This article was downloaded by: [The University of British Columbia]On: 22 July 2013, At: 09:59Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Psychiatry, Psychology and LawPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tppl20

Dangerous Decisions: Influence of JurorAttitudes and Defendant Appearanceon Legal Decision-MakingNatasha Korva a , Stephen Porter a , Brian P. O'Connor a , JuliaShaw a & Leanne ten Brinke aa Centre for the Advancement of Psychological Science and Law,University of British Columbia , Kelowna , CanadaPublished online: 28 Jun 2012.

To cite this article: Natasha Korva , Stephen Porter , Brian P. O'Connor , Julia Shaw &Leanne ten Brinke (2013) Dangerous Decisions: Influence of Juror Attitudes and DefendantAppearance on Legal Decision-Making, Psychiatry, Psychology and Law, 20:3, 384-398, DOI:10.1080/13218719.2012.692931

To link to this article: http://dx.doi.org/10.1080/13218719.2012.692931

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

© 2013 The Australian and New Zealand Association of Psychiatry, Psychology and Law

Psychiatry, Psychology and Law, 2013Vol. 20, No. 3, 384–398, http://dx.doi.org/10.1080/13218719.2012.692931

Dangerous Decisions: Influence of Juror Attitudes and DefendantAppearance on Legal Decision-Making

Natasha Korva*, Stephen Porter, Brian P. O’Connor, Julia Shaw and Leanne ten Brinke

Centre for the Advancement of Psychological Science and Law, University of British Columbia,Kelowna, Canada

According to the dangerous decisions theory (Porter, S., & ten Brinke, L. (2009).Dangerous decisions: A theoretical framework for understanding how judges assesscredibility in the courtroom. Legal and Criminological Psychology, 14(1), 119–134),intuitive evaluations of facial trustworthiness influence the interpretation of evidencepresented in courtroom settings. This study examined the role of individual differences inthis process. Participants were presented with crime narratives, accompanied by aphotograph of the supposed defendant that was previously rated as highly trustworthy oruntrustworthy. Following presentation of the evidence in each case, participants ratedthe defendant’s culpability and then completed questionnaires assessing potential biases.Participants endorsing justice–fairness were more likely to exonerate an untrustworthy-looking defendant, but less likely to exonerate a trustworthy-looking one. Individualswho held a strong racial bias, by contrast, were less likely to be influenced by exoneratingevidence, specifically for untrustworthy-looking defendants. These results suggest thatfaces varying in trustworthiness activate particular biases, and a tunnel vision approachto decision-making that may lead to wrongful convictions in a legal setting.

Key words: attitudes; bias; dangerous decisions; legal decisions; trustworthiness.

Based on another’s facial appearance,observers rapidly make inferences abouthis/her character prior to any interpersonalinteraction (Bull, 2006; Bull & Vine, 2003;Martelli, Majib, & Pelli, 2005; Willis &Todorov, 2006). One of the first inferencesmade upon viewing a stranger’s face is anassessment of his/her trait trustworthiness;observers come to instantaneous (less than1/10th of a second) and confident decisionsabout whether a face can be trusted (Baret al., 2006; Willis & Todorov, 2006), basedon cues gleaned from facial structure.

Evaluations of the trustworthiness of bothmale and female faces are based on a three-dimensional model relating to the face’sstructure: perceived dispositional valence(i.e., the trait-face structure suggests anger,happiness, etc.), dominance and attractive-ness (Oosterhof & Todorov, 2008; Porter,ten Brinke, & Mathesius, unpublished).For example, facial characteristics such aslarge round eyes, high eyebrows and asmall chin are considered baby-faced qua-lities, and are typically associated withkindness, warmth and honesty (Berry &

Correspondence: Stephen Porter, Centre for the Advancement of Psychological Science and Law,University of British Columbia – Okanagan, ASC II, ASC204, 3333 University Way, Kelowna, BC,V1V 1V7, Canada. Email: stephen.porter@ubc.ca

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 385

Dangerous Decisions: Influence of Juror Attitudes and DefendantAppearance on Legal Decision-Making

Natasha Korva*, Stephen Porter, Brian P. O’Connor, Julia Shaw and Leanne ten Brinke

Centre for the Advancement of Psychological Science and Law, University of British Columbia,Kelowna, Canada

According to the dangerous decisions theory (Porter, S., & ten Brinke, L. (2009).Dangerous decisions: A theoretical framework for understanding how judges assesscredibility in the courtroom. Legal and Criminological Psychology, 14(1), 119–134),intuitive evaluations of facial trustworthiness influence the interpretation of evidencepresented in courtroom settings. This study examined the role of individual differences inthis process. Participants were presented with crime narratives, accompanied by aphotograph of the supposed defendant that was previously rated as highly trustworthy oruntrustworthy. Following presentation of the evidence in each case, participants ratedthe defendant’s culpability and then completed questionnaires assessing potential biases.Participants endorsing justice–fairness were more likely to exonerate an untrustworthy-looking defendant, but less likely to exonerate a trustworthy-looking one. Individualswho held a strong racial bias, by contrast, were less likely to be influenced by exoneratingevidence, specifically for untrustworthy-looking defendants. These results suggest thatfaces varying in trustworthiness activate particular biases, and a tunnel vision approachto decision-making that may lead to wrongful convictions in a legal setting.

Key words: attitudes; bias; dangerous decisions; legal decisions; trustworthiness.

Based on another’s facial appearance,observers rapidly make inferences abouthis/her character prior to any interpersonalinteraction (Bull, 2006; Bull & Vine, 2003;Martelli, Majib, & Pelli, 2005; Willis &Todorov, 2006). One of the first inferencesmade upon viewing a stranger’s face is anassessment of his/her trait trustworthiness;observers come to instantaneous (less than1/10th of a second) and confident decisionsabout whether a face can be trusted (Baret al., 2006; Willis & Todorov, 2006), basedon cues gleaned from facial structure.

Evaluations of the trustworthiness of bothmale and female faces are based on a three-dimensional model relating to the face’sstructure: perceived dispositional valence(i.e., the trait-face structure suggests anger,happiness, etc.), dominance and attractive-ness (Oosterhof & Todorov, 2008; Porter,ten Brinke, & Mathesius, unpublished).For example, facial characteristics such aslarge round eyes, high eyebrows and asmall chin are considered baby-faced qua-lities, and are typically associated withkindness, warmth and honesty (Berry &

Correspondence: Stephen Porter, Centre for the Advancement of Psychological Science and Law,University of British Columbia – Okanagan, ASC II, ASC204, 3333 University Way, Kelowna, BC,V1V 1V7, Canada. Email: stephen.porter@ubc.ca

McArthur, 1986; Berry & Zebrowitz-McArthur, 1988; Zebrowitz, Voinescu, &Collins, 1996).

These evaluations play a major role insubsequent decisions about others in var-ious interpersonal contexts (Eberhardt,Davies, Purdie-Vaughns, & Johnson,2006; Olivola & Todorov, 2010) and cancontribute to prejudice and/or discrimina-tion (e.g., Porter & ten Brinke, 2009). Inlegal settings, such evaluations can holdtremendous consequences. For example,individuals with perceived baby-faced qua-lities receive relatively lenient sentences(Berry & Zebrowitz-McArthur, 1988),while attractive defendants are perceivedas more honest (Zebrowitz et al., 1996) andare less likely to be deemed guilty thantheir unattractive counterparts (Kulka &Kessler, 1978). Studies of racial bias in theUSA have shown that males rated as morestereotypically Black are more likely to besentenced to death, compared with theirless stereotypical counterparts (Eberhardtet al., 2006). Further, certain faces areconsidered congruent with specific types ofcrimes (Bull & McAlpine, 1998; Dumas &Teste, 2006), and have perceived stereo-typical features of rapists, armed robbersor murderers. Defendants whose faces arecongruent with facial characteristics com-monly associated with these crimes aremore likely to be found guilty in legalsettings, regardless of weak evidence (Du-mas & Teste, 2006). And defendants withan ‘‘untrustworthy’’-looking face are con-victed of murder (by mock jurors) based onless incriminating evidence than trust-worthy-looking defendants (Porter, tenBrinke, & Gustaw, 2010).

