Sharing the Responsibility: Victim Blaming as a Function of Crime Type and Victim Behavior ()
1. Introduction
People often blame victims of crime for their plight despite knowing better, using emotion more than logic to come to conclusions about culpability (Alicke, 2000) . While situational variables influence how blame for crimes is allocated to victims and to perpetrators, several inter- and intrapersonal characteristics of victims and perpetrators—as well as the persons making judgments of various crime scenarios—also influence the allocation of blame. Such characteristics include victim behavior (Felson & Palmore, 2018) , sexist stereotypes of victims (Capezza & Arriaga, 2008; Wiener et al., 2021) , and Beliefs in a Just World (Lerner & Miller, 1978; Lucas et al., 2007) . Additionally, people may not explicitly blame the victim but instead do so tacitly as a function of situational ambiguity and type of crime (Hafer et al., 2019) .
Crime type is important in judgments of how culpable a perpetrator is, as well as how responsible a victim is (Bieneck & Krahé, 2011; Felson & Palmore, 2018) . The majority of empirical work in the area of victim blaming focuses on sexual crimes. Research is mixed with regard to attributions for blame in rape, with some research (e.g., Brems & Wagner, 1994; Felson & Palmore, 2018 ) showing that rape victims are not uniformly assigned the most blame compared to victims of other crimes. For example, Felson and Palmore (2018) presented participants with scenarios that ended in various crimes (e.g., rape, robbery, homicide) and the participants attributed blame to the victim both indirectly and directly. Direct victim blaming occurred less in scenarios of rape than they did in robbery, supporting the notion that the type of crime is a factor contributing to the likelihood we will blame victims. Yet, other work (Beineck & Krahe, 2011) showed that rape victims were seen as more blameworthy and the perpetrator less blameworthy when compared to a robbery involving the same interactants. The victim was seen even more responsible—and the perpetrator less culpable—when there was a previous relationship between the two interactants, but only for the sexual crime.
What situational or personal variables lead to these differences in how much or whether victims are blamed? Wiener et al. (2021) suggest the degree of blame given to victims may, in part, be a function of the sex or gender of the person making the judgment. For example, men more than women blame victims (Felson & Palmore, 2018) , especially in cases involving sexual assault (Grubb & Harrower, 2009) , and people who exhibit more hostile sexism blame victims of acquaintance (although not stranger) rape more than those who do not show hostile sexism (Persson et al., 2018) . The work of Brems and Wagner (1994) showed that people with strong traditional sex-role stereotypes were more likely than those with nontraditional views to blame women for their victimization in unclear or ambiguous situations, whereas those with more nontraditional views assigned more responsibility to perpetrators. Indeed, we may be more empathetic to victims who seem similar to us or who have similar views (Putra et al., 2018) , but we are more punitive in cases where we rely on stereotypes of others. For example, Capezza and Arriaga (2008) showed that when a crime victim is a non-traditional woman, she was perceived negatively and given responsibility for her victimization because she was not sufficiently warm. Similarly, a rape victim who was a gay young man was seen as more responsible for his victimization than women in similar circumstances (Davies et al., 2009) . Sex trafficked survivors were more likely to be blamed if they engage in sex work as they were not seen as vulnerable due to their job status (Wiener et al., 2021) . Such findings suggest that the way people feel toward victims (particularly women victims) contributes to the attribution of blame and responsibility toward victims and offenders.
A victim’s behavior and relationship to a perpetrator is evaluated when deciding blame for crimes (Landstrom et al., 2016) . For example, victims of date rape are perceived to be more at fault than those of stranger rape (Pedersen & Stromwall, 2013) , but less at fault than women victims of seduction rape (Grubb & Harrower, 2009) . Even having a prior relationship with a perpetrator increases victim blame—but only for sexual crimes (Bieneck & Krahe, 2011) . Landstrom et al. (2016) examined how victim behavior affected the distribution of blame between victims and offenders. Participants read a scenario of a victim being sexually assaulted after being flirtatious or non-flirtatious, or sexually harassed online after online flirting (or not). Participants ascribed more blame to offenders than to victims in both cases, but also attributed more responsibility to victims of less serious crimes (online harassment) if those victims were being flirtatious.
