Abstract
Research has consistently shown that jurors are influenced by multiple schemas in cases of alleged sexual assault, including offense stereotypes and victim stereotypes. These schemas appear to be organized in a hierarchy, as victim stereotypicality seems to matter most in acquaintance assaults (counter-stereotypical offense). However, despite numerous studies demonstrating the impact of defendant stereotypes on juror perceptions of guilt for other crimes, to date, the impact of stereotypes about defendants (perpetrators) in cases involving sexual violence have been overlooked. As such, the current research aimed to build on the existing hierarchical schema model by systematically examining the influence of perpetrator stereotypes. Following pilot work, mock jurors’ (N = 163) read a rape scenario that varied in terms of offense stereotypicality (stereotypical, counter-stereotypical), victim stereotypicality (stereotypical, counter-stereotypical), and perpetrator stereotypicality (stereotypical, counter-stereotypical). Broadly consistent effects of offense stereotypicality and victim stereotypicality were observed across the outcome measures, such that the victim was perceived more positively and the perpetrator more negatively when the victim was described as being stereotypical and when the offense was described as stereotypical. However, contrary to past findings, the effect of victim stereotypicality did not differ as a function of offense stereotypicality. Furthermore, perpetrator stereotypicality did not influence perceptions in the stereotypical offense scenario. These findings suggest that contrary to the assertions of previous research, there is not a series of specific, individual stereotypes that impact attributions of blame, rather, there may be one underlying schema about consent that influences perceptions. These findings have important implications for how we address the effect of juror-held schemas on attributions of blame in cases of sexual assault.
Despite the severity and prevalence of rape and sexual assault, the rates of reporting, prosecution, and conviction are disappointedly low compared with any other crimes (Daly & Bouhours, 2010; Fileborn, 2011; Taylor, 2007). Although a number of legal and extra-legal factors contribute to this disparity (Clark & Quadara, 2010; Taylor, 2007), sexual assault–related schemas held by decision-makers, including jurors, influence the justice process. To eliminate the influence of these factors on juror decision-making, it is imperative to identify the specific schemas that jurors hold, and to determine how these schemas affect interpretations of the defendant, the complainant, and the case for crimes of sexual violence.
Due to the nature of rape and sexual assault, evidence is often circumstantial and convictions typically hinge on the contradictory accounts of the defendant and complainant (Temken & Krahé, 2008). When this happens, jurors have been found to turn to schemas, or stereotypes, to assist them to attend, to evaluate, and make sense of ambiguous information and apportion blame (Daly & Bouhours, 2010; McKimmie, Masser, & Bongiorno, 2014). Research has repeatedly demonstrated that offense-related and victim-related schemas influence impressions about the blameworthiness of the defendant and complainant in cases of rape and sexual assault (Abrams, Viki, Masser, & Bohner, 2003; Masser, Lee, & McKimmie, 2010; McKimmie et al., 2014; Schuller, McKimmie, Masser, & Klippenstine, 2010). However, despite numerous studies demonstrating the impact of defendant stereotypes on juror perceptions of guilt for other crimes (Bodenhausen, 1988; Dean, Wayne, Mack, & Thomas, 2000; Gordon, Bindrim, McNicholas, & Walden, 1988; C.S. Jones & Kaplan, 2003), to date, the impact of defendant (perpetrator) stereotypes in cases involving sexual violence have been largely ignored. The current research aimed to address this gap by systematically examining the influence of offense stereotypes, victim stereotypes, and perpetrator stereotypes on case evaluations, and perceptions of defendant and complainant behavior in the context of an alleged case of rape.
Rape Myths
Until recently, much of the research on perceptions of sexual violence focused on the influence of rape myths (Horvath & Brown, 2013; Krahé, Bieneck, & Scheinberger-Olwig, 2007; Lonsway & Fitzgerald, 1994). Conceptually linked to a belief in a “just world” (Lerner, 1980), rape myths are set of stereotypical beliefs that serve to justify and dismiss male sexual aggression against women (Bohner, Eyssel, Pina, Siebler, & Viki, 2009; Brownmiller, 1975; Gerger, Kley, Bohner, & Siebler, 2007; Grubb & Turner, 2012), absolve the perpetrator, and increase the victim’s responsibility for the assault (Burt, 1980; Grubb & Turner, 2012). While rape can and does involve victims and perpetrators of all genders, it remains a gendered crime that is predominantly committed by male perpetrators against female victims (Fileborn, 2011). Due to the gendered nature of the crime, it is perhaps unsurprising that rape myth endorsement is consistently positively associated with stronger sex role stereotyping (Payne, Lonsway, & Fitzgerald, 1999), hostile attitudes toward women (Costin, 1985; Grubb & Turner, 2012; Suarez & Gadalla, 2010), and blaming the female victim more and the male perpetrator less in sexual assault cases (Finch & Munro, 2005; Grubb & Turner, 2012; Temken & Krahé, 2008)
Research on sexual violence has focused on these myths; however within this research, beliefs about the offense (offense stereotypes) and about the victim (victim stereotypes; McKimmie et al., 2014) have often been conflated. Recent research has started to address this by disentangling the various stereotypes and assessing their effects. However, within this literature there has been a focus on offense stereotypes (Abrams et al., 2003; Viki, Abrams, & Masser, 2004) and victim stereotypes (Lonsway & Fitzgerald, 1994; Masser et al., 2010; McKimmie et al., 2014; Schuller et al., 2010). In contrast, perpetrator stereotypes have been relatively neglected (Pollard, 1992).
Offense Stereotypes
To reach a verdict, jurors are required to assess whether or not an offense matches a specific and often narrow legal definition (Finkel & Groscup, 1997). According to Pennington and Hastie (1992), jurors do this by organizing trial information into a coherent narrative, which is influenced by the offense prototype, commonly referred to as a stereotype, that helps them to interpret the evidence to construct a plausible series of events. Jurors often construct more than one “story,” switching between them to engage in a “goodness-of-fit” test to help determine which constructed narrative best matches the features required to reach a verdict (Pennington & Hastie, 1992). The more stereotypical and coherent a story is, the more likely a guilty verdict will result (Finkel & Groscup, 1997; Pennington & Hastie, 1992).