Dangerous Decisions Theory

Although interpersonal evaluations basedon appearance can be consequential, evi-dence suggests that split-second evaluationsof trustworthiness are often inaccurate,providing only a slight advantage over

guessing (ten Brinke&Porter, 2011). Porter,England, Juodis, and ten Brinke (2008)asked participants to rate the trustworthi-ness of violent America’s Most Wantedcriminals and comparatively trustworthyphilanthropists. Participants were able todiscriminate group membership onlyslightly above the level of chance. Similarly,Bond, Berry, and Omar (1994) found thatappearance-based impressions of honestyaccounted for only a ‘‘kernel of truth’’(4% of the variance) in the target’s actualwillingness to engage in deceptive behavior(see also Zebrowitz et al., 1996).

Although trustworthiness assessmentsare of questionable validity, it has beenproposed that they set in motion a power-ful decision-making process, characterizedby natural human biases, tunnel vision andoverconfidence. The dangerous decisionstheory (DDT; Porter & ten Brinke, 2009)predicts that interpersonal judgments oftrustworthiness occur instantaneouslyupon seeing a face (Willis & Todorov,2006), which subjectively may be experi-enced as ‘‘intuition’’. While it is possiblethat these intuitive judgments are correct,incorrect impressions can lead to biased (or‘‘dangerous’’) decisions concerning thetarget. This rapid process of trustworthi-ness assessment likely served the functionof reducing danger to our human ances-tors. While the initial assessment is im-plicit, the high-stakes involved withtrustworthiness decisions in a court settingare likely to generate an increasinglyconscious engagement of the defendant’scharacteristics. However, the initial impres-sion of a defendant’s trustworthiness islikely to have an enduring influence on themanner in which new information concern-ing the target is interpreted and assimilatedby judges and jurors. Specifically, the initialevaluation can influence subsequent infer-ences concerning the defendant (or otherwitness) by making decision-making abouthim/her increasingly irrational (Kahneman& Tversky, 1982). This can then generate a

Juror Attitudes and Defendant Appearance

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

386 N. Korva et al.

non-critical, ‘‘tunnel vision’’ assimilationof potentially ambiguous or contradictoryevidence concerning the defendant. This isakin to setting a ‘‘confirmation bias’’ intoeffect, wherein jurors seek to confirm theirinitial evaluation of trustworthiness (oruntrustworthiness) by attending only toinformation that supports this notion (e.g.,Pyszczynski & Greenberg, 1987; Snyder &Stukas, 1999).

In a study of criminal investigators,Ask and Granhag (2007) found strongsupport for this ‘‘asymmetrical skepti-cism’’, the tendency to be more skepticalabout evidence that runs counter to one’sprior belief than evidence consistent withthe belief. While most judges and jurorsprobably are circumspect in their efforts tomake the correct decisions concerningcredibility, it may be possible to work toohard in this context; high motivation canexacerbate the level of bias in decisionsabout credibility. Porter, McCabe, Wood-worth, and Peace (2007) identified amotivational impairment effect such thata high level of motivation in a deceptiondetection task was negatively associatedwith accuracy (also see Ask & Granhag,2007). A high level of motivation such asthat felt by a judge or juror, coupled withthe complexity of credibility assessment,may serve to increase the power of theinitial perception of trustworthiness andcreate tunnel vision decision-making. DDTis related to several other theoreticalframeworks, including the ‘‘blame-valida-tion’’ process suggested in Alicke’s (2000)culpable control theory and the motiva-tionally driven inferences proposed inTetlock and colleagues’ conception of the‘‘intuitive prosecutor’’ (Tetlock et al.,2007). The fundamental assumption ofthe culpable control theory is that evidenceconcerning negative events is scrutinizedfor its contribution to personal control(culpability) and is spontaneously evalu-ated for its favorableness (Alicke, 2000;Alicke, Buckingham, Zell, & Davis, 2008).

These spontaneous evaluations then incitea blame-validation process, wherein evi-dence concerning the event is reviewed in amanner that favors ascribing blame to thedefendant who evokes the most negativeaffect. This is similar to the assumptionunderlying the tunnel vision induced byinitial impressions of trustworthiness in theDDT model. Another theory based on aframework of attributions of responsibility,the ‘‘intuitive prosecutor’’ (Tetlock et al.,2007), explores how individuals come tomake character attributions for normviolators. According to this framework,we have evolved to be social watchdogsthat punish norm violators and havedeveloped a ‘‘fair-but-biased-yet-correcti-ble’’ (FBC; Tetlock, 2002) method ofdealing with them. The FBC methodsuggests that we see ourselves as fairindividuals, yet we are often biased deci-sion-makers, and when we notice our ownbiases we attempt to self-correct ourcognitions. The self-correction piece ofthis framework has not been adequatelyexplored in a legal decision-making con-text. Research to date has not found aspontaneous end to the tunnel vision thatensues in many legal proceedings. UnlikeDDT, these frameworks are based on thenotion that we have a desire to inferpersonal control and responsibility toothers’ behavior (similar to the ‘‘justworld’’ framework; e.g., Hafer & Begue,2005). DDT simply suggests that, based onevolutionary mechanisms, we make intui-tive judgments of ‘‘friend or foe’’ based onappearance that lead to biased decision-making. Although they differ in a variety ofways, all of these theories suggest thatimplicit cognitions and values influence theinterpretation of defendant informationand lead to generally adverse attributionsof blame. DDT is favorable over compet-ing theories for explaining bias in legaldecision-making for this study because itexplains the isolated effects of appearanceon judicial biases.

N. Korva

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 387

non-critical, ‘‘tunnel vision’’ assimilationof potentially ambiguous or contradictoryevidence concerning the defendant. This isakin to setting a ‘‘confirmation bias’’ intoeffect, wherein jurors seek to confirm theirinitial evaluation of trustworthiness (oruntrustworthiness) by attending only toinformation that supports this notion (e.g.,Pyszczynski & Greenberg, 1987; Snyder &Stukas, 1999).

In a study of criminal investigators,Ask and Granhag (2007) found strongsupport for this ‘‘asymmetrical skepti-cism’’, the tendency to be more skepticalabout evidence that runs counter to one’sprior belief than evidence consistent withthe belief. While most judges and jurorsprobably are circumspect in their efforts tomake the correct decisions concerningcredibility, it may be possible to work toohard in this context; high motivation canexacerbate the level of bias in decisionsabout credibility. Porter, McCabe, Wood-worth, and Peace (2007) identified amotivational impairment effect such thata high level of motivation in a deceptiondetection task was negatively associatedwith accuracy (also see Ask & Granhag,2007). A high level of motivation such asthat felt by a judge or juror, coupled withthe complexity of credibility assessment,may serve to increase the power of theinitial perception of trustworthiness andcreate tunnel vision decision-making. DDTis related to several other theoreticalframeworks, including the ‘‘blame-valida-tion’’ process suggested in Alicke’s (2000)culpable control theory and the motiva-tionally driven inferences proposed inTetlock and colleagues’ conception of the‘‘intuitive prosecutor’’ (Tetlock et al.,2007). The fundamental assumption ofthe culpable control theory is that evidenceconcerning negative events is scrutinizedfor its contribution to personal control(culpability) and is spontaneously evalu-ated for its favorableness (Alicke, 2000;Alicke, Buckingham, Zell, & Davis, 2008).

