A test of gender–crime congruency on mock juror decision-making

Abstract The purpose of this study was to investigate whether jurors would be biased in favor of guilt when a defendant’s gender was congruent with stereotypes associated with certain crimes (i.e. a gender–crime congruency effect) and the role of juror gender in informing such an effect. A gender balanced sample (N = 200) of participants read a six-page fabricated grand theft of a motor vehicle or shoplifting trial transcript, in which we manipulated defendant gender. Results did not support the prediction that a woman charged with shoplifting and a man charged with auto theft would yield harsher decisions among same-gender mock jurors. However, there was a significant juror gender by crime-type interaction effect on defendant impressions. For jurors who were women, shoplifting was associated with more positive defendant impressions, with no such effect for men. While this study did not provide evidence of a gender–crime congruency effect, future researchers should consider other crime types and moderator variables.


Introduction
Gender can convey significant behavioral expectations. For instance, society tends to associate masculinity with being strong, aggressive, and dominant, while women may be branded as weak, caring,

PUBLIC INTEREST STATEMENT
Research shows that men are most likely to be charged with personal crimes and women are most likely to be charged with minor property crimes. Gender role expectations could contribute to that difference. In this experiment, participants made decisions in response to a fictional case in which we manipulated the type of crime and whether the defendant was a man or a woman. We expected that a woman charged with shoplifting and a man charged with auto theft would produce harsher decisions among mock jurors who were the same gender as the defendant. Results did not show this pattern. However, jurors who were women had more positive impressions of a defendant in a shoplifting case as compared to an auto theft case; men did not show any such pattern. In sum, gender and crime type did not jointly influence verdict decisions, but researchers should look at patterns for other cases. and passive (Wiest & Duffy, 2013). These distinctions perpetuate beliefs and attitudes that carry social consequences (Glick & Fiske, 1996). In particular, gender is strongly predictive of interaction with the criminal justice system (Farnworth & Teske, 1995). Researchers have shown that men account for 80% of arrests for most crime types (U.S. Department of Justice, 2012). Compared to women, men are five times more likely to be charged with personal crimes (e.g. murder, sexual assault). In contrast, women are most likely to be charged with minor property crimes (e.g. shoplifting; U.S. Department of Justice, 2012). One possible contributor to this difference is that jurors might be biased in favor of guilt when a defendant's gender is congruent with stereotypes associated with certain crimes. This phenomenon, known as crime congruency, has been demonstrated with respect to race, such that defendants possessing characteristics that match those expected of the typical offender receive harsher treatment (Jones & Kaplan, 2003;Maeder, Yamamoto, McManus, & Capaldi, 2016;Skorinko & Spellman, 2013;Sunnafrank & Fontes, 1983). Only a handful of studies have examined the potential for a gender-crime congruency effect; findings indicate that, in general, men might be perceived as fitting crime stereotypes more so than are women (McKimmie, Masters, Masser, Schuller, & Terry, 2013;Strub & McKimmie, 2016). However, for certain crime types, these effects might depend on juror gender (Bagby, Parker, Rector, & Kalemba, 1994;Burke, Ames, Etherington, & Pietsch, 1990;Devine & Caughlin, 2014;Quas, Bottoms, Haegerich, & Nysse-Carris, 2002). The purpose of this study was to investigate the combined effects of juror and defendant gender in two different types of criminal trials: shoplifting and auto theft.