Although the majority of rapes are committed by an acquaintance of the victim in a private residence, a “real” or stereotypical rape is depicted as a crime committed by a stranger in a deserted public place (DuMont, Miller, & Myhr, 2003). Consistent with this stereotype, it has been found that perpetrators of stranger rape (stereotypical rape) are perceived as more culpable for the offense than perpetrators of acquaintance rape (counter-stereotypical rape; Krahé, 1991; Pollard, 1992). In contrast, victims of acquaintance rape are often perceived as less genuine (Lonsway & Fitzgerald, 1994; Pollard, 1992; Temken & Krahé, 2008) and are blamed more for their assault than victims of stranger rape (Abrams et al., 2003; Masser et al., 2010; Schuller et al., 2010).
In cases of rape, deciding whether there was an absence of mutual consent for engaging in sexual activity is the critical feature for reaching a verdict (Fileborn, 2011). During the trial, jurors are asked to establish whether the offense meets several legal criteria: (a) that an act of sexual penetration occurred, (b) that the victim did not consent to the act, and (c) that the defendant was aware that the victim did not consent (Fileborn, 2011). In cases of stranger rape, the nature of the assault implies that consensual sex is highly unlikely and defendants often deny that intercourse occurred. As a consequence, the main issue that is often raised as a defense is misidentification, with the question becoming “Was it him?” rather than “Did she consent?” (Spohn & Holleran, 2001). However, in acquaintance rape, where the assault may be hidden in heterosexual dating scripts (Clark & Quadara, 2010; Ellison & Munro, 2009), defendants commonly argue that the victim consented (Bryden & Lengnick, 1997). When this happens, jurors may turn to other factors, such as victim stereotypes, to determine what happened (Masser et al., 2010; McKimmie et al., 2014). Jurors may then use these stereotypes to assess the victim’s credibility and behavior to establish whether the event was consensual or not (Masser et al., 2010; McKimmie et al., 2014; Olsen-Fulero & Fulero, 1997).
Disentangling Gender and Victim Stereotypes
Although it is reasonable to expect that jurors evaluate and interpret the behavior and characteristics of both the perpetrator and the victim when making attributions of blame, the focus of the research has remained largely on the victim (Pollard, 1992; Viki et al., 2004). For a victim to be perceived as “real,” and therefore less blameworthy, she must adhere to several behavioral standards. For example, she must have done nothing to warrant the assault, vigorously resisted the perpetrator (and be physically injured whilst doing so), report the rape immediately to police, and be appropriately emotionally traumatized after the event (Lonsway & Fitzgerald, 1994; Temken & Krahé, 2008). However, victims’ responses to assaults vary, and many victims do not behave in ways that fit these narrowly prescribed standards (Alderden & Ullman, 2012).
As with offense stereotypes, victims who deviate from these stereotypes are viewed as counter-stereotypical (Masser et al., 2010), less credible (Schuller & Stewart, 2000), and are more likely to be blamed for their assault (Buddie & Miller, 2001). However, victim stereotypes have often also been conceptualized as violations of gender stereotypes (Abrams et al., 2003; Masser et al., 2010). For example, traditional sex role stereotypes assert that women are the “sexual gatekeepers” and must therefore remain chaste and respectable (Ellison & Munro, 2009; Masser et al., 2010). When victim behavior is contrary to these stereotypes in that she fails to resist the attack, dresses provocatively, has had prior sexual experiences, and has drunk alcohol, victims are viewed as having acted in a way that precipitated the assault or having consented to intercourse (Ellison & Munro, 2009; Payne et al., 1999; Pollard, 1992). Furthermore, a victim’s post-assault behavior, such as emotional demeanor, has also been found to influence the extent to which the victim is perceived to be stereotypical (Schuller et al., 2010). Victims who express greater emotional distress following an attack are perceived to be more credible (Buddie & Miller, 2001), and when a victim is described as tearful and upset, more blame is attributed to the perpetrator (Schuller et al., 2010).
However, this overlap of victim and gender stereotypes is a potential confound, as both “good” women and “real” victims are expected to uphold many of the same behavioral standards (Abrams et al., 2003; Masser et al., 2010). Furthermore, this overlap makes it difficult to identify the specific stereotype to target in an intervention. Masser et al. (2010) independently manipulated gender and victim stereotypes in a case of acquaintance rape. To manipulate gender typicality, the victim was described as being a widowed mother who had either left her children home alone (gender counter-stereotypical) or had left them with a trusted baby sitter (gender stereotypical) while she attended a party. The results showed that gender stereotypicality only influenced attributions of blame when the victim acted counter to the victim stereotype (e.g., she said “no” and asked him to stop but did not physically resist her attacker). Masser et al. (2010) concluded that perceivers may switch between a hierarchy of stereotypes—first turning to victim stereotypes and then “stepping down” to other stereotypes, such as gender, when the victim stereotype fails to provide a coherent framework for the events.
Gender stereotypes aside, McKimmie et al. (2014) found further support for a hierarchical schema model when they “stepped up” the schema hierarchy to include offense stereotypes, depicting the assault as either a stranger (stereotypical) or acquaintance (counter-stereotypical) rape. In this study, victim stereotypes had different effects depending on whether the assault was portrayed as stereotypical (stranger) or counter-stereotypical (acquaintance). Specifically, when the victim was described as counter-stereotypical in an acquaintance rape scenario, she was perceived less positively and the defendant was perceived more positively and less likely to be guilty compared with when the victim was described as stereotypical. However, it was found that victim stereotypicality had little to no effect on ratings of guilt and blame in a stranger rape scenario.