These spontaneous evaluations then incitea blame-validation process, wherein evi-dence concerning the event is reviewed in amanner that favors ascribing blame to thedefendant who evokes the most negativeaffect. This is similar to the assumptionunderlying the tunnel vision induced byinitial impressions of trustworthiness in theDDT model. Another theory based on aframework of attributions of responsibility,the ‘‘intuitive prosecutor’’ (Tetlock et al.,2007), explores how individuals come tomake character attributions for normviolators. According to this framework,we have evolved to be social watchdogsthat punish norm violators and havedeveloped a ‘‘fair-but-biased-yet-correcti-ble’’ (FBC; Tetlock, 2002) method ofdealing with them. The FBC methodsuggests that we see ourselves as fairindividuals, yet we are often biased deci-sion-makers, and when we notice our ownbiases we attempt to self-correct ourcognitions. The self-correction piece ofthis framework has not been adequatelyexplored in a legal decision-making con-text. Research to date has not found aspontaneous end to the tunnel vision thatensues in many legal proceedings. UnlikeDDT, these frameworks are based on thenotion that we have a desire to inferpersonal control and responsibility toothers’ behavior (similar to the ‘‘justworld’’ framework; e.g., Hafer & Begue,2005). DDT simply suggests that, based onevolutionary mechanisms, we make intui-tive judgments of ‘‘friend or foe’’ based onappearance that lead to biased decision-making. Although they differ in a variety ofways, all of these theories suggest thatimplicit cognitions and values influence theinterpretation of defendant informationand lead to generally adverse attributionsof blame. DDT is favorable over compet-ing theories for explaining bias in legaldecision-making for this study because itexplains the isolated effects of appearanceon judicial biases.

While DDT predicts only adverse con-sequences of incorrect implicit evaluations,previous research has also suggested thatimplicit judgments can be accurate. Workby Ambady and colleagues on ‘‘thin slices’’of human behavior has demonstrated thatpeople are able to form accurate impres-sions that predict certain ecologically validoutcomes from brief observations (Amba-dy, 2010), including sexual orientation(Ambady, Hallahan, & Conner, 1999)teaching performance (Ambady & Ro-senthal, 1993) and deception (e.g., Ambady& Rosenthal, 1992). This effect is generallysmall but significant (average d¼ 0.39), andis impeded by tasks that interfere withintuition, such as deliberation or activejudgments of character (Ambady, 2010;Hartwig & Bond, 2011). While thesefindings are important, literature to datethat has examined the accuracy and im-plications of trustworthiness judgments offaces per se has found detrimental effects ofthese (often incorrect) judgments in foren-sic settings.

Research addressing the assessment offactors underlying these initial assessmentsof trustworthiness and specific conse-quences for legal decision-making is criti-cal. In the first study of DDT (Porter et al.,2010), pilot participants rating photos ofCaucasian male faces on several featuresincluding trustworthiness, attractiveness,baby-facedness, facial symmetry and kind-ness. Faces deemed to be the most and leasttrustworthy were then used for the mainstudy. Participants were presented with acase description of a crime, along with aphoto of a trustworthy- or untrustworthy-looking male face. Finally, participantswere provided with 10 increasingly incri-minating pieces of evidence and asked toendorse either a guilty or not guilty verdict(based on a reasonable doubt criterion)following the presentation of each. Partici-pants required less evidence to reach aguilty verdict for defendants with untrust-worthy-looking faces than those with

trustworthy-looking faces. This experimen-tal format provides the template for thepresent study.

The present study examined the effectsof participant biases and attitudes towardsthe legal system, and how they relate to theDDT and legal decision-making. Findingsto date suggest that untrustworthy-lookingfaces put into motion a form of tunnelvision decision-making that exaggeratesthe importance of incriminating evidenceand undervalues that of exculpating infor-mation (Porter & ten Brinke, 2009). How-ever, we think that it is likely that anindividual juror’s beliefs and biases alsomay contribute to this process. To test thishypothesis, we included two measures thatassess attitudes and biases towards thelegal system: the Pretrial Juror AttitudeQuestionnaire (PJAQ; Lecci & Myers,2008) and the Justice–Vengeance Scale(JVS; Ho, ForsterLee, ForsterLee, &Crofts, 2002). The PJAQ and JVS wereselected over other instruments becausethey comprehensively assess factors ofdirect relevance to legal decision-making,and fit within the time constraints of thisstudy. Both questionnaires measure atti-tudes related to the justice system, whichcan affect both the initial assessments oftrustworthiness (e.g., racial bias) and themanner in which evidence is interpreted. Inline with the DDT pattern found by Porteret al. (2010), it was expected that jurorswho endorsed intuition and ‘‘facial profil-ing’’ (e.g., high scorers on subscales of thePJAQ and JVS assessing interpersonalbiases and subjective reasoning) wouldremain confident in the defendant’s guilteven in the presence of exonerating evi-dence. Further, it was expected that parti-cipants who recognized the detrimentaleffects of legal decision-making (e.g.,wrongful convictions), and scored high onmeasures of justice–fairness (and otherscales evaluating unbiased legal attitudes)are more likely to strive for objectiveness intheir decisions.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

388 N. Korva et al.

Pilot Study

A pilot study was conducted to obtainratings on several facial characteristics of aset of photographs of female faces, afterwhich the most and least trustworthy-looking faces were used as defendants inthe main study. Ratings were comparedwith those of male faces used in Porteret al. (2010) to examine the similarity offacial characteristics across gender.

Method

Participants

Participants in the pilot study were 20individuals from a convenience sample (10females, 10 males), with an average age of22.9 years (SD¼ 3.56).

Materials

Twenty photographs of Caucasian femaleswere chosen from the Karolinska set ofphotographs (Lundqvist, Flykt, & Ohman,1998) to be rated on various characteristics.The images depict female faces between theages of 20 and 30 exhibiting neutralexpressions. Twenty male images from thesame image set were previously rated in apilot study for Porter et al. (2010). Parti-cipants viewed the female faces in an onlinecomputer-based experiment.

Procedure

Participants were presented with all 20faces in a computer-generated randomizedmanner, and were asked to rate each faceon attractiveness and trustworthiness on aseven-point scale (1¼ not at all, 7¼ very).Additionally, participants provided anestimate of the age (1¼ 20, 7¼ 50þ),aggressiveness (1¼ not at all, 7¼ very)and likelihood of committing a crime(1¼ not at all, 7¼ very) for each face.Because only a limited set of images wasavailable in the database used for thisstudy, faces were selected from the larger

database based on the researcher’s impres-sions of trustworthiness. Computer-gener-ated faces examining the specific facialfeatures associated with trustworthinesshave already been conducted (e.g., Ooster-hof & Todorov, 2008), but lack the realismand generalizability associated with actualphotos.

Results

Based on average trait inferences made bythe pilot study participants, the most andleast trustworthy-looking faces were cho-sen from the rated images to be used in themain study. The most trustworthy femaleface (M¼ 4.45, SD¼ 1.23) and the leasttrustworthy female face (M¼ 3.00,SD¼ 1.21), were significantly different ontrustworthiness ratings, t(19)¼ 3.51(p5 .01).

Trustworthy-looking faces were consid-ered to be more attractive (t(19)¼ 3.75,p5 .001), less aggressive (t(19)¼ 2.79,p5 .01) and less likely to commit crime(t(19)¼ 2.84, p5 .01) than the perceiveduntrustworthy faces. The two male facestaken from the Porter et al. (2010) studywere also previously reported to also differsignificantly. The most trustworthy maleface (M¼ 4.70, SD¼ 0.99) and least trust-worthy male face (M¼ 2.63, SD¼ 1.45)were rated as significantly different fromone another on trustworthiness, (t(26)¼6.99, p5 .01). Other attributions based onthe face, including attractiveness, baby-facedness, symmetry and kindness alsowere rated as significantly different(p values 4.05). Figure 1 depicts the finalfaces selected for the main part of thisstudy.

Main Study

Method

Participants

Participants in the main study were 98adults (79 female), including both

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 389

Pilot Study

A pilot study was conducted to obtainratings on several facial characteristics of aset of photographs of female faces, afterwhich the most and least trustworthy-looking faces were used as defendants inthe main study. Ratings were comparedwith those of male faces used in Porteret al. (2010) to examine the similarity offacial characteristics across gender.

Method

Participants

Participants in the pilot study were 20individuals from a convenience sample (10females, 10 males), with an average age of22.9 years (SD¼ 3.56).

Materials

Twenty photographs of Caucasian femaleswere chosen from the Karolinska set ofphotographs (Lundqvist, Flykt, & Ohman,1998) to be rated on various characteristics.The images depict female faces between theages of 20 and 30 exhibiting neutralexpressions. Twenty male images from thesame image set were previously rated in apilot study for Porter et al. (2010). Parti-cipants viewed the female faces in an onlinecomputer-based experiment.