Several researchers (e.g., Chapin & Coleman, 2017; Harber et al., 2015 ) have noted that Belief in a Just World (BJW) is related directly to victim blaming. BJW is defined as the idea that consequences are a function of actions, such that good people generally have positive outcomes, while people reap negative consequences for bad behavior (Lerner & Miller 1978) . The construct reflects the extent to which people see others’ outcomes as a function of intent, effort and the idea that get what they deserve. Research suggests that believing in a Just World reflects an aim for a controlled (and controllable) life, reducing the threat of being punished for something you did not do or being rewarded inappropriately (Hafer & Bègue, 2005) . The original BJW construct and corresponding measurement has been divided into two constructs: procedural justice and distributive justice (Lucas et al., 2007) . The procedural justice aspect focuses on the belief that rules and processes of the world are fair, whereas distributive justice refers to the belief that the outcomes of these rules are fair on a personal level, i.e., people get what they deserve (Lucas et al., 2007) . Lerner and Miller (1978) suggest that strong BJW directly affects how we react to the suffering of others, as well as how we determine if they are innocent or guilty in certain situations; BJW is thus a contributing factor in victim blaming. Specifically, even when logic tells us otherwise, we tend to see that victims are to some degree responsible for their misfortune.
BJW has predictable outcomes for judgments of responsibility for crime. As noted, many researchers show that as BJW increases so, too, does victim blaming (Hafer et al., 2019; Landstrom et al., 2016; Pederson & Strömwall, 2013) . In turn, victim blaming can lead to justifications for poor treatment of victims, such a bullying (Chapin & Coleman, 2017) . BJW can increase victim blaming because dissonance results when considering another’s bad fortune in relation to just world beliefs, and dissonance is resolved by blaming a victim. However, if people describe, ruminate, and discuss what they think and feel about the victim’s misfortunes, victim blaming is reduced (Harber et al., 2015) , perhaps because people put themselves into another’s shoes and begin to feel empathetic.
In sum, victim blame is likely under many circumstances: when it is possible that a person could have done something to prevent victimization or acted in a way that increased its likelihood, when crimes are not very serious, when people hold strong stereotypes about members of the victim’s group, and when there is a high BJW. Our study examined several of the aforementioned variables to address questions about attribution of responsibility to a victim when she was swindled into losing money compared to when the same amount of money was lost due to theft. Moreover, we manipulated the woman victim’s behavior to have a short-term (possibly sexual) relationship with the perpetrator or not. We predicted that when the victim acted in ways that allowed increased her chances of being victimized she would be seen as more culpable than if she had not, particularly if she had a short-term relationship with the perpetrator.
2. Method
2.1. Participants and Design
A total of 91 college students (43 men, 48 women) were assigned randomly to read a scenario of either a burglary or swindle allegedly committed by a man. The woman victim was reported to have a flirtatious interaction with the man earlier and then after the interaction either invited him to her dorm room afterwards interacting or not. These manipulations thus resulted in a 2 × 2 (Crime Type: Swindle, Burglary × Victim Behavior: Return to Room, No Return to Room) between-participants design.
2.2. Stimulus Materials
Two scenarios portrayed the accused man (Daniel) breaking into a dorm room of a woman (Kylie) and stealing $2000 worth of Apple products the day after he had a flirtatious interaction with the victim. However, one scenario clearly noted that the woman invited the accused to her dorm room after their interaction; the other did not. The other two scenarios described the target man stealing the victim’s credit card information after he convinced her to allow him to use the card to pay for a meal. He spent $2000 on Apple products. The scenarios are shown in the Appendix.
2.3. Dependent Measures
Recall. After reading the scenarios, participants completed four multiple-choice questions asking what happened in the scenario to ensure that they were attending to the details. These were sufficiently straightforward that we believe that they also served to prompt participants to recall salient details. The questions were the exact same for each scenario, and focused on worth of items in the crime, items stolen, what the man did, and the actual charge levied.