Gender and juror decision-making
Researchers have shown that mock juror judgments may differ as a function of both defendant and juror gender (Devine & Caughlin, 2014;Mazzella & Feingold, 1994). In terms of defendant gender, many studies have focused on sexual assault, predominantly demonstrating leniency for women as compared to men (Cramer, 1999;Henning & Feder, 2005;McCoy & Gray, 2007;Pozzulo, Dempsey, Maeder, & Allen, 2010). Specifically, mock jurors tend to perceive the defendant as more culpable and the victim as more credible when the defendant is a man (Rogers & Davies, 2007). However, findings from major meta-analyses are mixed. Mazzella and Feingold (1994) asserted that, in general, jurors are more likely to perceive men as guilty due to their overrepresentation among offenders. In their meta-analysis of gender-based jury studies, they found a small but a significant pattern of harshness toward male defendants. In contrast, Devine and Caughlin's (2014) more recent metaanalysis found an effect size close to zero, indicating that defendant gender may not play a large role in juror decision-making. Notably, they also uncovered different patterns for student and community mock jurors. Community studies yielded a slightly greater likelihood of a guilty verdict for a woman as compared to a man, with the opposite effect for student participants.
There is some evidence that, as jurors, women are harsher in their decisions than are men (ForsterLee, ForsterLee, Horowitz, & King, 2006), especially for child abuse, sexual assault, or domestic abuse cases (e.g. Bagby et al., 1994;Burke, Ames, Etherington, & Pietsch, 1990;Quas, Bottoms, Haegerich, & Nysse-Carris, 2002). Wayne, Riordan, and Thomas (2001) demonstrated that female mock jurors are particularly harsh with female defendants. In same-gender and cross-gender harassment cases, women perceived the behavior of a female defendant to be more inappropriate, serious, and offensive than the same behavior performed by a male defendant. One possible explanation for these findings stems from Marques, Yzerbyt, and Leyens' (1988) black sheep effect theory. Derived from social identity theory (Tajfel & Turner, 1986), this explanation suggests that jurors categorize defendants as members of their in-group or out-group (in this case, based on gender). The black sheep effect posits that under certain conditions-such as when defendants deviate from prescribed norms-evaluations from members of one's in-group are more extreme. As a response to norm violations, jurors are likely to socially distance themselves by displaying harsher treatments for the in-group defendant, in comparison to the equivalent out-group defendant.
Research assessing the interaction between juror and defendant gender suggests that female jurors might only treat female defendants less favorably for certain types of crimes. Foley (1993) found that in sexual assault trials, female mock jurors were more likely to find the male defendant guilty than the female defendant. Given that women are more likely to be victims of sex crimes (U.S. Department of Justice, 2012), it is possible that female mock jurors are likely to empathize with the victim. Thus, women may perceive a male defendant's actions as more heinous in comparison to men. Devine and Caughlin's (2014) meta-analysis also found that the effect of juror gender is moderated by case type, such that cases involving sexual violence tend to elicit harsher verdicts from female jurors than male jurors, but there were few to no gender differences in other case types. Taken together, the literature on the effects of gender in the courtroom suggests that both juror gender and defendant gender have the potential to influence jurors' verdicts. However, aside from sexual assault cases it is unclear under what specific contexts gender stereotypes are most likely to influence juror judgments.