As a follow up to these findings, the analysis of qualitative data revealed that participants considered different factors when reaching their verdict in the stranger compared with the acquaintance rape scenarios. Consistent with the hierarchical schema model, mock jurors appeared to first assess whether the assault was consistent with the offense stereotype. When the assault did not match the crime stereotype, in the acquaintance rape case, participants tended to focus on victim behaviors to judge whether consent was given or implied. In contrast, when the assault matched the offense stereotype, as was the case in stranger rape scenario, participants tended to focus on the perpetrator and his behavior.
While these findings hold important implications for understanding how and when victim stereotypes influence juror perceptions in cases of sexual assault, they also highlight a gap in our current knowledge. Given that the qualitative findings of McKimmie et al. (2014) suggest that jurors switch focus from the perpetrator to the victim in cases of stranger and acquaintance assault, it is likely that perpetrator characteristics play a larger role in this decision-making process than has been previously considered. However, as research has predominantly focused on victim stereotypes, it is unclear what characterizes a stereotypical perpetrator and how these attributes might interact with offense prototypes and victim stereotypes (Pollard, 1992; Viki et al., 2004).
Defendant and Perpetrator Stereotypes
At a broader level, research on juror decision-making has focused on the influence of stereotypes about defendants (or perpetrators) on perceptions of guilt for a variety of offenses. It has been repeatedly demonstrated that characteristics such as race (Gordon et al., 1988; C. S. Jones & Kaplan, 2003; Mitchell, Haw, Pfeifer, & Meissner, 2005), ethnicity (Bodenhausen & Wyer, 1985), and gender (Dean et al., 2000; McKimmie, Masters, Masser, Schuller, & Terry, 2013) significantly influence juror perceptions. While these effects have mostly been conceptualized as a function of negative stereotypes, the effect of belonging to a negatively stereotyped group is contingent on the type of crime alleged to have been committed (Gordon et al., 1988; C. S. Jones & Kaplan, 2003). Specifically, defendants who possess characteristics that match those stereotypically associated with the crime they are alleged to have committed are more likely to be found guilty and receive harsher punishment recommendations (Gordon et al., 1988). For example, Black defendants charged with burglary and White defendants charged with embezzlement receive longer recommended sentences compared with Black defendants charged with embezzlement and White defendants charged with burglary (Gordon et al., 1988).
One explanation for this defendant-crime congruency effect is through stereotype-activation theory. This theory suggests that when a defendant matches the stereotype associated with a particular crime, jurors engage in less effortful processing of case information. Instead, they rely on the negative stereotype to attend to and process evidence that is consistent with their expectation of guilt (Gordon et al., 1988). When this happens, jurors are more likely to make dispositional attributions for the behavior of the defendant (Bodenhausen, 1988; C. S. Jones & Kaplan, 2003). However, when a defendant is incongruent with a negative crime stereotype, jurors engage in more effortful processing of case information and are more likely to consider situational factors for the defendant’s behavior (Gordon et al., 1988).
We propose that the defendant-crime congruency effect may help to explain the focus on the perpetrator (who is the defendant) in cases of stranger rape and the focus on the victim in cases of acquaintance rape. Specifically, when the crime is depicted as a stranger rape, jurors are likely to turn to perpetrator characteristics to assist them to answer the central question of whether the defendant is the alleged perpetrator. Whereas when the crime is depicted as an acquaintance rape, and is therefore hidden in heterosexual dating scripts, jurors may switch to victim characteristics to determine whether the interaction was consensual or not (Masser et al., 2010).
The Current Research
To effectively counter juror bias related to sexual assault schemas, we must first establish how and when particular beliefs about sexual assault impact juror decision-making. Rape myths portray typical perpetrators as psychologically abnormal men who are aggressive toward their victim (Krahé, 1991; Payne et al., 1999). However, it is currently not well understood how perceivers decide whether a perpetrator of sexual assault is psychologically “abnormal” compared with a “normal” man (Clark & Quadara, 2010; Pollard, 1992). Furthermore, although there have been some efforts to disentangle victim and gender stereotypes (Masser et al., 2010), the distinction between these stereotypes for characteristics outside of motherhood is yet to be established. As such, the current research aimed to (a) disentangle and identify the specific beliefs that differentiate a typical man from a typical perpetrator of rape, (b) further disentangle and identify the specific beliefs that differentiate a typical woman from a typical victim of rape, and (c) build on the previously established hierarchical schema model by independently manipulating the established offense prototypes, victim stereotypes, and perpetrator stereotypes.
Once these stereotypes had been established, it was expected that victim stereotypicality would have a greater effect on judgments in the counter-stereotypical (acquaintance) assault scenario than in the stereotypical (stranger) assault scenario (McKimmie et al., 2014). Specifically, a counter-stereotypical victim of an acquaintance assault would be perceived more negatively and the defendant more positively and seen as less likely to be guilty compared with when the victim was depicted as stereotypical (Hypothesis 1 [H1]). However, consistent with the defendant-crime congruency effect (Gordon et al., 1988), it was also expected that perpetrator stereotypicality would have a greater effect in the stereotypical stranger assault scenario than in the counter-stereotypical acquaintance scenario. Specifically, it was predicted that a stereotypical perpetrator in the stranger assault scenario would be seen as more likely to be guilty and perceived more negatively and the victim perceived more positively compared with when the perpetrator was depicted as counter-stereotypical (Hypothesis 2 [H2]). We expected no impact of victim stereotypicality in the stranger rape scenario, and no impact of perpetrator stereotypicality in the acquaintance rape scenario.
Method
Pilot Study 1
To identify the specific beliefs that differentiate typical women from typical rape victims and typical men from typical rape perpetrators, two preliminary studies were conducted. While Masser et al. (2010) began to address this, it is not fully understood which gender stereotypes outside of those associated with motherhood are distinct from those of a typical rape victim. Furthermore, although some rape myths depict a perpetrator as an aggressive, psychologically abnormal man, the specific beliefs that distinguish a typical man from a rape perpetrator have not been established (Clark & Quadara, 2010; Pollard, 1992). As such, a pilot study was conducted to determine current beliefs about rape victims and rape perpetrators, and to check which beliefs were more indicative of gender-based stereotypes. A second preliminary study was conducted to check the effectiveness of these beliefs as part of the proposed manipulation for the main study.