Perceptions of guilt/responsibility. Questions to assess the guilt, likelihood of similar criminal behavior, and responsibility of the people involved were taken individually on seven-point scales with opposite-meaning points. Questions tapped the likelihood Daniel had committed similar crimes in the past (1 very likely to 7 not at all likely), likelihood Daniel may escalate his crimes to become more serious (1 not at all likely to 7 very likely), how aggressive Daniel was (1 very to 7 not at all), how responsible Kylie was for the crime (1 not at all to 7 very), and the likelihood Kylie would learn from the situation (1 very likely to 7 not at all likely). We also gave a final two-response option questions asking how Daniel should be sentenced should he be found guilty: with probation, or jail time.
Belief in a Just World Scale (JWS; Lucas et al., 2007 ). The Just World Scale (JWS) has two subcategories of distributive justice and procedural justice, both of which address the extent people see others’ outcomes as a function of intent, effort and the idea that “people get what they deserve.” Procedural justice taps how people perceive the rules of life and if they are fair; the distributive justice subscale assesses whether people see that equitable outcomes exist and are spread along people. There were 10 questions, five for procedural and five for distributive, that participants answered on a seven-point scale with endpoints 1 strongly agree to 7 strongly disagree. For example, questions assessed whether people feel that others generally earn what they get in the world, receive outcomes they deserve, usually receive fair outcomes, they deserve the things they are accorded, receive the outcomes they are due, are treated according to fair processes, that people use fair procedures in dealing with others, and are evaluated with fair methods. The questions were present in one of two counterbalanced random orders within conditions. There was internal reliability at α = 0.90 for distributive justice and α = 0.89 for procedural justice.
2.4. Procedure
Participants accessed a link to Kwiksurveys, where they were presented with directions and consent. Following consent, they read one of four scenarios after being prompted to read carefully because they would be unable to return to the scenario after they moved on to the recall questions. After the recall questions, participants completed the major dependent measures, and then moved on to the JWS. There were two counterbalanced presentations of the dependent measures and the JWS, within condition. Finally, participants provided their self-characterized gender and/or sex and their ethnicity/race. After all data were collected and analyzed, students were debriefed via email.
3. Results
3.1. Preliminary Analyses
We first examined the effects of crime type and victim behavior on the three judgments of the perpetrator and the two judgments of the victim in two separate 2 × 2 × 2 (Participant Sex × Crime Type: Swindle, Burglary × Victim Behavior: Room Invite/No Room Invite) Multivariate Analyses of Variance (MANCOVA), holding constant recall and both of the JWS scales. None of the covariates was significant in either MANCOVA, and the pattern of results found paralleled exactly our results without the covariates (reported below). Moreover, neither JWS D or P were significantly related to victim blame, both rs(93) < 0.01, both ps > 0.92. Thus, JWS and recall were dropped from remaining analyses. There were also no significant main or interactive effects with participant sex, so this variable was dropped from the foregoing analyses.
3.2. Influence of Crime Type and Victim Behavior on Perceptions of the Perpetrator
To examine the effects of crime type and victim behavior on past criminal behavior, escalation of future crimes, and aggression of the perpetrator, a 2 × 2 (Crime Type: Swindle, Burglary x Victim Behavior: Room Invite/No Room Invite) Multivariate Analysis of Variance (MANOVA) was calculated. The analysis showed a significant multivariate main effect for crime type, F(3, 89) = 5.24, p = 0.002,
= 0.15, for whether Kylie invited Daniel to her dorm room, F(3, 89) = 11.53, p < 0.001,
= 0.28, and an interaction, F(3, 89) = 9.85, p < 0.001,
= 0.25. The univariate means and standard deviations are located in Table 1.