Crime congruency
Previous research shows that defendants are convicted more often and given harsher sentences when on trial for crimes that are stereotype-consistent with some personal characteristic. This crime congruency has primarily been examined in terms of defendant race (e.g. Gordon, 1990;Gordon, Bindrim, McNicholas, & Walden, 1988;Jones & Kaplan, 2003). More specifically, defendants possessing characteristics that match those expected for the typical offender are more likely to receive harsher treatment, because their behavior is perceived to be a result of internal dispositions; that is, jurors tend to rely on blame-related attributions (causal explanations for an event, Jones & Harris, 1967). Conversely, defendants possessing characteristics thought to be incongruent for the typical offender are likely to receive greater leniency because their behavior is perceived to be a consequence of situational factors (Gordon, 1990).
Scholars contend that the differential treatment of men and women can be explained in part by jurors' gender roles expectations (Thompson, 2010). Gender roles refer to the behaviors, activities, and traits that are deemed socially appropriate for each sex (Lips, 2005). For instance, in Western society, men are traditionally expected to demonstrate aggression and toughness, whereas women are expected to demonstrate nurturance and compassion (Best & Williams, 1990;Eagly & Steffen, 1984;Sheriffs & McKee, 1957;Wiest & Duffy, 2013). Such effects have been shown for perceptions of expert testimony. McKimmie, Newton, Terry, and Schuller (2004) found that mock jurors tended to view an expert witness more favorably when he/she testified in a case domain that was more consistent with gender stereotypes (i.e. automobile industry for a man and cosmetics industry for a woman). Hence, a match between a trial party's gender and gender role expectations might be more compelling for jurors.
It seems that prejudice toward male defendants may arise from the congruence that mock jurors perceive between the gender of the defendant and the crime (Steffensmeier & Kramer, 1982). This may result from the fact that male stereotypic characteristics (e.g. aggression, dominance) are more closely associated with criminal behavior compared to female stereotypic characteristics (e.g. nurturance, passivity; Steffensmeier & Kramer, 1982). Therefore, mock jurors might be more likely to attribute the deviant behavior of a man to internal dispositions, making his presence in the courtroom more plausible than a woman charged with a similar crime. However, some crime types do not feature such male-stereotypic characteristics. For instance, shoplifting could be perceived as a more passive crime, and shopping is an activity stereotypically associated with women (Dholakia & Chiang, 2003). Thus, a woman charged with shoplifting might constitute a more automatic match than a woman charged with auto theft.
However, it is also theoretically plausible that a mismatch between gender role and crime type will result in harsher treatment. More recent findings suggest that deviating from gender role stereotypes can have detrimental effects for defendants. Eagly and Karau's (2002) role congruity theory contends that women who commit grave violent crimes exhibit traits that are highly incongruent with feminine ideals (e.g. nurturance, compassion; Gilbert, 2002). As such, women perceived to have committed a violent crime are likely to be evaluated more negatively than a male counterpart. Supporting this theory, Wiest and Duffy (2013) found that mock jurors' perceptions of female defendants varied as a function of how well they fit traditional gender role expectations. More specifically, women were more likely to be found guilty when they violated gender role expectations of femininity using violence or deception. These findings are consistent with Thompson (2010), who found that greater incongruity was perceived when women violated their expected stereotypes, because men's behaviors were deemed to be more accepted in society and set the standard for normalcy. In their test of gender-crime congruency, Strub and McKimmie (2016) found that a woman who was portrayed using masculine terms yielded more negative evaluations of the defense's evidence in comparison to when she was described in feminine terms. However, whether a man was described in masculine or feminine terms did not significantly influence evaluations of the defense. They also found that men were rated as more consistent with stereotypes about murderers as compared to women. Strub and McKimmie (2016) concluded that women tend to be compared against gender stereotypes, whereas men are compared against offender stereotypes. In sum, it is possible for both gender congruence and incongruity to yield harsher judgments.

Current study
The literature on the effects of gender in the courtroom suggests that both juror gender and defendant gender have the potential to influence juror decisions (Devine & Caughlin, 2014;Mazzella & Feingold, 1994). Research has demonstrated that jurors may hold negative biases against defendants whose behavior is believed to be congruent or incongruent with gender role expectations. In following, given that these gender biases threaten a defendant's right to a fair trial, it is necessary to examine the ways in which gender role expectations impact on juror judgments in different types of criminal trials. The purpose of the present study was to understand how defendant gender, juror gender, and gender role expectations affected juror decision-making in shoplifting and auto theft cases.
Hypothesis 1: Dichotomous verdict decision. Drawing from the literature on gender differences in legal decision-making (e.g. Duke & Desforges, 2007;ForsterLee et al., 2006), we expected that cases in which the defendant's gender was congruent with the crime stereotype (male for grand theft of a motor vehicle, female for shoplifting) would yield a greater likelihood of a guilty verdict.
Hypothesis 2: Causal attributions. In line with the verdict hypothesis, we expected that gendercrime congruent conditions would elicit greater blame attributions (Jones & Harris, 1967).
Hypothesis 3: Defendant impressions. Finally, we predicted that defendant impressions would be more negative in gender-crime congruent conditions.

Trial transcript
Participants read a six-page fabricated grand theft of a motor vehicle or shoplifting trial transcript. Defendant gender was manipulated using names (Carl/Kelly), titles (Mr/Ms), and pronouns (he/his/ him, she/hers/her). The grand theft of a motor vehicle case was meant to represent a stereotypically masculine crime, while the shoplifting case was meant to represent a stereotypically feminine crime (a pilot study [N = 100] informed the selection of these two crimes based on their gender stereotypicality). Each transcript included opening and closing statements from the Prosecution and Defense, and testimony from/cross-examination of the defendant and witnesses. The auto theft case involved an accusation that the defendant stole a car from a used car dealership, for which the salesperson at the scene could not make a positive identification. The shoplifting case described a store clerk's allegation that the defendant stole a watch after trying it on in a jewelry store and fleeing. To ensure control of crime severity, the monetary value of the item stolen (car/watch) in each case was held constant ($7386). A pilot study (N = 52) of the trial transcripts revealed that there were no initial verdict biases for either the theft of a motor vehicle (χ² = .57, p = .45; 16 guilty, 12 not guilty) or shoplifting (χ² = .67, p = .41; 14 guilty, 10 not guilty) case. Participants read the California Penal Code criteria for the charges, as well as instructions regarding the burden of proof and reasonable doubt. They were then asked to provide a dichotomous verdict (not guilty, guilty).