Procedure

Participants were presented with all 20faces in a computer-generated randomizedmanner, and were asked to rate each faceon attractiveness and trustworthiness on aseven-point scale (1¼ not at all, 7¼ very).Additionally, participants provided anestimate of the age (1¼ 20, 7¼ 50þ),aggressiveness (1¼ not at all, 7¼ very)and likelihood of committing a crime(1¼ not at all, 7¼ very) for each face.Because only a limited set of images wasavailable in the database used for thisstudy, faces were selected from the larger

database based on the researcher’s impres-sions of trustworthiness. Computer-gener-ated faces examining the specific facialfeatures associated with trustworthinesshave already been conducted (e.g., Ooster-hof & Todorov, 2008), but lack the realismand generalizability associated with actualphotos.

Results

Based on average trait inferences made bythe pilot study participants, the most andleast trustworthy-looking faces were cho-sen from the rated images to be used in themain study. The most trustworthy femaleface (M¼ 4.45, SD¼ 1.23) and the leasttrustworthy female face (M¼ 3.00,SD¼ 1.21), were significantly different ontrustworthiness ratings, t(19)¼ 3.51(p5 .01).

Trustworthy-looking faces were consid-ered to be more attractive (t(19)¼ 3.75,p5 .001), less aggressive (t(19)¼ 2.79,p5 .01) and less likely to commit crime(t(19)¼ 2.84, p5 .01) than the perceiveduntrustworthy faces. The two male facestaken from the Porter et al. (2010) studywere also previously reported to also differsignificantly. The most trustworthy maleface (M¼ 4.70, SD¼ 0.99) and least trust-worthy male face (M¼ 2.63, SD¼ 1.45)were rated as significantly different fromone another on trustworthiness, (t(26)¼6.99, p5 .01). Other attributions based onthe face, including attractiveness, baby-facedness, symmetry and kindness alsowere rated as significantly different(p values 4.05). Figure 1 depicts the finalfaces selected for the main part of thisstudy.

Main Study

Method

Participants

Participants in the main study were 98adults (79 female), including both

university students and other individualsfrom western Canada, with an average ageof 21.27 years (SD¼ 4.64). Participants

were offered a choice of one classcredit or $10 as compensation forparticipating.

Figure 1. Depiction of face stimuli used in main study. Images 1 and 2 depict the trustworthy maleand female faces. Images 3 and 4 depict the untrustworthy male and female faces.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

390 N. Korva et al.

Materials

All participants viewed and rated the entireset of 20 female images used in the pilotstudy on trustworthiness, attractiveness,etc. Because of the number of evaluations,participants were asked to provide for eachimage, and the number of images, it isunlikely that this process made our hy-potheses transparent or adversely affectedthe validity of our results. This is sub-stantiated by results found by Porter et al.(2010), who assigned only half of theirparticipants to provide similar pre-ratingsand found no effect of this prior exposureon subsequent legal decision-making. Thisre-evaluation of images by all participantsserved to confirm that the images weselected were consistently rated as verytrustworthy and very untrustworthy.

The female photographs that had beenrated as being the lowest and highest intrustworthiness from the pilot study wereselected for use in the main study, alongwith the most and least trustworthy malephotographs from Porter et al.’s (2010)initial study. Two serious crime vignettes(robbery and murder; violent attack result-ing in murder), and two less serious crimevignettes (car theft; fraud), as well as 11pieces of evidence for each crime, selectedfor Porter et al.’s (2010) study wereutilized.

Two questionnaires were used in ourmain study to assess participant attitudesand biases towards the legal system. Thesequestionnaires were presented upon com-pletion of all other study components so asto not bias participants and to keep thepremise of our experiment from being tootransparent. While it is possible thatevaluating evidence in a case prior tocompleting these questionnaires may havesome effect on scores, the researchersbelieve this is still an important constructto examine. The PJAQ (Lecci & Myers,2008) identifies individual differences inlegal attitudes that assess how trial infor-mation is processed in legal decision-

making. Lecci and Myers (2008) suggestthat there are certain biases and beliefs thatinfluence how final legal judgments arereached, and the PJAQ targets thesespecific attitudes to investigate the effectsthey have on legal decisions. This 29-itemquestionnaire includes the following sub-scales: conviction proneness, system con-fidence, cynicism toward the defense, racialbias, social justice and innate criminality.The presence of these six subscales hasbeen established by means of cross-valida-tion (i.e., all fit indices are above .90) onsamples summing 600 participants (Lecci &Myers, 2008). The JVS (Ho et al., 2002)assesses two predominant motives (justiceand vengeance) that may be aroused by thelegal system. Ho et al. (2002) suggest that itis difficult to establish whether decisionsare guided by justice or vengeance motives.Thus, the subscales are designed to inves-tigate the two motives separately anddistinguish between their influences onlegal decisions. The JVS is a 16-itemquestionnaire and includes four subscales:vengeance–sentence, vengeance–emotion,justice–fairness and justice–legal. Thesefour subscales have shown to be reliable(i.e., Cronbach’s alpha for each subscaleis at least .70), and have been cross-validated (i.e., all fit indices are close toor above .9) in two samples totaling over500 participants.

Procedure

Before commencing with the experiment,all participants were asked to rate the pilotstudy images on various attributes, includ-ing trustworthiness and attractiveness.

Participants were randomly assigned toone of two crime seriousness conditions:Serious (murder) or Less Serious (fraud,theft), a between-subjects variable. In addi-tion, participants were randomly assignedinto one of two defendant gender condi-tions (male, female), also a between-sub-jects variable. Defendant trustworthiness

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 391

Materials

All participants viewed and rated the entireset of 20 female images used in the pilotstudy on trustworthiness, attractiveness,etc. Because of the number of evaluations,participants were asked to provide for eachimage, and the number of images, it isunlikely that this process made our hy-potheses transparent or adversely affectedthe validity of our results. This is sub-stantiated by results found by Porter et al.(2010), who assigned only half of theirparticipants to provide similar pre-ratingsand found no effect of this prior exposureon subsequent legal decision-making. Thisre-evaluation of images by all participantsserved to confirm that the images weselected were consistently rated as verytrustworthy and very untrustworthy.

The female photographs that had beenrated as being the lowest and highest intrustworthiness from the pilot study wereselected for use in the main study, alongwith the most and least trustworthy malephotographs from Porter et al.’s (2010)initial study. Two serious crime vignettes(robbery and murder; violent attack result-ing in murder), and two less serious crimevignettes (car theft; fraud), as well as 11pieces of evidence for each crime, selectedfor Porter et al.’s (2010) study wereutilized.

Two questionnaires were used in ourmain study to assess participant attitudesand biases towards the legal system. Thesequestionnaires were presented upon com-pletion of all other study components so asto not bias participants and to keep thepremise of our experiment from being tootransparent. While it is possible thatevaluating evidence in a case prior tocompleting these questionnaires may havesome effect on scores, the researchersbelieve this is still an important constructto examine. The PJAQ (Lecci & Myers,2008) identifies individual differences inlegal attitudes that assess how trial infor-mation is processed in legal decision-

making. Lecci and Myers (2008) suggestthat there are certain biases and beliefs thatinfluence how final legal judgments arereached, and the PJAQ targets thesespecific attitudes to investigate the effectsthey have on legal decisions. This 29-itemquestionnaire includes the following sub-scales: conviction proneness, system con-fidence, cynicism toward the defense, racialbias, social justice and innate criminality.The presence of these six subscales hasbeen established by means of cross-valida-tion (i.e., all fit indices are above .90) onsamples summing 600 participants (Lecci &Myers, 2008). The JVS (Ho et al., 2002)assesses two predominant motives (justiceand vengeance) that may be aroused by thelegal system. Ho et al. (2002) suggest that itis difficult to establish whether decisionsare guided by justice or vengeance motives.Thus, the subscales are designed to inves-tigate the two motives separately anddistinguish between their influences onlegal decisions. The JVS is a 16-itemquestionnaire and includes four subscales:vengeance–sentence, vengeance–emotion,justice–fairness and justice–legal. Thesefour subscales have shown to be reliable(i.e., Cronbach’s alpha for each subscaleis at least .70), and have been cross-validated (i.e., all fit indices are close toor above .9) in two samples totaling over500 participants.