Participants and method
Participants were 152 members of the community (51% women) ranging in age from 18 to 77 years (M = 43.16, SD = 15.02). They were presented with 34 statements constructed from victim, perpetrator, and gender stereotypes that had been given in open-ended responses by participants in preliminary work and had been gathered from the literature (Clark & Quadara, 2010; Pollard, 1992; Viki et al., 2004). Half of the statements related to women and rape victims and the other half to men and perpetrators. Participants rated the extent to which they agreed that the statement was indicative of (a) a typical rape victim, or a typical woman or (b) a typical man, or a typical rape perpetrator on a 7-point Likert scale from 1 (strongly disagree) to 7 (strongly agree). Example items describing either a victim stereotype or a gender stereotype about women were “has low self-esteem” and “flirts in pubs or nightclubs.” Example items describing either a perpetrator stereotype or a gender stereotype about a man were that he “is angry toward women” and “behaves strategically to obtain sexual intercourse.”
Results
The responses were analyzed with a series of paired t-tests to test for absolute differences between the mean scores on ratings of typicality for either a woman, or a victim, or a man, or a perpetrator. The analyses showed that typical rape victims were seen as significantly more weak or vulnerable, t(125) = 9.27, p < .001; frightened, helpless, and scared, t(125) = 13.32, p < .001; sad and depressed, t(125) = 12.18, p < .001; ashamed and embarrassed, t(125) = 15.39, p < .001; and to have lower self-esteem, t(123) = 9.67, p < .001, than women in general. It was also found that typical rape perpetrators were seen as significantly more aggressive, t(124) = 15.07, p < .001, and angry toward women, t(125) = 17.54, p < .001; insensitive to the feelings of others, t(124) = 17.54, p < .001; impulsive, t(124) = 8.05, p < .001; and significantly less socially skilled, t(125) = −7.05, p < .001, compared with men in general. On the basis of these results, these characteristics were used to form part of the scenario for the next phase of preliminary testing.
Pilot Study 2
Participants and method
Thirty-two undergraduate psychology students (14 men, 18 women) aged 18 to 28 years (M = 20.00, SD = 2.66) participated in this study for partial course credit. On arrival at the laboratory, participants were told that the study involved written testimony regarding several cases of alleged rape and were asked to provide informed consent. They were then asked to follow the instructions presented on a computer via the Qualtrics survey platform. All participants were presented with three pairs of hypothetical scenarios depicting details of a stereotypical and counter-stereotypical victim, perpetrator, and offense. In all scenarios it was alleged that the defendant grabbed the victim and tried to take her clothes off and that the victim told the defendant to stop, but he did not listen to her and instead held her down and raped her. The details of the manipulations of the stereotypes within the scenarios are as presented below. Participants were told that the scenarios were excerpts from six distinct and unrelated cases of alleged rape. The presentation of all scenarios was randomized to account for order effects.
Manipulation of offense stereotypicality
The first pair of scenarios manipulated offense stereotypicality with the case being described as either a stranger (stereotypical) or an acquaintance (counter-stereotypical) rape, based on Abrams et al. (2003), Masser et al. (2010), and McKimmie et al. (2014). In the stereotypical offense (stranger) scenario, participants were told that as the victim walked to her car after a party, a man who she did not know and had not met before approached and subsequently assaulted her. In the counter-stereotypical offense (acquaintance), participants were told that the victim and perpetrator knew each other through mutual friends and the assault occurred in the victim’s home after the defendant walked her home from a party. After each offense excerpt was presented, participants were asked to indicate how well the people in the scenario knew one another from 1 (not at all well) to 7 (very well). Participants were also asked to indicate whether the victim and perpetrator in the scenario were either strangers to, or acquaintances of, one another in a forced choice question as the relationship between the victim and perpetrator has been identified as a feature of offense stereotypicality (Abrams et al., 2003; Masser et al., 2010; McKimmie et al., 2014). A paired samples t test indicated that the offense manipulation was successful, t(31) = −13.53, p < .001, as participants thought that the victim and perpetrator knew one another significantly less in the stranger assault scenario (M = 1.03, SD = 0.18) compared with the acquaintance assault scenario (M = 3.84, SD = 1.17). In addition, all participants correctly classified the stereotypical assault as a stranger assault and the counter-stereotypical assault as an acquaintance assault.
Manipulation of victim stereotypicality
The second set of scenarios manipulated victim stereotypicality via a written summary of the statement made by the victim during an interview with police. In the stereotypical victim scenario, the victim was said to appear timid, withdrawn, and as emotional when talking about the assault. In the counter-stereotypical victim scenario, the victim was said to appear composed and calm, and was not emotional when detailing the events of the assault. Participants were then asked to rate the extent to which they perceived the victims to be similar to a stereotypical victim of rape, how similar they were to other victims of rape, and how similar they were to other people who have been raped from 1 (not at all) to 7 (very). These items comprised the measure of participants’ perceptions of the victims’ stereotypicality (α = .84, stereotypical condition; α = .86, counter-stereotypical condition). The manipulation of victim stereotypicality was successful, t(31) = 10.39, p < .001, as participants perceived the stereotypical victim (M = 5.67, SD = 0.87) as more stereotypical than the counter-stereotypical victim (M = 2.81, SD = 1.05).