Causal attributions
Participants were asked to indicate the degree to which they thought the defendant was responsible for the crime, was able to foresee the crime, and intended to commit the crime. Responses were indicated on a scale from 1 (not at all) to 9 (very much). These items showed good internal consistency, α = .78, and so they were averaged into a single variable.

Defendant impressions
To indicate their overall impressions of the defendant, participants rated him/her with regards to 15 dimensions (e.g. perceived honesty, likability, cruelty) on nine-point scales. The ratings for the 15 dimensions were averaged, with negative items reverse coded, resulting in higher scores reflecting more positive impressions. This measure demonstrated strong internal consistency, α = .92.

Social roles questionnaire
The Social Roles Questionnaire (Baber & Tucker, 2006) was used to assess participants' gender role attitudes. The scale includes 13 items that measure participants' beliefs that certain roles are associated with specific genders and the degree to which participants think about gender in non-dichotomous ways. We created a mean score variable, with higher scores indicating more traditional gender role attitudes. The scale demonstrated strong internal consistency, α = .88.

Procedure
This study received clearance from the university's ethics review board. Participants were first screened for jury eligibility via a brief demographics questionnaire. Eligible participants proceeded to the study, which was conducted online using Qualtrics software. After providing informed consent, participants were asked to play the role of a mock juror and carefully read through one of four randomly assigned transcripts online, in which the type of crime and gender of the defendant were manipulated. Participants were then asked to complete a series of opinion related measures that were in the form of a juror questionnaire. After completing another brief demographics questionnaire, participants were debriefed, thanked for their participation, and provided with a completion code to receive compensation.

Verdict
To determine whether mock juror gender (man/woman), defendant gender (man/woman), and type of crime (theft of a motor vehicle/shoplifting) would interact to influence participants' verdict decision (guilty/not guilty), we conducted a 2×2×2×2 hierarchical log-linear analysis (HILOG). This analysis revealed only an interaction between type of crime and verdict, χ 2 (1, N = 200) = 20.11, p < .001, v = .33. Participants who read about a defendant charged with theft of a motor vehicle were significantly more likely to vote guilty (76%) than those who read about a defendant charged with shoplifting (45%). No other significant effects emerged. Table 1 displays the verdict counts and percentages for each experimental condition.

Causal attributions
We conducted a 2 (defendant gender: man, woman) by 2 (juror gender: men, women) by 2 (crime type: shoplifting, auto theft) analysis of covariance (ANCOVA) using gender role attitudes as the covariate and causal attributions as the dependent variable. Results did not reveal any significant effects. Table 2 displays the results.

Defendant Impressions
We conducted a 2 (defendant gender: man, woman) by 2 (juror gender: men, women) by 2 (crime type: shoplifting, auto theft) analysis of covariance (ANCOVA) using gender role attitudes as the covariate and defendant impressions as the dependent variable. Table 3 displays the results. There was a significant relationship between gender role attitudes and defendant impressions, F(1, 191) = 4.68, p = .032, η 2 partial = .024, with more traditional gender role attitudes associated with more negative defendant impressions. There was a significant main effect of crime type, F(1, 191) = 23.65, p < .001, η 2 partial = .11. This effect was qualified by a significant crime type by juror gender interaction effect, F(1, 191) = 5.85, p = .017, η 2 partial = .03. Simple effects tests demonstrated that among women, there was a significant effect of crime type, t (97)