Procedure

Before commencing with the experiment,all participants were asked to rate the pilotstudy images on various attributes, includ-ing trustworthiness and attractiveness.

Participants were randomly assigned toone of two crime seriousness conditions:Serious (murder) or Less Serious (fraud,theft), a between-subjects variable. In addi-tion, participants were randomly assignedinto one of two defendant gender condi-tions (male, female), also a between-sub-jects variable. Defendant trustworthiness

was a within-subjects variable, such thateach participant viewed one trustworthyand untrustworthy-looking face. Afterparticipants were randomly assigned intothese conditions, they were presented withone of two counterbalanced crime scenar-ios (according to their condition) in whichone of the photographed defendants (ran-domly assigned) had been tried for thecrime (construed as being real) in a foreignjurisdiction.

Before presentation of the evidence, thelegal concept of reasonable doubt wasdisplayed on the computer screen to theparticipants who were to act as jurors, andthey were asked to apply it in theirdecision-making. Participants then werepresented with the pre-arranged evidenceset corresponding to their case file. Eachcase included 11 pieces of evidence; the firstfive were ambiguous (e.g., ‘‘John was in thesame city as the crime on the day of themurder’’); the next five were increasinglyincriminating (e.g., ‘‘The accused hadscratches on his hands and arms’’); andthe final piece of evidence offered wasthoroughly exonerating (e.g., ‘‘DNA evi-dence from underneath one of the victim’sfingernails did not match the accused’’).After each piece of evidence was presented,the participant was instructed to select averdict of not guilty or guilty beyond areasonable doubt. They also were asked torate their confidence in the verdict. Next,the second case and photograph werepresented in the same manner.

Participants then were asked to fill outthe two questionnaires that examine theirattitudes and biases towards the legalsystem.

Results

A manipulation check was conducted toconfirm that the selected photographs wererated similarly on trustworthiness in themain and pilot studies. Paired samples t-tests were conducted to ensure that

trustworthy and untrustworthy-lookingfaces were rated significantly differently.There were significant differences betweenthe trustworthy and untrustworthy femalefaces, t(97)¼ 3.20, p5 .01, and betweenthe trustworthy and untrustworthy malefaces, t(97)¼ 3.53, p5 .01. Further, asdesired, there was no overall significantdifference in the trustworthiness ratings ofmale and female faces (for the two mostand least trustworthy photos). We alsoexamined whether there was an interactionbetween trustworthiness and sex in theprediction of trustworthiness ratings. Ananalysis of variance involving one within-subjects factor (high vs. low trust) and onebetween-subjects factor (sex) revealed nosignificant interaction effect, F(1,45)¼ 0.78,p¼ .781.

Independent samples t-tests were thenconducted to determine whether the find-ings from the main study and the pilotstudy converged. There were no significantdifferences between the main study andpilot study in the trustworthiness ratings oftrustworthy and untrustworthy male faces,or in the ratings of trustworthy femalefaces. However, untrustworthy femalefaces were rated less trustworthy in themain study than in the pilot study,t(97)¼ 2.61, p5 .05. All of these weredesirable findings that indicate strongmanipulation of the variables of interest.Crime seriousness had no effect on verdictsfor trustworthy or untrustworthy defen-dants and so these conditions (severe andless severe crimes) were collapsed.

Pearson correlations were computed toexplore the relationships between question-naire subscales and decision-making whenpresented with exonerating evidence fortrustworthy and untrustworthy-lookingdefendants (see Table 1).1 When partici-pants were presented with an untrust-worthy-looking defendant, PJAQ systemconfidence, racial bias and JVS vengeance–emotion scores were positively correlatedwith guilty verdicts, regardless of

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

392 N. Korva et al.

exonerating evidence to the contrary, pvalues 5 .05. By contrast, JVS justice–fairness scores were negatively correlatedwith guilty verdicts following exoneratingevidence for untrustworthy-looking defen-dants, p5 .05. The other subscales of theJVS and PJAQ were not significantlycorrelated with legal decision-making foruntrustworthy faces. When participantswere presented with a trustworthy-lookingdefendant, there were no significant rela-tionships between the questionnaire sub-scales and verdict choice after presentationof exonerating evidence.

The Generalized Estimating Equations(GEE) procedure in STATA 12 (www.stata.com/) was then used to conductbinary logistic regressions predicting ver-dict (innocent or guilty) after exoneratingevidence was presented. Defendant trust-worthiness was a within-subjects predictorvariable, and scores on the PJAQ and JVSwere used as potential moderators of theinfluence of defendant trustworthiness onverdict. We explored the nature of signifi-cant interactions for these variables by

graphing the logistic regression equationsfor the two levels of trustworthiness inrelation to the continuous PJAQ and JVSmoderator variables. The margins proce-dure in STATA 12 was used to assesssimple main effects and the points alongmoderator variable continuums for whichthere were significant differences betweenthe high and low trustworthinessconditions.

The first GEE binary logistic regres-sion involved defendant trustworthiness,PJAQ racial bias, and their interaction aspredictors of verdict. The effect fordefendant trustworthiness did not reachsignificance, p4.05, but there was asignificant effect for PJAQ racial bias,b¼7.18, w2(N¼ 86)¼ 3.79, p¼ .05, oddsratio (OR)¼ .83. Higher scores on thePJAQ racial bias subscale were associatedwith an increased probability of convict-ing the defendant.

There also was a significant interactionbetween racial bias and defendant trust-worthiness in the prediction of verdict,b¼ .29, w2(N¼ 86)¼ 4.28, p5 .05, OR¼1.34. This interaction indicates that theslopes for racial bias in the prediction ofverdict were significantly different for thehigh versus low trustworthiness conditions.A plot of the interaction appears in Figure2. When defendant trustworthiness waslow, the effects for racial bias weredramatic and the simple slope was signifi-cant, z¼ 2.25, p¼ .024. Higher scores onracial bias were associated with higherprobabilities of a guilty verdict. In fact,the highest probability of a guilty verdictoccurred when racially biased participantsevaluated a defendant with a low trust-worthiness face. By contrast, when defen-dant trustworthiness was high, the simpleslope for racial bias was negative, but notstatistically significant, z¼ 1.41, p¼ .16. Inother words, for defendants with trust-worthy faces, there was a non-significanttendency for higher scores on racial bias tobe associated with lower guilty verdict

Table 1. Correlations (r) between PJAQ andJVS subscales and verdict decisions (0¼ notguilty/exonerate, 1¼ guilty) after presentation ofexonerating evidence.

Trustworthydefendant

Untrustworthydefendant

PJAQSystem Confidence .15 .21*Conviction

Proneness.15 .16

Cynicism Towardthe Defense

.00 .13

Racial Bias 7.10 .25*Social Justice .10 7.18JVSVengeance–

Emotion7.05 .25*

Vengeance–Sentence

7.01 .16

Justice–Fairness .05 7.29**Justice–Legal 7.02 7.08

*p5 .05; **p5 .01.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 393

exonerating evidence to the contrary, pvalues 5 .05. By contrast, JVS justice–fairness scores were negatively correlatedwith guilty verdicts following exoneratingevidence for untrustworthy-looking defen-dants, p5 .05. The other subscales of theJVS and PJAQ were not significantlycorrelated with legal decision-making foruntrustworthy faces. When participantswere presented with a trustworthy-lookingdefendant, there were no significant rela-tionships between the questionnaire sub-scales and verdict choice after presentationof exonerating evidence.

The Generalized Estimating Equations(GEE) procedure in STATA 12 (www.stata.com/) was then used to conductbinary logistic regressions predicting ver-dict (innocent or guilty) after exoneratingevidence was presented. Defendant trust-worthiness was a within-subjects predictorvariable, and scores on the PJAQ and JVSwere used as potential moderators of theinfluence of defendant trustworthiness onverdict. We explored the nature of signifi-cant interactions for these variables by

graphing the logistic regression equationsfor the two levels of trustworthiness inrelation to the continuous PJAQ and JVSmoderator variables. The margins proce-dure in STATA 12 was used to assesssimple main effects and the points alongmoderator variable continuums for whichthere were significant differences betweenthe high and low trustworthinessconditions.