Follow-up univariate tests were conducted using a Bonferroni correction, setting the acceptable alpha level at 0.017 for three dependent variables. The likelihood of Daniel having similar past criminal behavior as a function of crime type was not different if he purportedly committed a burglary or a swindle, F(1, 91) = 5.44, MSE = 2.27, p = 0.022, but he was viewed as having done a similar crime previously if he also happened to have been invited to the room, F(1, 91) = 27.12, MSE = 1.86, p < 0.001,
= 0.23. However, this effect was qualified by a significant interaction, F(1, 91) = 12.38, MSE = 2.02, p < 0.001,
= 0.12. This interaction is depicted in Figure 1. Post-hoc comparisons showed that participants believed that Daniel was more likely to have committed a swindle in the
![]()
Table 1. Means and standard deviations for attributions to the perpetrator as a function of crime type and room invitation.
Note. Means with different subscripts differ at p < 0.05.
![]()
Figure 1. Means and 95% CIs for attributions of similar past behavior on the part of the perpetrator.
past when invited to Kylie’s room (M = 5.42 [95% CI 4.81, 6.03] vs. 2.72 [95% CI 2.12, 3.32] for no invitation), p < 0.01, but no differences in judgments of whether this was a recurring behavior pattern were seen among those who read that Daniel committed the burglary (Ms = 3.09 [95% CI 2.46, 3.71] for no invitation vs. 3.61 [95% CI 2.99, 4.23] for an invitation), p = 0.20.
The perception that Daniel would escalate his behavior if not caught was higher if he had committed a burglary rather than a swindle, F(1, 91) = 9.45, p = 0.003,
= 0.09, but not influenced by whether he had been invited back to the room, F(1, 91) = 4.41, p = 0.039. A significant interaction qualified the Crime Type main effect, F(1, 91) = 17.40, p = 0.008,
= 0.08. Post-hoc testing showed that concern about escalation of criminal behavior was not influenced by whether there was an invitation to the room—but only when the crime was a burglary (Ms = 4.74 for no invitation and 4.91 for an invitation), p > 0.50. If the crime was a swindle then participants perceived criminality would escalate if he was not invited to the room (M = 4.64 [95% CI 4.10, 5.18]) more so than if he were (M = 3.29 [95% CI 2.74, 3.84]), p < 0.01.
Judgments of how aggressive Daniel is was not influenced by either crime type (swindle or burglary) or victim behavior, both Fs(1, 91) ≤ 0.73, both ps ≥ 0.395. However, the interaction was significant, F(1, 91) = 6.22, p = 0.014,
= 0.06. Once again, post-hoc testing located meaningful differences when the alleged crime was a swindle rather than a burglary. Specifically, concern for how aggressive the perpetrator was when there was no room invitation was higher than when there was (Ms = 4.56 [95% CI 4.00, 5.13] vs. 3.58 [95% CI 3.01, 4.16]), p < 0.04, but there were no differences when the crime was a burglary (Ms = 3.87 [95% CI 3.55, 4.38] for swindle vs. 4.35 [95% CI 3.81, 4.62] for burglary), p = 0.25.
3.3. Influence of Crime Type and Flirting on Responsibility and Likeliness to Learn for the Victim
We examined the effects of crime type and victim behavior on victim responsibility and her likelihood of learning from the situation for the victim in a 2 × 2 (Crime Type: Swindle, Burglary × Victim Behavior: Room Invite/No Room Invite). There was a significant multivariate main effect of crime type, F(2, 90) = 32.88, p < 0.001,
= 0.42, victim behavior, F(2, 90) = 23.28, p < 0.001,
= 0.34, and the interaction, F(2, 90) = 14.03, p < 0.001,
= 0.24. The univariate means and standard deviations are located in Table 2.
We used a Bonferroni correction for two dependent measures, set to alpha = 0.025, for our follow-up univariate tests. Means and standard deviations from this analysis are located in Table 2. For attributions of victim responsibility, there was a significant main effect of crime type, F(1, 91) = 44.09, MSE = 2.12, p < 0.001,
= 0.33, as well as victim behavior, F(1, 91) = 5.30, p < 0.024,
= 0.06. The woman victim was seen as more responsible when she was swindled rather than burgled (Ms = 4.09 [95% CI 3.68, 4.51] vs. 2.11 [95% CI 1.68, 2.54], respectively), and also more to blame for her situation when she invited the man perpetrator to her room (Ms = 3.47 [95% CI 3.02, 3.87] vs. 2.79 [95% CI 2.34, 3.17], for invitation vs. no invitation), although this latter effect was small. The interaction was not significant, F(1, 91) = 0.98, p = 0.324, indicating that the victim’s behavior was scrutinized in the same manner for crime type.