Manipulation of perpetrator stereotypicality
In the final pair of scenarios, perpetrator stereotypicality was varied. Participants were presented with a summary of the police interview with the defendant. In the stereotypical perpetrator scenario, the defendant was described as being socially awkward and becoming angry and unsympathetic during questioning. In the counter-stereotypical perpetrator condition, the defendant was described as behaving in the opposite manner. Participants were asked to rate the perpetrator on three items based on the victim stereotypicality measure (e.g., the extent to which they perceived the perpetrator to be similar to a stereotypical perpetrator of rape), with these items comprising the measure of participants’ perceptions of the perpetrators stereotypicality (α = .80, stereotypical; α = .86, counter-stereotypical). The manipulation of perpetrator stereotypicality was also successful, t(31) = 6.00, p < .001, as participants perceived the stereotypical perpetrator (M = 4.92, SD = 1.11) to be significantly more stereotypical than the counter-stereotypical perpetrator (M = 3.04, SD = 1.09).
Main Study
Participants and design
Participants (N = 163) were recruited from undergraduate psychology courses in Queensland, Australia, and participated for partial course credit. Fifteen participants were excluded from the analyses for failing to meet the 18-year minimum age criteria for the study (n = 11) and exiting the study before completing the dependent measures (n = 4). The final sample comprised 148 participants (107 women, 41 men) aged 18 to 42 years (M = 20.05, SD = 3.59). No participant had previously served as a member of a jury.
Participants were randomly allocated to one of the eight conditions formed by the between-subjects manipulation of offense stereotypicality (stereotypical, counter-stereotypical), victim stereotypicality (stereotypical, counter-stereotypical), and perpetrator stereotypicality (stereotypical, counter-stereotypical).
Materials and procedure
Participants were recruited and tested via the same procedure used in the second pilot study. In this study, participants read a 460 to 480 word summary of a (fictional) rape trial. The summary detailed that Luke (the defendant) was charged with the rape of Sarah (the complainant). They were told that during the trial, the Prosecutor called Sarah to the stand to testify about the events that led up to the alleged rape. In all the scenarios Sarah testified that she had been at a birthday party with some friends on the evening of May the 17th and that the assault occurred afterward.
Offense, victim, and perpetrator stereotypicality were manipulated using the text assessed in the second pilot study. To improve the ecological validity of the manipulations, further information about the victim and perpetrator’s behavior during the assault was provided. This information was adapted from the scenarios used by Masser et al. (2010) and McKimmie et al. (2014), and was consistent with the level of detail that rape victims are commonly asked to provide during cross examination (Brereton, 1997). For example, participants in the stereotypical victim condition were also told that as well as telling Luke to stop, Sarah physically resisted him. In contrast to this, participants in the counter-stereotypical condition were told that despite telling Luke to stop, Sarah did not physically resist him. Participants in the stereotypical perpetrator condition were told that despite Sarah telling Luke to stop, he held her down with aggressive force and eventually penetrated her. Conversely, participants in the counter-stereotypical perpetrator condition were told that during the alleged assault, Luke held Sarah down with non-aggressive force and eventually penetrated her.
Manipulation checks
The success of the manipulations of victim stereotypicality (α = .90) and perpetrator stereotypicality (α = .95) were assessed using the same items that were used to assess the constructs in the second pilot study. For consistency, and to gain a multi-item of measure of offense stereotypicality, participants were also asked to indicate the extent to which the events in the scenario were similar to a typical act of rape, how similar they were like a typical act of rape, and to what extent they would categorize the events in this case as a typical rape from 1 (not at all) to 7 (very), α = .92.
Perceptions about the victim
Participants were asked to indicate the extent to which they held the victim accountable for the assault using Abrams’ and colleagues (2003) seven-item victim blame scale (α = .92). Example items include “How much sympathy do you feel for Sarah”(reverse scored) and “To what extent should Sarah have behaved differently” from 1 (not at all) to 7 (very much), α = .91.
Next, four items assessing perceptions of case strength were included to assess participant perceptions’ of the victim’s version of events. Participants were asked to rate their perceptions of the Prosecution’s case on the following dimensions: unconvincing (reverse scored), believable, credible, and strong. Response options ranged from 1 (not at all) to 7 (very), α = .79.
Perceptions about the perpetrator
To assess perceptions of the perpetrator, participants were asked to indicate the extent to which they held the perpetrator accountable for the assault. The six-item perpetrator blame scale was adapted from Abrams et al. (2003) and included items such as “How much do you think Luke should blame himself for what happened.” Response options ranged from 1 (not at all) to 7 (very much). One item was reverse scored (α = .81).
Next, participants were asked how dangerous they perceived Luke to be. An eight-item scale was constructed using items from the Perceived Dangerousness of Offenders Scale (Quinsey & Cyr, 1986) and the Attitudes Toward Prisoners Scale (Melvin, Gramling, & Gardner, 1985). The scale was designed to gain a better understanding of the beliefs underlying perpetrator stereotypes. To do this, participants were asked to what extent they agreed or disagreed with statements such as “Luke is a hostile person” and “Luke is unlikely to be violent” from 1 (strongly disagree) to 7 (strongly agree), α = .90. Two of the items were reverse scored. Participants were also asked about the defendant’s guilt likelihood. Specifically, participants were asked to indicate how likely they thought it was that the defendant committed rape from 1 (not at all likely) to 7 (very likely).
Rape myth acceptance (RMA)
Costin’ R scale (1985) was used to measure RMA. Participants were asked to indicate their level of agreement or disagreement with 18 statements using 1 (strongly disagree) to 7 (strongly agree) scales. The scale measures a number of rape-related beliefs such as, “Most charges of rape are unfounded” and “No healthy adult female who resists vigorously can be raped by an unarmed man.” Higher scores indicate greater acceptance of rape myths, with five items reverse scored (α = .71).
Following completion of these measures, a final screen appeared and informed participants that the study was complete. Participants were then debriefed and offered the opportunity to discuss and ask questions about the study with the experimenter in private.
Results
Preliminary Analysis
RMA
To confirm that there were no systematic differences in levels of RMA as a function of the independent variables, a 2 (offense: stereotypical, counter-stereotypical) by 2 (victim: stereotypical, counter-stereotypical) by 2 (perpetrator: stereotypical, counter-stereotypical) between-participants analysis of variance (ANOVA) with RMA as the dependent variable was conducted. There were no significant main effects or interactions (all Fs > .110).