The first GEE binary logistic regres-sion involved defendant trustworthiness,PJAQ racial bias, and their interaction aspredictors of verdict. The effect fordefendant trustworthiness did not reachsignificance, p4.05, but there was asignificant effect for PJAQ racial bias,b¼7.18, w2(N¼ 86)¼ 3.79, p¼ .05, oddsratio (OR)¼ .83. Higher scores on thePJAQ racial bias subscale were associatedwith an increased probability of convict-ing the defendant.

There also was a significant interactionbetween racial bias and defendant trust-worthiness in the prediction of verdict,b¼ .29, w2(N¼ 86)¼ 4.28, p5 .05, OR¼1.34. This interaction indicates that theslopes for racial bias in the prediction ofverdict were significantly different for thehigh versus low trustworthiness conditions.A plot of the interaction appears in Figure2. When defendant trustworthiness waslow, the effects for racial bias weredramatic and the simple slope was signifi-cant, z¼ 2.25, p¼ .024. Higher scores onracial bias were associated with higherprobabilities of a guilty verdict. In fact,the highest probability of a guilty verdictoccurred when racially biased participantsevaluated a defendant with a low trust-worthiness face. By contrast, when defen-dant trustworthiness was high, the simpleslope for racial bias was negative, but notstatistically significant, z¼ 1.41, p¼ .16. Inother words, for defendants with trust-worthy faces, there was a non-significanttendency for higher scores on racial bias tobe associated with lower guilty verdict

Table 1. Correlations (r) between PJAQ andJVS subscales and verdict decisions (0¼ notguilty/exonerate, 1¼ guilty) after presentation ofexonerating evidence.

Trustworthydefendant

Untrustworthydefendant

PJAQSystem Confidence .15 .21*Conviction

Proneness.15 .16

Cynicism Towardthe Defense

.00 .13

Racial Bias 7.10 .25*Social Justice .10 7.18JVSVengeance–

Emotion7.05 .25*

Vengeance–Sentence

7.01 .16

Justice–Fairness .05 7.29**Justice–Legal 7.02 7.08

*p5 .05; **p5 .01.

probabilities, and for low scores on racialbias to be associated with higher guiltyverdict probabilities. The STATA 12 mar-gins procedure revealed that the discrepan-cies between the two logistic regressionlines in Figure 2 were statistically signifi-cant at or above PJAQ racial bias scores of12.7 and at or below PJAQ racial biasscores of 4.15.

The second GEE binary logistic regres-sion involved defendant trustworthiness,JVS justice–fairness, and their interactionas predictors of verdict. In this case,untrustworthy defendants were more likelyto be found guilty after exonerating evi-dence than were trustworthy defendants,b¼ 4.98, w2(N¼ 86)¼ 4.18, p5 .05, OR¼145.68. High scores on the JVS justice–fairness scale were associated with fewerguilty verdicts than were low scores,

b¼ .177, w2(N¼ 86)¼ 6.26, p5 .05,OR¼ 1.19. There also was a significantinteraction between defendant trustworthi-ness and JVS in the prediction of verdict,b¼7.226, w2(N¼ 86)¼ 3.94, p5 .05,OR¼ .80. A plot of this interaction ap-pears in Figure 3. Once again, the biggesteffect (steepest slope) occurred in the lowtrust condition, z¼ 2.61, p¼ .009. Partici-pants who scored low on justice–fairnesswere the persons most likely to provideguilty verdicts. Participants who scoredhigh on justice–fairness were much lesslikely to provide guilty verdicts whenevaluating the same faces. The simple slopefor JVS justice–fairness and verdict wasweak and not significant in the high trustcondition, z¼ 0.65, p¼ .514. The discre-pancies between the two logistic regressionlines in Figure 3 were statistically

Figure 2. Trust and PJAQ racial bias predicting the probability of a guilty verdict afterpresentation of exonerating evidence.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

394 N. Korva et al.

significant at or below JVS justice–fairnessscores of 17.

The Pearson correlation between PJAQracial bias and JVS justice–fairness was7.33, p¼ .002, indicating some overlapbetween the two attitude variables. Theabove GEE analyses were therefore bothrun again to determine whether the effectsfor PJAQ racial bias remained significantafter controlling for JVS justice–fairness,and to determine whether the effects forJVS justice–fairness remained significantafter controlling for PJAQ racial bias. Theanswer was ‘‘yes’’ in both cases. Thepattern of significant and non-significanteffects did not change for these covariateanalyses. Finally, GEE binary logisticregression analyses for the JVS systemconfidence and vengeance–emotion

subscales did not reveal significant interac-tions with defendant trustworthiness.

Discussion

As predicted by the DDT, participantsattributed more positive traits to trust-worthy faces relative to untrustworthyfaces. Trustworthy faces were consideredto be more attractive, kinder, less aggres-sive, and less likely to commit crimes. Inaddition, individuals endorsing justice–fairness were more likely to exonerate aperceived untrustworthy defendant. Incontrast, more racially biased participantswere less likely to be influenced by exon-erating evidence for an untrustworthy-looking defendant. It is also important tonote that although there was some overlap

Figure 3. Trust and JVS justice–fairness predicting the probability of a guilty verdict afterpresentation of exonerating evidence.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 395

significant at or below JVS justice–fairnessscores of 17.

The Pearson correlation between PJAQracial bias and JVS justice–fairness was7.33, p¼ .002, indicating some overlapbetween the two attitude variables. Theabove GEE analyses were therefore bothrun again to determine whether the effectsfor PJAQ racial bias remained significantafter controlling for JVS justice–fairness,and to determine whether the effects forJVS justice–fairness remained significantafter controlling for PJAQ racial bias. Theanswer was ‘‘yes’’ in both cases. Thepattern of significant and non-significanteffects did not change for these covariateanalyses. Finally, GEE binary logisticregression analyses for the JVS systemconfidence and vengeance–emotion

subscales did not reveal significant interac-tions with defendant trustworthiness.

Discussion

As predicted by the DDT, participantsattributed more positive traits to trust-worthy faces relative to untrustworthyfaces. Trustworthy faces were consideredto be more attractive, kinder, less aggres-sive, and less likely to commit crimes. Inaddition, individuals endorsing justice–fairness were more likely to exonerate aperceived untrustworthy defendant. Incontrast, more racially biased participantswere less likely to be influenced by exon-erating evidence for an untrustworthy-looking defendant. It is also important tonote that although there was some overlap

Figure 3. Trust and JVS justice–fairness predicting the probability of a guilty verdict afterpresentation of exonerating evidence.

between our racial bias and justice–fairnessvariables, the effects for each variablesremained the same when statistically con-trolling for the other variable. We havethus found two variables with independenteffects on legal decision-making, and wesuspect that other attitude variables may bediscovered in further research.

While the observed effects were statis-tically significant, they were not large andpractical applications of the findingsshould be approached with caution. Ourresearch indicates a detrimental effect oftrustworthiness evaluations on legal deci-sion-making. But our findings regardingwhen, and for whom, these effects arestrongest should be replicated and exploredin further research. The results are never-theless in accordance with previous workon tunnel vision, and they suggest thatuntrustworthy faces activate personalbiases related to the importance of appear-ance-based assessments in legal decisions.The subsequent tunnel vision reduces thelikelihood that exonerating evidence will beconsidered in evaluations of guilt orinnocence (Vrij, 2008), which ultimatelymay contribute to wrongful convictions.We suspect that the effect sizes that wereobserved in the present laboratory, on-paper study are likely underestimates of theeffect sizes for these same variables thatoccur for real-world crimes and defen-dants, where more is on the line andeverything is more salient.