The analysis focusing on how much the victim learned from her behavior produced three significant effects. There was a significant main effect for crime type, F(1, 91) = 21.41, MSE = 3.06, p < 0.001,
= 0.19, which revealed that participants perceived the woman victim would be more likely to learn from the experience if she had been swindled rather than burgled, and the main effect for victim behavior showed that she would learn more from the experience if she had invited him to her room rather than if she had not, F(1, 91) = 41.31, p < 0.001,
= 0.31, perhaps suggesting that participants saw her behavior as contributing to her victimization. However, these effects were qualified by a significant interaction, F(1, 91) = 27.22, p < 0.001,
= 0.23. This interaction is shown in Figure 2. Post-hoc tests showed that participants did not believe that the woman victim would differentially learn from the situation if the crime in question was a burglary, regardless of whether she invited the man to her room
![]()
Table 2. Means and standard deviations of crime type and amount of flirting on responsibility and likeliness to learn for the victim.
Note. Means with different subscripts differ at p < 0.05.
![]()
Figure 2. Means and 95% CIs for attributions of learning from the incident on the part of the victim.
(M = 3.09 [95% CI 2.36, 3.81]) or if she did not (M = 2.65 [95% CI 1.93, 3.38]), p > 0.25. But if the crime was a swindle, she was seen as more likely to learn from the situation if she had invited the man to her room (M = 6.63 [95% CI 5.92, 7.34]) than if she had not (M = 2.44 [95% CI 1.75, 3.15]), p < 0.001.
3.4. Consequences for the Perpetrator as a Function of Crime Type and Victim Behavior
Two separate chi-square tests were used to examine whether participants were more likely to assign probation or a jail sentence based on crime type and victim behavior. Neither the test for crime type, χ2(1, N = 95) = 0.07, p = 0.793, nor the test for victim behavior, χ2(1, N = 95) = 4.48, p = 0.115), was significant.
4. Discussion
Our results showed that crime type and victim behavior played a role in judgments of both the victim and the alleged perpetrator. Across all measures of perceptions of the perpetrator, judgments were affected only by whether he committed a swindle, in which case he was seen as less aggressive and less likely to escalate his crimes if he was invited to the room, but more likely to have done the same deed in the past if he had. The results suggest that people perceived his swindle to be a crime of convenience, with him taking advantage of a person who allowed herself to be taken advantage of. Across both measures about the victim—her responsibility and whether she would learn from her experience— no differential attributions were seen when she was burgled, but if she was swindled and invited the perpetrator to her room she was seen as more responsible but also more likely to learn from the situation than if she had not.
Our results harmonize with previous research (Landstrom et al., 2016) related to the severity of a crime as it relates to judgments of blame—specifically, a very serious crime typically does not lead to a lot of victim blame. Although the same items that were worth the same amount of money were taken in each case, the crime of burglary was likely deemed more serious. The swindle may have been seen as a crime of opportunity, rather than one of planning, and in that case the woman victim was perceived as having some responsibility by inviting the man to her dorm room. Yet, paradoxically, participants did not differentially assign probation or jail time based on the crime.
Our results also support the idea that victim blame is more likely when the victim has had a relationship with the perpetrator, although typically such blame is only likely when the crime is related to sex (Bieneck & Krahe, 2011) . It is possible that many of our participants (college students) empathized with the victim and related to her and her situation, and thus did not see her as culpable for the burglary. Dorm room burglaries of electronics are not uncommon and perhaps participants could put themselves in her shoes, showing more empathy (see Harber et al., 2015 ), reasoning that something like that could have happened to anyone (Putra et al., 2018) .