Manipulation Checks
To assess the effectiveness of the offense, victim, and perpetrator manipulations, a series of 3 (condition) by 2 (stereotypicality) between groups ANOVAs were conducted on the measures of offense stereotypicality, victim stereotypicality, and perpetrator stereotypicality.
Offense stereotypicality
As expected, there was a significant main effect of offense stereotypicality, F(1, 143) = 6.22, p < .01, η2 = .04, with participants reporting that the stranger assault scenario was more stereotypical (M = 5.06, SD = 1.20) than the acquaintance assault scenario (M = 4.60, SD = 1.31). However, contrary to expectations, there was also a main effect of victim stereotypicality, F(1, 143) = 21.07, p < .001, η2 = .12, indicating that participants perceived the offense to be more stereotypical when the victim was described as stereotypical (M = 5.23, SD = 1.14) compared with counter-stereotypical (M = 4.27, SD = 1.22). The main effect of perpetrator stereotypicality was not significant, F(1, 143) = 4.51, p = .070, η2 = .01, nor were there any significant interactions (Fs < .593).
Victim stereotypicality
Consistent with a successful manipulation, there was a significant main effect of victim stereotypicality, F(1, 143) = 28.46, p < .001, η2 = .17, with participants perceiving the stereotypical victim (M = 5.57, SD = 1.29) as more stereotypical than the counter-stereotypical victim (M = 4.20, SD = 1.65). However, there was also a main effect of offense stereotypicality, F(1, 143) = 3.90, p < .05, η2 = .02, such that the victim was perceived as more stereotypical when the offense was depicted as a stranger assault (M = 5.27, SD = 1.53), compared with an acquaintance assault (M = 4.71, SD = 1.64). The main effect of perpetrator stereotypicality was non-significant, F(1, 143) = .09, p = .771, η2 = .00, and there were no significant interactions (Fs < .115).
Perpetrator stereotypicality
Contrary to results from the pilot study, the main effect of perpetrator stereotypicality was not significant, F(1, 142) = 3.44, p
Exploratory analyses
Given that the manipulations of offense, victim, and perpetrator stereotypicality did not work as intended, zero-order correlations were calculated on the measured stereotypicality variables: perceptions of offense stereotypicality, perceptions of victim stereotypicality, and perceptions of perpetrator stereotypicality. The aim of this was to explore the possibility that the manipulated variables were having simultaneous, rather than interdependent, effects on participants’ perceptions of stereotypicality. As shown in Table 1, perceptions of offense stereotypicality, perceptions of victim stereotypicality, and perceptions of perpetrator stereotypicality were all significantly positively correlated with one another. This shows that the extent to which evaluations of stereotypicality were made on one dimension was positively related to evaluations of stereotypicality being made on the other dimensions. For example, the more the participants perceived the victim as stereotypical, the more they perceived the perpetrator and the offense as stereotypical. This was also the case for each combination of variables.
Intercorrelations Among Measured Stereotypicality Variables.
Note. All measures are scaled so that higher numbers represent more of the construct.
p < .05. **p < .01. ***p < .001.
Main Analysis
Case evaluations
A series of 2 (offense stereotypicality) by 2 (victim stereotypicality) by 2 (perpetrator stereotypicality) between-participants analyses of covariance (ANCOVAs) with RMA as the covariate (see Masser et al., 2010 and McKimmie et al., 2014) were conducted on the measures assessing victim blame, case strength, likelihood that the perpetrator committed rape, perpetrator blame, and perceived perpetrator dangerousness. Higher scores of RMA were significantly associated with higher levels of victim blame, lower ratings of the prosecution’s case strength, lower perceptions of likelihood that the perpetrator committed rape, lower perpetrator blame, and lower perceptions of perpetrator dangerousness (effect sizes are presented in Table 2).
Means and Significant Main Effects of Offense Stereotypicality, Victim Stereotypicality, and Perpetrator Stereotypicality.
Note. Pairs of means are significantly different as indicated by *p < .05. **p < .01. ***p < .001; df (1, 147). RMA = rape myth acceptance.
The results of the ANCOVAs on the remaining measures showed consistent effects. There were significant main effects of offense stereotypicality, F(1, 147) = 15.49, p < .001, η2 = .07, and victim stereotypicality, F(1, 147) = 23.24, p < .001, η2 = .09, for victim blame, indicating that participants were less likely to blame the victim, when the scenario was depicted as a stranger rape compared with an acquaintance rape and when the victim was described as stereotypical compared with counter-stereotypical (see Table 2).
Significant main effects for offense stereotypicality, F(1, 147) = 5.87, p < .05, η2 = .03, and victim stereotypicality, F(1, 147) = 31.75, p < .001, η2 = .17, were also found for perceptions of likelihood that the perpetrator committed rape. Participants were more likely to perceive the perpetrator as having committed rape when the scenario was depicted as a stranger rape compared with an acquaintance rape and when the victim was described as stereotypical compared with counter-stereotypical. Main effects of offense stereotypicality, F(1, 147) = 25.67, p < .001, η2 = .13, and victim stereotypicality, F(1, 147) = 10.34, p < .01, η2 = .05, were also found for perpetrator blame. Participants blamed the perpetrator more when the offense was described as a stranger rape compared with an acquaintance rape and when the victim was described as stereotypical compared with counter-stereotypical.There was also a significant main effect of victim stereotypicality on perceptions of prosecution case strength, F(1, 147) = 8.91, p < .01, η2 = .06. Participants were more likely to perceive the prosecution’s case as stronger when the victim was depicted as stereotypical compared with counter-stereotypical. Finally, there were significant main effects of offense stereotypicality, F(1, 147) = 14.12, p < .001, η2 = .08; victim stereotypicality, F(1, 147) = 17.23,p < .001, η2 = .09; and perpetrator stereotypicality, F(1, 147) = 11.60, p < .01, η2 = .06, on perceived perpetrator dangerousness. Participants were more likely to see the perpetrator as dangerous when the scenario was depicted as a stranger rape compared with an acquaintance rape, when the victim was depicted as stereotypical compared with counter-stereotypical, and when the perpetrator was described as stereotypical compared with counter-stereotypical. However, contrary to H1, there were no significant interactions between offense stereotypicality and victim stereotypicality on any of the dependent measures (all ps > .05). Further, there were no significant interactions on any of the dependent measures between offense stereotypicality and perpetrator stereotypicality (all Fs >.955; see Table 2). As such, H2 was not supported.