Previous research has found that de-fendant race is a strong predictor of legaldecision-making (e.g., Sweeney & Haney,1992). For example, Eberhardt et al.(2006) found that males rated higher instereotypical blackness were more likely tobe sentenced to death for a murdercompared with less stereotypically Blackdefendants. Here, racial bias was a strongnegative predictor of exoneration whenpresented with an untrustworthy-lookingdefendant. Participants obtaining highscores on racial bias were less likely to

render a verdict of innocence in generalafter being presented with exoneratingevidence, despite the fact that all defen-dants were Caucasian. Racial bias appar-ently encouraged tunnel vision decision-making, biasing decisions regarding un-trustworthy-looking defendants. This ef-fect is presumed to be associated with ageneral reliance on ‘‘gut instinct’’, stereo-typing and a tendency to ‘‘judge a bookby its cover’’ among racially biasedindividuals. Lecci and Myers (2008) foundthat the PJAQ racial bias scale was aconsistent predictor even when race orother background was not specified.Therefore, they suggested that the racialbias scale is related to a more generaltendency to make (unfounded) inferencesabout defendants. It is also possible thatthe untrustworthy faces appeared raciallydifferent to racially biased persons, eventhough the skin color of the faces waswhite (see the faces in Figure 1). Incontrast, individuals who scored higheron the justice–fairness subscale of the JVSwere more likely to exonerate perceiveduntrustworthy-looking defendants. Giventhat this scale measures objectivity in legaldecisions, it was expected that individualswho endorsed these unbiased and objec-tive attitudes would be more likely toexonerate perceived untrustworthy faces.However, people more often exonerateddefendants when they were perceived asuntrustworthy compared with trust-worthy, suggesting that individuals whostrive for objectivity may overcompensatein an attempt to make a fair and unbiaseddecision. These individuals may be overlyconscious of the need for objectivity inlegal decisions, which may be potentiallybiasing their verdicts in the oppositedirection.

More generally, the DDT model wassupported by the main effect of trust-worthiness. Trustworthy defendants werefar more likely to be exonerated thanuntrustworthy defendants when variance

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

396 N. Korva et al.

associated with personal beliefs (i.e., jus-tice–fairness) had been accounted for.Further, trustworthy-looking defendantsalso were rated highly on various positivetraits (e.g., kind and unlikely to commitcrime), while untrustworthy-looking indi-viduals were assigned predominantly nega-tive traits (e.g., unkind and likely tocommit crime). In general, the resultssuggested that character evaluations, basedin part on facial appearance, interact withvarious biases and attitudes to shape themanner in which evidence of an individualis assessed.

Limitations of this study include the useof self-reports. Both the PJAQ and the JVSare self-report questionnaires that could besusceptible to underestimates or overesti-mates on various items. For example,certain questions pertain to the importanceof being objective when making legaldecisions or sentencing a guilty defendant.Such questions may be subject to socialdesirability biases, although this may bemitigated by the anonymity and confiden-tiality assured to participants. Alterna-tively, implicit measures of racial biasmay measure the construct more accu-rately, as it is based on reactions ofparticipants that are difficult to control(Klauer & Teige-Mocigemba, 2007). An-other potential limitation to this study wasthe use of still photographs and writtendescriptions of the evidence. To enhanceecological validity, further research couldutilize a videotaped trial to imitate a morerealistic legal scenario. Further, the exclu-sive use of Caucasian defendants in thisstudy should be extended by includingvariations of defendant race in futurestudies, especially given that we foundthat a strong impact of racial bias on legaldecision-making.

The research presented here supportsthe DDT and the notion that legal deci-sion-makers formulate implicit assump-tions regarding defendant culpabilitybased on facial appearance. These findings

have important implications for under-standing errors that happen in criminaltrials and manner in which false convic-tions might occur.

Note

1. Potential defendant gender differences werealso explored; however, no significant re-lationships emerged.

References

Alicke, M.D. (2000). Culpable control and thepsychology of blame. Psychological Bulletin,126, 556–574, doi:10.1037/0033-2909.126.4.556.

Alicke, M.D., Buckingham, J., Zell, E., &Davis, T. (2008). Culpable control andcounterfactual reasoning in the psychologyof blame. Personality and Social PsychologyBulletin, 34, 1371–1381, doi: 10.1177/0146167208321594.

Ambady, N. (2010). The perils of pondering:Intuition and thin slice judgments. Psycho-logical Inquiry, 21, 271–278, doi: 10.1080/1047840X.2010.524882.

Ambady,N., Hallahan,M., &Conner, B. (1999).Accuracy of judgments of sexual orientationfrom thin slices of behavior. Journal ofPersonality and Social Psychology, 77, 538–547, doi: 10.1037/0022-3514.77.3.538.

Ambady, N., & Rosenthal, R. (1992). Thinslices of expressive behavior as predictors ofinterpersonal consequences: A meta-analy-sis. Psychological Bulletin, 111, 256–274,doi:10.1037/0033-2909.111.2.256.

Ambady, N., & Rosenthal, R. (1993). Half aminute: Predicting teacher evaluations fromthin slices of nonverbal behavior andphysical attractiveness. Journal of Person-ality and Social Psychology, 64, 431–441,doi: 10.1037/0022-3514.64.3.431.

Ask, K., & Granhag, P.A. (2007). Motivationalbias in criminal investigators’ judgments ofwitness reliability. Journal of Applied SocialPsychology, 37, 561–591.

Bar, M., Neta, M., & Linz, H. (2006). Veryfirst impressions. Emotion, 6, 269–278, doi:10.1037/1528-3542.6.2.269

Berry, D., & McArthur, L. (1986). Perceivingcharacter in faces: The impact of age-relatedcraniofacial changes on social perception.Psychological Bulletin, 100, 3–18, doi: 10.1037/0033-2909.100.1.3.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

Juror Attitudes and Defendant Appearance 397

associated with personal beliefs (i.e., jus-tice–fairness) had been accounted for.Further, trustworthy-looking defendantsalso were rated highly on various positivetraits (e.g., kind and unlikely to commitcrime), while untrustworthy-looking indi-viduals were assigned predominantly nega-tive traits (e.g., unkind and likely tocommit crime). In general, the resultssuggested that character evaluations, basedin part on facial appearance, interact withvarious biases and attitudes to shape themanner in which evidence of an individualis assessed.

Limitations of this study include the useof self-reports. Both the PJAQ and the JVSare self-report questionnaires that could besusceptible to underestimates or overesti-mates on various items. For example,certain questions pertain to the importanceof being objective when making legaldecisions or sentencing a guilty defendant.Such questions may be subject to socialdesirability biases, although this may bemitigated by the anonymity and confiden-tiality assured to participants. Alterna-tively, implicit measures of racial biasmay measure the construct more accu-rately, as it is based on reactions ofparticipants that are difficult to control(Klauer & Teige-Mocigemba, 2007). An-other potential limitation to this study wasthe use of still photographs and writtendescriptions of the evidence. To enhanceecological validity, further research couldutilize a videotaped trial to imitate a morerealistic legal scenario. Further, the exclu-sive use of Caucasian defendants in thisstudy should be extended by includingvariations of defendant race in futurestudies, especially given that we foundthat a strong impact of racial bias on legaldecision-making.

The research presented here supportsthe DDT and the notion that legal deci-sion-makers formulate implicit assump-tions regarding defendant culpabilitybased on facial appearance. These findings

have important implications for under-standing errors that happen in criminaltrials and manner in which false convic-tions might occur.

Note

1. Potential defendant gender differences werealso explored; however, no significant re-lationships emerged.

References

Alicke, M.D. (2000). Culpable control and thepsychology of blame. Psychological Bulletin,126, 556–574, doi:10.1037/0033-2909.126.4.556.

Alicke, M.D., Buckingham, J., Zell, E., &Davis, T. (2008). Culpable control andcounterfactual reasoning in the psychologyof blame. Personality and Social PsychologyBulletin, 34, 1371–1381, doi: 10.1177/0146167208321594.

Ambady, N. (2010). The perils of pondering:Intuition and thin slice judgments. Psycho-logical Inquiry, 21, 271–278, doi: 10.1080/1047840X.2010.524882.

Ambady,N., Hallahan,M., &Conner, B. (1999).Accuracy of judgments of sexual orientationfrom thin slices of behavior. Journal ofPersonality and Social Psychology, 77, 538–547, doi: 10.1037/0022-3514.77.3.538.

Ambady, N., & Rosenthal, R. (1992). Thinslices of expressive behavior as predictors ofinterpersonal consequences: A meta-analy-sis. Psychological Bulletin, 111, 256–274,doi:10.1037/0033-2909.111.2.256.