We had hypothesized, based on our literature review, that victim blame would be higher if the victim had a relationship with the perpetrator, yet such blame was seen only for the swindle. That is, when the victim may have implicitly had a sexual relationship with the perpetrator, she was blamed more for being swindled, a finding that aligns with that of Landstrom et al. (2016) , who showed that for non-serious crimes, if flirting was involved, the victim is blamed more. Having a short-term sexual liaison with the perpetrator could be considered as a mistake she made in a situation completely under her control, and that would allow her to be seen as more blameworthy (Felson & Palmore, 2018; Hafer et al., 2019) . Future research may address whether these patterns hold only when a woman is the crime victim and a man is the perpetrator, rather than with some other combination of interactants.
Belief in a Just World (BJW) did not change our results, despite that research has shown a clear relationship between higher levels of BJW and victim blaming (Hafer et al., 2019; Pederson & Strömwall, 2013) . While BJW can lead to victim blame, it can also lead to more negative judgments of a victim in general (Chapin & Coleman, 2017) . Therefore, we would have expected to see an overall negative halo toward the woman victim as a function of BJW. Yet, our participants’ BJW scores were normally distributed (i.e., skewness −0.33 and −0.41 for distributive and procedural, respectively), and generally in the middle of the 7-point scale (Ms = 3.85 and 4.09 for distributive and procedural, respectively). One limitation of our study is the timing of the BJW scales, which occurred after the reading of the scenario and completion of the measures about the victim and perpetrator. This presentation order may have created experimental demand: participants looked at questions concerning whether the rules of life are fair, whether people get the outcomes that they are due, or that legal processes are generally equitable and may have been cued to not express strong opinions after having made evaluations of a crime where the victim may have been perceived as being (at least in part) culpable. To better examine the relationship of BJW and blame, BJW should be assessed a different way or at a different time, or perhaps the target scenarios could have been created with more ambiguity for victim behavior.
There are other personal variables about participants that we did not measure that may help shed light on the differential patterns of blame ascription based on crime type and victim behavior, and which should be considered in future research. First, our sample was comprised solely of college students, whose perceptions of the similar-aged perpetrator and victim may be more lenient than older persons because they can better empathize with the situation. Next, we could have measured sexism (as did Persson et al., 2018 ) and sex-role stereotype adherence (Brems & Wagner, 1994) , which could have increased blame when the victim was swindled after she had invited the perpetrator to her room, particularly in men participants. Future research may also focus on implicit measures, such as personal narratives or explanations, which could help to locate differences in attributions for culpability because people are unlikely to provide explicit blame to victims because they know that they should not appear biased (Hafer et al., 2019) . Yet our results (using explicit measures) were strong and showed clearly that participants perceived that a woman might be partially at fault for her victimization if she allowed herself to be manipulated after a (likely) sexual encounter, but they also recognized that she did not deserve blame when she was robbed, despite her likely sexual encounter. Thus, our findings indicate that while some progress has been made reducing blame to women crime victims, people nonetheless decided that responsibility for some crimes is shared by the victim and perpetrator, despite that the crime was ultimately the sole responsibility of the one who committed it.
Appendix: Scenarios
Both scenarios: Daniel Smith and Kylie Johnson attend different colleges and were attending a campus party at Kylie’s school and got into several long conversations at the night went on. They both had a lot to drink. Kylie and Daniel were flirting with each other pretty strongly and eventually they went back to Kylie’s room.
Swindle scenario: At one point Daniel convinced Kylie to allow him to use her debit card to pay for a meal. He then stole all of her card information and proceeded to use her money to spend $2000 on a MacBook computer and airpod headphones. He was identified and charged with grand larceny after public safety reported the incident to the police.
Burglary scenario: The next day Daniel broke into Kylie’s dorm room and stole $2000 worth of Apple products, including a MacBook computer and airpod headphones. After stealing the merchandise, Daniel tried to sell his fellow peers the stolen belongings. He was identified and charged with grand larceny after public safety reported the incident to the police.