Discussion
Having identified stereotypes of victim and perpetrators separate from gender stereotypes in our preliminary research, the aim of our main study was to experimentally examine the influence of these stereotypes in conjunction with offense stereotypes on perceptions of defendant and complainant behavior in a case of sexual assault. In line with the hierarchical schema model and consistent with McKimmie et al. (2014), it was predicted that victim stereotypicality would have the greatest effect on judgments in the counter-stereotypical (acquaintance) assault scenario compared with the stereotypical (stranger) assault scenario. More specifically, we expected that the counter-stereotypical victim in the acquaintance assault scenario would be perceived more negatively and the perpetrator more positively and less likely to be guilty, compared with when the victim was described as counter-stereotypical.
Results showed broadly consistent effects of offense stereotypicality and victim stereotypicality across the measures, such that participants perceived the victim more positively and the perpetrator more negatively when the offense was described as being stereotypical and when the victim was described as stereotypical. However, contrary to past findings (Masser et al., 2010; McKimmie et al., 2014), the effect of victim stereotypicality did not differ as a function of offense stereotypicality. As such, H1 was not supported.
Consistent with the defendant-crime congruency effect (Gordon et al., 1988), we also proposed that perpetrator stereotypicality would have the greatest effect on judgments in the stereotypical (stranger) assault scenario compared with the counter-stereotypical (acquaintance) assault scenario. Such that, a stereotypical perpetrator in the stranger assault scenario would be perceived more negatively and more likely to be guilty, and the victim would be perceived more positively, compared with when the perpetrator was depicted as counter-stereotypical. While perpetrator stereotypicality influenced perceptions of the perpetrator’s dangerousness, whereby participants thought that the perpetrator was more dangerous when he was stereotypical compared with counter-stereotypical, this pattern did not differ depending on the stereotypicality of the offense, nor did it hold for any of the other evaluative measures. As such, H2 was also not supported.
Although the main hypotheses were not supported, the manipulation checks revealed some intriguing results. Despite the success of the perpetrator stereotypicality manipulation in the second pilot study, this manipulation did not have the anticipated effect in the main study. Instead, in the main study, victim stereotypicality affected how the perpetrator was perceived. Specifically, the perpetrator was perceived as more stereotypical when the victim was stereotypical compared with when the victim was counter-stereotypical. While this may help to explain why perpetrator stereotypicality had little effect across the measures, it also suggests that when perceivers are asked to make judgments in the context of information about both the perpetrator and victim being presented simultaneously, victim stereotypicality contaminates perceptions of the perpetrator. For example, when participants were given information about just the perpetrator alone in the second pilot study, he was perceived as significantly stereotypical. Whereas when this same information was provided alongside information about the victim in the main study, it was the information about the victim that impacted his perceived stereotypicality.
The manipulation checks also revealed further unexpected spill over effects across the manipulations. An unexpected main effect of victim stereotypicality was found on perceptions of offense stereotypicality, as well as a main effect of offense stereotypicality on victim stereotypicality. These results suggest that when presented together, the manipulations of victim stereotypicality and offense stereotypicality influence participants’ perceptions on both of these dimensions. Although these findings are inconsistent with research that has suggested that these stereotypes are independent of one another (Masser et al., 2010; McKimmie et al., 2014), they appear somewhat consistent with a spreading activation model (Kunda & Thagard, 1996). According to this model, when a concept such as victim stereotypicality is activated, this activation can spread along to influence, or constrain, interconnected and related concepts, such as the stereotypicality of the perpetrator and the offense. For example, the force that the perpetrator uses is likely to look more aggressive and more like the behavior of a typical perpetrator if a victim is also described as physically resisting him. However, if the victim does not physically resist, this same behavior may be interpreted as being less aggressive and something more akin to “forceful passion.” Similarly, the counter-stereotypical victim stereotype may be activating consent-related schemas, making the offense look more like an acquaintance assault, and therefore more likely to be consensual. Given that the manipulation of perpetrator stereotypicality influenced perceptions of offense and victim stereotypicality to a lesser extent than the manipulation of victim stereotypicality, it is likely that that the “network” emanating from a stereotypical perpetrator may be less well formed.
The bivariate correlations calculated on the perceived stereotypicality variables provide further support for the possibility of a spreading activation model. This analysis showed that responses to the variables assessing perceptions of stereotypicality were positively associated. Such that, when the victim was perceived as stereotypical, the perpetrator was also perceived as more stereotypical, and so was the offense. This was also the case for the other combinations of the perceived stereotypicality variables. Taken together, the results of the pilot and main studies suggest that contrary to previous findings, participants do not “step down” through a hierarchy of individual schemas, with a differential focus on the victim and perpetrator, depending on the offense stereotype (Masser et al., 2010; McKimmie et al., 2014). Instead, they suggest that the distinctions between the independent stereotypes identified in previous research (Masser et al., 2010; McKimmie et al., 2014) and in the first pilot study may in fact be arbitrary. For example, it appears as though information about the victim and the offense may be simultaneously activating an overarching collection of consent-related schemas, which in turn may be influencing participant evaluations of the victim and perpetrator.