Ambady, N., & Rosenthal, R. (1993). Half aminute: Predicting teacher evaluations fromthin slices of nonverbal behavior andphysical attractiveness. Journal of Person-ality and Social Psychology, 64, 431–441,doi: 10.1037/0022-3514.64.3.431.

Ask, K., & Granhag, P.A. (2007). Motivationalbias in criminal investigators’ judgments ofwitness reliability. Journal of Applied SocialPsychology, 37, 561–591.

Bar, M., Neta, M., & Linz, H. (2006). Veryfirst impressions. Emotion, 6, 269–278, doi:10.1037/1528-3542.6.2.269

Berry, D., & McArthur, L. (1986). Perceivingcharacter in faces: The impact of age-relatedcraniofacial changes on social perception.Psychological Bulletin, 100, 3–18, doi: 10.1037/0033-2909.100.1.3.

Berry, D., & Zebrowitz-McArthur, L. (1988).What’s in a face? Facial maturity and theattributionof legal responsibility.Personalityand Social Psychology Bulletin, 14, 23–33,doi: 10.1177/0146167288141003.

Bond, C.F., Berry, D.S., & Omar, A. (1994).The kernel of truth in judgments of decep-tiveness. Basic and Applied Social Psychol-ogy, 15, 523–534, doi:10.1207/s15324834basp1504_8.

Bull, R. (2006). Detecting lies and deceit: Thepsychology of lying and the implicationsfor professional practice. Journal of Com-munity & Applied Social Psychology, 16,166–167.

Bull, R., & McAlpine, S. (1998). Facial appear-ance and criminality. In A. Memon, A. Vrij& R. Bull (Eds.), Psychology and law.Truthfulness, accuracy and credibility (pp.59–76). London: McGraw-Hill.

Bull, R., & Vine, M. (2003). Attractive peopletell the truth: Can you believe it? Posterpresented at the Annual Conference of theEuropean Association of Psychology andLaw, Edinburgh, UK.

Dumas, R., & Teste, B. (2006). The in?uence ofcriminal facial stereotypes on juridic judg-ments. Swiss Journal of Psychology, 65, 237–244.

Eberhardt, J.L., Davies, P.G., Purdie-Vaughns,V.J., & Johnson, S. (2006). Looking death-worthy: Perceived stereotypicality of blackdefendants predicts capital-sentencing out-comes. Psychological Science, 17, 383–386,doi:10.1111/j.1467-9280.2006.01716.x

Hafer, C.L., & Begue, L. (2005). Experimentalresearch on just-world theory: problems,developments, and future challenges. Psy-chological Bulletin, 131, 128–167, doi: 10.1037/0033-2909.131.1.128.

Hartwig, M., & Bond, C.R. (2011). Why do lie-catchers fail? A lens model meta-analysis ofhuman lie judgments. Psychological Bulletin,137, 643–659, doi:10.1037/a0023589.

Ho, R., ForsterLee, L., ForsterLee, R., &Crofts, N. (2002). Justice versus vengeance:Motives underlying punitive judgements.Personality and Individual Differences, 33,365–378, doi:10.1016/S0191-8869(01)00161-1.

Kahneman, D., & Tversky, A. (1982). Thepsychology of preferences. Scientific Amer-ican, 246, 160–173.

Klauer, K.C., & Teige-Mocigemba, S. (2007).Controllability and resource dependence inautomatic evaluation. Journal of Experi-mental Social Psychology, 43, 648–655.

Kulka, R., & Kessler, J. (1978). Is justice reallyblind? The influence of litigant physicalattractiveness on juridical judgment. Journalof Applied Social Psychology, 8, 366–381,doi: 10.1111/j.1559-1816.1978.tb00790.x.

Lecci, L., & Myers, B. (2008). Individualdifferences in attitudes relevant to jurordecision making: Development and valida-tion of the Pretrial Juror Attitude Ques-tionnaire (PJAQ). Journal of Applied SocialPsychology, 38, 2010–2038, doi:10.1111/j.1559-1816.2008.00378.x.

Lundqvist, D., Flykt, A., & Ohman, A. (1998).The Karolinska Directed Emotional Faces.Stockholm: Karolinska Institute, Psychol-ogy Section, Department of ClinicalNeuroscience.

Martelli, M., Majib, J.M., & Pelli, D.G. (2005).Are faces processed like words? A diagnos-tic test for recognition by parts. Journal ofVision, 5, 58–70.

Olivola, C.Y., & Todorov, A. (2010). Elected in100 milliseconds: Appearance-based traitinferences and voting. Journal of NonverbalBehavior, 34(2), 83–110, doi: 10.1007/s10919-009-0082-1.

Oosterhof, N.N., & Todorov, A. (2008). Thefunctional basis of face evaluation. Proceed-ings of the National Academy of Sciences ofthe United States of America, 105, 11087–11092, doi: 10.1073/pnas.0805664105.

Porter, S., England, L., Juodis, M., & tenBrinke, L. (2008). Is the face a window tothe soul? Investigation of the validity ofintuitive judgments of the trustworthiness ofhuman faces. Canadian Journal of Beha-vioural Science, 40, 171–177, doi:10.1037/0008-400X.40.3.171.

Porter, S., McCabe, S., Woodworth, M., &Peace, K.A. (2007). Genius is 1% inspira-tion and 99% perspiration? Or is it? Aninvestigation of the effects of motivationand feedback on deception detection.Legal and Criminological Psychology, 12,297–309.

Porter, S., & ten Brinke, L. (2009). Dangerousdecisions: A theoretical framework forunderstanding how judges assess credibilityin the courtroom. Legal and CriminologicalPsychology, 14, 119–134, doi: 10.1348/135532508X281520.

Porter, S., ten Brinke, L., & Gustaw, C. (2010).Dangerous decisions: The impact of firstimpressions of trustworthiness on the eva-luation of legal evidence and defendantculpability. Psychology, Crime & Law, 16,477–491, doi: 10.1080/10683160902926141.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3

398 N. Korva et al.

Pyszczynski, T., & Greenberg, J. (1987). To-ward an integration of cognitive and moti-vational perspectives on social inference: Abiased hypothesis-testing model. In L. Ber-kowitz, (Ed.), Advances in experimentalsocial psychology (Vol. 20, pp. 297–340).New York: Academic Press.

Snyder, M., & Stukas, A.A. (1999). Interperso-nal processes: The interplay of cognitive,motivational, and behavioral activities insocial interaction. Annual Review of Psy-chology, 50, 273–303.

Sweeney, L.T., & Haney, C. (1992). The influ-ence of race on sentencing: A meta-analyticreview of experimental studies. BehavioralSciences and the Law, 10, 179–195.

ten Brinke, L. & Porter, S. (2011, in press).Friend or foe? The role of intuition ininterpersonal trustworthiness and vulner-ability assessments. In Columbus, F. (Ed.)Psychology of Intuition. New York: NovaScience Publishers.

Tetlock, P.E. (2002). Social functionalist frame-works for judgment and choice: Intuitivepoliticians, theologians, and prosecutors.Psy-chological Review, 109, 451–471, doi:10.1037/0033-295X.109.3.451

Tetlock, P.E., Visser, P.S., Singh, R., Polifroni,M., Scott, A., Elson, S., Mazzocco, P., &Rescober, P. (2007). People as intuitiveprosecutors: The impact of social-controlgoals on attributions of responsibility.Journal of Experimental Social Psychology,43, 195–209, doi:10.1016/j.jesp.2006.02.009.

Vrij, A. (2008). Detecting lies and deceit: Pitfallsand opportunities (2nd edn). New York:Wiley.

Willis, J., & Todorov, A. (2006). First impres-sions: Making up your mind after a 100-msexposure to a face. Psychological Science,17, 592–598. doi:10.1111/j.1467-9280.2006.01750.x.

Zebrowitz, L.A., Voinescu, L., & Collins, M.A.(1996). ‘Wide-eyed’ and ‘crooked-faced’:Determinants of perceived and real honestyacross the life span. Personality and SocialPsychology Bulletin, 22, 1258–1269, doi:10.1177/01461672962212006.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f B

ritis

h C

olum

bia]

at 0

9:59

22

July

201

3