus contrast European Journal of Social Psychology 2353ndash62

Tesser A amp Martin L (1996) The psychology of evaluation In ET Higgins amp AW Kruglanski(Eds) Social psychology Handbook of basic principles (pp 400ndash432) New York Guilford

Thompson EP Roman RJ Moskowitz GB Chaiken S amp Bargh JA (1994) Accuracy moti-vation attenuates covert primingThe systematic reprocessing of social information Jour-nal of Personality and Social Psychology 66 474ndash489

Vanman EJ Paul B Y Ito TA amp Miller N (1997) The modern face of prejudice and struc-tural features that moderate the effect of cooperation on affect Journal of Personality andSocial Psychology 73 941ndash959

Weary G Tobin SJ amp Reich DA (2001) Chronic and temporary distinct expectancies as com-parison standards Automatic contrasts in dispositional judgments Journal of Personalityand Social Psychology 80 365ndash380

Wegener D T Dunn M amp Tokusato D (2001) The flexible correction model Phenomenologyand the use of naiumlve theories in avoiding or removing bias In GBMoskowitz (Ed) Cog-nitive social psychology The Princeton Symposium on the Legacy and Future of Social Cognition(pp277ndash290) Mahwah NJ Lawrence Erlbaum

Wegener DT amp Petty RE (1995) Flexible correction processes in social judgment The role ofnaive theories in corrections for perceived bias Journal of Personality and Social Psychology68 36ndash51

Wegener DT amp Petty RE (1997) The flexible correction model The role of naive theories ofbias in bias correction In M Zanna (Ed) Advances in Experimental Social Psychology (Vol29 pp 141ndash208) San Diego Academic Press

Wegener DT Petty RE amp Dunn M (1998) The metacognition of bias correction Naiumlve theo-ries of bias and the Flexible Correction Model In V Yzerbyt G Lories amp B Dardenne(Eds) Metacognition Cognitive and social dimensions (pp 202ndash227) London UK Sage

Wegner DM amp Bargh JA (1998) Control and automaticity in social life In D Gilbert S Fiskeamp G Lindzey (Eds) The Handbook of Social Psychology (4th ed Vol 1 pp 446ndash496) NewYork NY McGrawndashHill

Winkielman P amp Schwarz N (1996 May)Contrast and assimilation with and without awarenessAtest of the inclusionexclusion model Paper presented at the annual meeting of the Midwest-ern Psychological Association Chicago

Wilson TD amp Brekke N (1994) Mental contamination and mental correction Unwanted influ-ences on judgments and evaluations Psychological Bulletin 116 117ndash142

Wittenbrink B Judd CM amp Park B (1997) Evidence for racial prejudice at the implicit leveland its relationship with questionnaire measures Journal of Personality and Social Psychol-ogy 72 2 262ndash274

Wyer NA Sherman JW amp Stroessner SJ (2000) The roles of motivation and ability in con-trolling the consequences of stereotype suppression Personality and Social Psychology Bul-letin 26 13ndash25


vant to everyday social encounters What are the consequences of catego-rization What are peoplersquos automatic responses to members ofstereotyped groups Theories of modern (McConahay 1986) aversive(Gaertner amp Dovidio 1986) and subtle racism (Pettigrew amp Meertens1995)all statethatprejudice lingers despite its reduction in expressed atti-tudes and stereotype content arising from changed norms of social inter-actionBased on these conceptualizations it is often surmised that explicitresponses to category members will appear to be nonprejudiced whereasthe implicit and automatically accessed responses will still be prejudicedSuch incongruency between implicit and explicit attitudes is implied forexample in Devinersquos (1989)seminal work According to her model nega-tive stereotypes are activated automatically upon categorization regard-less of peoplersquos prejudice level Their use however can be consciouslycontrolled if motivation and cognitive capacity allow for the suppressionof the automatic prejudiced response Other studies support theincongruency of implicit and explicit attitudes Banaji and Greenwald(1995) for example found that implicit gender bias in judgments of fameand explicit gender measures were not correlated (see also Banaji ampHardin 1996 for a review see Blair 2001) Such findings suggest that ste-reotypes are activated automatically in an allndashorndashnone manner and re-gardless of peoplersquos beliefs about the target group Deriving from thepremise that stereotypes are activated automatically upon categorization(Allport 1954 Bargh 1999 Devine 1989Greenwald amp Banaji 1995)is re-search investigating the effects of stereotype suppression (eg Monteith1993 Wyer Sherman amp Stroessner 2000) The conscious suppression ofprejudiced responses elicited by the automatic and seemingly inevitableactivation of negative stereotypes should curb unwanted stereotyping injudgments and behaviors (but cf Macrae Bodenhausen Milne amp Jetten1994) in a process of selfndashregulation eventually leading to the automatiza-tion of such suppression itself (see Monteith Sherman amp Devine 1998andMonteith amp Voils 2001 for reviews)

At present evidence regarding the congruency or incongruency of im-plicit and explicit attitudes is contradictory and not yet reconciled (seeBlair 2001) In contrastwith the findings reported above several studieshave observed automatic facilitation effects to be correlated with scoreson a prejudice measure (Lepore amp Brown 1997 Locke MacLeod ampWalker 1994 Wittenbrink Judd amp Park 1997 See also Fazio JacksonDunton amp Williams 1995 Kawakami Dion amp Dovidio 1998) Investi-gating the relation between categorizationand stereotyping Lepore and




































RESULTS










4

45

5

55

6

65

7

75



High-Prejudice

Low-Prejudice




DISCUSSION






4

4 5

5

5 5

6

6 5

7

7 5























RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION





























































































































































RESULTS










4

45

5

55

6

65

7

75



High-Prejudice

Low-Prejudice




DISCUSSION






4

4 5

5

5 5

6

6 5

7

7 5























RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION































































































































4

45

5

55

6

65

7

75



High-Prejudice

Low-Prejudice




DISCUSSION






4

4 5

5

5 5

6

6 5

7

7 5























RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION





























































































































DISCUSSION






4

4 5

5

5 5

6

6 5

7

7 5























RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION





























































































































4

4 5

5

5 5

6

6 5

7

7 5























RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION












































































































































RESULTS















4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION





































































































































4

45

5

55

6

65

7

75








DISCUSSION









GENERAL DISCUSSION




























































































































DISCUSSION









GENERAL DISCUSSION































































































































GENERAL DISCUSSION



























































































































the role of awareness: divergent automatic stereotype ...faculty.weber.edu/eamsel/research...

Documents