In line with the story model (Pennington & Hastie, 1992) and sexual script theory (Frith & Kitzinger, 2001), it is proposed that the more this collection of activated schemas matches a rape script, the less likely participants may be to blame the victim. Alternatively, the more the collection of activated schemas matches a consensual heterosexual sex script, the more likely participants may be to blame the victim. This explanation is in line with past research, which has found that there is considerable overlap between university student’s heterosexual rape and seduction scripts (Littleton & Axsom, 2003). In these rape and seduction scripts, respondents describe both situations as beginning with a man and a woman who are strangers toward one another. The man offers a compliment to the women, and following this, the man then engages in unwanted sexual activity with the woman (Littleton & Axsom, 2003). In addition, normative consensual sexual scripts are often male-initiated, male-dominated and contain elements of “token” female resistance (Littleton, Axsom, & Yoder, 2006). As such, it appears that in both the acquaintance and stranger assaults, participants’ may have relied on the victim’s response to the perpetrator’s advances to help them answer the legally relevant question of whether there was consent.
Strengths and Limitations
While this research represents an advance in our knowledge about how people may be using stereotypes about sexual assault to form evaluations of the assault, it is not without its limitations. Although the results of the pilot studies in conjunction with the main study allowed us to draw some conclusions about the individual and combined effects of the manipulations, the data from the current study did not allow for an unambiguous test of the spreading activation model. Based on the findings in past studies using the same or similar manipulations (Abrams et al., 2003; Masser et al., 2010; McKimmie et al., 2014; Schuller et al., 2010; Viki et al., 2004), we expected the manipulations used in this study to be independent of each other. As such, the second pilot study was designed to confirm that the specific operationalizations used in this study were effective, not whether they were independent. These pilot scenarios included only the information relevant for each of the manipulations. Given this, we are somewhat limited in terms of our interpretation of the results.
In addition, it is possible that the display order of the information about the victim and the perpetrator impacted on the results obtained. Given that the order of stereotype presentation has been found to impact impression formation (E. E. Jones & Goethals, 1972), it is possible that the manipulation of perpetrator stereotypicality may have had a stronger effect if this information was presented first in the scenario. For example, the primacy effect in judgment means that information acquired early on in a judgment process can disproportionally influence any information that is presented thereafter (Asch, 1946; Englich, Mussweiler, & Strack, 2005; Kerstholt & Jackson, 1998). It has been suggested that the primacy effect can result in a process of anchoring and adjustment, whereby perceivers use the initial piece of information about the target as an “anchor” and then adjust their interpretation about any following information in relation to this (Tversky & Kahneman, 1974). As the victim’s testimony was always provided before the perpetrator’s testimony, it is possible that participants may have “anchored” their focus on her behavior and then adjusted their perceptions about the perpetrator based on this. Given that the present research does not provide any data to differentiate between the spreading activation model and an anchor-and-adjustment process, it is suggested that this is an area that future research might focus on.
An additional consideration for future research regards the methodology used in the current study. In the interests of practical application, a move away from scenario-based descriptions toward more realistic materials may provide a more ecologically valid assessment of the influence of the experimental manipulations. While scenario-based methods allow for stricter experimental control, more naturalistic settings, such as mock video interviews or live trial simulations, may influence the extent to which the stereotypes are relied on for categorization (Anderson & Doherty, 2008). However, as past research has found that participants respond similarly to video and vignette manipulations of victim stereotypicality (Calhoun, Cann, Selby, & Magee, 1981; Winkel & Koppelaar, 1991), we would not expect this to effect overall conclusions about blame attributions. Further, due to the low number of male compared with female participants, gender effects were not assessed, as we could not reliably test for systematic effects. Although past research has found some differential effects between men and women on perceptions of victim blame, consistent moderation effects have not been found (McKimmie & Masser, 2010). As such, gender would not be expected to qualify the hypothesized interactions.
It is also worth noting that the current study only included scenarios in which there was a male perpetrator and a female victim. As such, our results, and the conclusions that can be drawn from them, are limited to this specific gender dyad. To address this limitation, future research should investigate how sexual scripts and stereotypes about gender further influence juror decision-making in cases of rape where there are other variations of victim and perpetrator gender.
Conclusions and Implications
The aim of the current research was to further disentangle victim and perpetrator stereotypes from gender stereotypes and expand on the current hierarchical schema model to experimentally examine the influence of offense, victim, and perpetrator stereotypes on perceptions of blame in a case of sexual assault. The results from the current study suggest that the relationship between offense, victim, and perpetrator stereotypes is more complex than previously thought. Rather than mock jurors’ engaging in a “step down” process through a series of hierarchical stereotypes to aid decision-making, the results are suggestive of a spreading activation model (e.g., Kunda & Thagard, 1996) or an anchor and adjustment heuristic process (e.g., Tversky & Kahneman, 1974). They suggest that it is not a series of specific, individual stereotypes that impact attributions of blame, rather that there may be one underlying schema about consent that influences perceptions. It appears as though mock jurors are trying to answer the legally relevant question of whether there was consent in both the stranger and the acquaintance assaults. At a broader level, this research may also provide some support for education and public policy initiatives, which are designed to target problematic beliefs within current heterosexual consensual sex scripts. For example, there are a number of campaigns and legislative initiatives that have begun to adopt an “affirmative consent” approach to sexual scripts, framing consent as an enthusiastic “yes,” rather than as a lack of resistance (Coy, Kelly, Vera-Gray, Garner, & Kanyeredzi, 2016). Given that the results of the current research suggest that the existing “no means no” model of consent does not sufficiently reduce third-party perceptions of blame, even for a stereotypical assault, an approach such as this may help to challenge the idea of women’s “token” resistance and reduce the overlap between rape and heterosexual seduction scripts. However, at this stage further work is needed. Future research should seek to systematically test the spreading activation and anchor and adjust frameworks to determine the relationship between these stereotypes and the direct impact that they may have on juror decision-making outcomes.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by an Australian Research Council Discovery Grant (DP1093581).
Author Biographies