Discussion
This study was designed to extend our understanding of the role defendant and juror gender play in informing juror decisions in two types of criminal trials: shoplifting and auto theft. We expected a three-way interaction between juror gender, defendant gender, and crime type. Specifically, we expected cases where the defendant's gender was congruent with crime stereotypes (grand theft of a motor vehicle for a man, shoplifting for a woman) to yield a greater likelihood of a guilty verdict, greater blame attributions, and more negative defendant impressions. In line with the black sheep effect (Marques et al., 1988), we predicted that this effect would be stronger for jurors of the same gender.
Results did not support these predictions. There was a significant effect of crime type on verdicts, responsibility ratings, and defendant impressions, with harsher decisions for the auto theft as compared to the shoplifting case. Given that the monetary amount in each case was held constant, these findings may reflect that mock jurors perceived the grand theft of a motor vehicle case to be more severe in nature. However, juror and defendant gender did not independently or jointly impact on verdicts or defendant impressions. Interestingly, there was a significant juror gender by crimetype interaction effect on defendant impressions. For jurors who were women, shoplifting was associated with more positive defendant impressions, with no such effect for men. This effect could be suggestive of a form of similarity leniency (i.e., more favorable judgments for an in-group member), with women having more positive impressions of the crime that is stereotypically associated with women.
Overall, the current study provides no evidence of gender-crime congruency. The general lack of gender effects on verdict decisions supports findings from Devine and Caughlin's (2014) meta-analysis, which indicated that gender might not be as predictive as previous estimates had indicated. However, we cannot discount the possibility that lack of power contributed to the null findings. Further, the auto theft case appeared to be guilty-leaning in this sample, and so a case featuring greater ambiguity might yield different results. Accordingly, we do not wish to prematurely dismiss the idea of gender-crime congruency in this context. Findings from Strub and McKimmie (2016)  suggested that gender stereotype-based descriptors of a defendant can significantly influence mock juror judgments. Specifically, a woman charged with a male stereotypic crime was judged more harshly when described in masculine as compared to feminine terms (Strub & McKimmie, 2016). Further, we only tested for the effect of gender role attitudes, which leaves a wide array of other potential covariates unaccounted for. Therefore, future researchers should incorporate defendant gender stereotype manipulations as well as alternative moderator variables. Researchers might also consider testing other case types that are highly gendered, such as ones involving sex work.

Limitations and future directions
Several limitations to the present study should be noted in closing. As with other types of simulation research, ecological validity is an important consideration. First, to access a diverse population, we conducted this study online, which might differ significantly in comparison to a real trial process. Further, as it is difficult to emulate the social psychological stressors present when rendering a verdict decision, participants in the current study were not burdened with the emotional consequences of their decisions. Nonetheless, the use of web-based studies has been shown to be comparable, if not more representative, to that of in-lab research (Birnbaum, 2000;Buchanan & Smith, 1999;Krantz & Reeshad, 2000;Riva, Teruzzi, & Anolli, 2003) and to provide the benefit of reduced social desirability concerns (Maeder, Yamamoto, & McManus, 2017). Thus, we argue that they provide a sound basis to examine juror decision-making.
The current study also made use of a trial transcript, as opposed to a videotape or re-enactment of the proceedings. While this modality may be critiqued for having limited contextual information, research examining the differences between various modes of presentation is inconclusive (Bornstein, 1999). Perhaps most notably, in actual court cases, jurors are provided with the opportunity to deliberate. This process provides juries with the opportunity to exchange beliefs, which may in turn influence subsequent judgments. Although the use of individual mock juror judgments helps decipher individual jurors' decision-making process, future studies should investigate the influence of gender after a deliberation period. Researchers have found that deliberations may act to diminish or enhance individual juror bias (Kerr, Hymes, Anderson, & Weathers, 1995). Further, to ensure the findings are not unique to these two cases, replication of this study should be undertaken using different types of stereotypically male and stereotypically female crimes to strengthen and expand the results.
Notwithstanding these limitations, the present findings underscore the dynamic relationship between gender expectations and mock juror judgments. When selected to participate in the judicial process, jurors might carry with them existing gender expectations that both guide their perceptions of the defendant and shape their judicial decisions. Thus, the courtroom constitutes an influential setting in which to investigate gender biases and a defendant's right to a fair and impartial trial.