The Impact of Class and Sexuality-Based Stereotyping on Rape Blame

On any given weekday, on any given basic television channel, one can find a number of television programs that portray a stereotypical version of low-socioeconomic (SES) women’s lives and issues. These shows mainly include “real-life” court shows, talk shows that focus on paternity testing and/or cheating spouses, and most recently, Here Comes Honey Boo Boo. ¹ Upon viewing such programs, one is left with an indelible image of a low-SES woman. She is loud. She is stupid. She is promiscuous. She is either overweight and unattractive or blonde and sexy in a “trashy sort of way.” She can easily have 10 children, but not know who any of the fathers are.

This form of programming supports the main stereotypes about low-SES women that they are promiscuous, fertile, and immoral (Cozzarelli, Tagler, & Wilkinson, 2002), and in fact, have children in order to gain support from the state (Bullock, 1995). These classist stereotypes often interact with racist and sexist depictions as well (Bullock, Wyche, & Williams, 2001), usually resulting in a portrayal of a neglectful, oversexed woman. Even when she is characterized in a well-intentioned, sympathetic manner, she is still seen as having too many children with little control over her sexual desires. Although the stereotypes about the low-SES woman serve as entertainment, they simultaneously serve as a constant reminder that she is still less moral than her high-SES counterpart, and only worthy of derision. Take again, for example, Here Comes Honey Boo Boo. In this reality program, “Mama June,” who has four children from three men, serves as the real focus of the show, serving “sketti” (pasta, ketchup, and butter) to her children, including a teenage daughter who is pregnant. Throughout the series, we see Mama June date, have a pregnancy scare, and give her daughters “sex talks” although one of her daughters is already a mother. Throughout the series, we are both implicitly and explicitly reminded of Mama June’s (and, to some extent, her daughter’s) promiscuity. Despite averaging millions of viewers (Hare, 2014), The Learning Channel (TLC) was forced to cancel the show after it was revealed that Mama June was dating a convicted child molester. What was once entertaining “trash” became too aversive for viewers and advertisers. Both the content and downfall of the show reinforce the same message, that low-SES women are promiscuous and immoral.

Although low-SES women are stereotyped as being sexually promiscuous, to date, there has been no direct psychological research investigating the impact of these stereotypical images. Namely, how classism and sexual stereotyping may relate to perceptions of low-SES women in situations where perceived sexuality may be an important factor in making judgments about them. One such situation may be in cases of sexual assault and rape. Scholars have shown that sexual stereotyping can lead to justifications of sexual assault (George & Martinez, 2002; Greene, 2000). Ferguson and colleagues (2005) found that participants who were exposed to television images of promiscuous women later perceived a survivor ² of sexual harassment as less traumatized, and more blameworthy, than participants who had previously been exposed to images of nonpromiscuous women. Given the stereotypes about low-SES women’s sexuality, the role perceived promiscuity plays in perceptions of survivors, and the fact that low-SES women are far more likely to be survivors of sexual assault than high-SES women (Bureau of Justice Statistics, 2008), it is important to examine the role stereotyping plays in the perception of low-SES women who are survivors of sexual assault. Although there has been no direct research on this topic to date, it is possible that because low-SES women are often perceived as being promiscuous, and that alleged promiscuity leads to more victim blame in cases of sexual assault, that classist stereotypes about sexuality could lead to more victim blame and derogation for low-SES survivors. Additionally, preexisting classist beliefs may moderate these effects, so that a person who scores high in classism may hold the most negative views of a low-SES survivor.

Victim Blame and Classism

Impression formation plays a large role in the courtroom, where many traits of perpetrators and survivors can affect public opinion as well as judge and jury decision-making. In the case of rape, when a survivor knows his or her attacker (Pollard, 1992), is a gay man or a lesbian (White & Kurpius, 2002), or is unattractive (Gerdes, Dammann, & Heilig, 1988), he or she is seen as more culpable in the attack and is attributed more blame for the rape. This form of victim blaming and questioning of the survivor’s credibility is referred to as secondary victimization and often prevents rape survivors from prosecuting their attackers (Ullman, 1996). Literature investigating attitudes toward rape survivors shows that women are more sympathetic and less likely to endorse rape myths than men (Struckman-Johnson & Struckman-Johnson, 1992), as are Whites compared to Latinos (Jimenez & Abreu, 2003), Asian Americans (Lee & Cheung, 1991; Yamawaki & Tschanz, 2005), and African Americans, although these differences become nonsignificant when education and income are controlled, suggesting that higher education and higher income lead to more sympathetic views (Nagel, Matsuo, McIntyre, & Morrison, 2005).

Comparing interracial and intraracial rape scenarios with Black and White perpetrators and survivors, George and Martinez (2002) found that higher levels of racism in male participants predicted more victim blame overall, whereas for female participants, levels of racism moderated victim blame in cases of interracial acquaintance rape only. Overall, in cases of interracial rapes, female survivors were seen as less credible and more culpable than survivors of intraracial rape. The authors argue that stereotypes about African American sexuality led to these effects, because female African American survivors may be seen as promiscuous, and White female survivors of African American perpetrators may be stereotyped as being “less reputable” than White women who do not socialize with African American men.

Differences in survivor and perpetrator status can also influence judgment about the rape and the rape survivor. Yamawaki, Darby, and Queiroz (2007) examined the role of status, as operationalized by education and career differences, in cases of heterosexual rape. To assess such status differentials on perceptions of rape, participants read one of two scenarios in which a male assailant and female survivor varied in status. In the first condition, the assailant had a master of business administration degree from an elite university and was a CEO of a successful company, and the survivor was a student from a local university. In the second condition, the survivor’s and assailant’s credentials were reversed. Following the scenario, participants completed a series of materials that measured victim blame, perpetrator culpability, rape minimization, recommended sentencing as well as hostile and benevolent sexism. Results suggested that when the male attacker was of higher status than the female survivor, the survivor was more likely to be blamed for the rape and the rape was more likely to be minimized. Both effects were moderated by hostile sexism, with participants who scored higher in hostile sexism showing more negative attitudes toward the rape survivor.

Similar to Yamawaki et al. (2007), other researchers have investigated the role of hostile and benevolent sexism for moderating effects of victim blame in cases of rape. The endorsement of benevolent sexism and traditional gender roles can lead to more victim blame (Abrams, Viki, Masser, & Bohner, 2003; Viki & Abrams, 2002; White & Kurpius, 2002), as can other beliefs and attitudes, such as levels of racism, in cases of interracial rapes (George & Martinez, 2002), and ethnic prejudice and misogynistic attitudes in cases of interethnic rapes (Murthi, 2009).

One such attitude that has not been widely studied in regard to victim blaming is classism. Courtrooms are a setting in which classism influences judgments of low-SES individuals. Previous research has shown that male defendants who are presented as low-SES as compared to high-SES are seen as more blameworthy (Gleason & Harris, 1976) and receive harsher punishment (Mazzella & Feingold, 1994; Osborne & Rappaport, 1985). Individual factors, such as a belief in a “just world,” may moderate such classist effects, so that participants who endorse more just world attitudes believe low-SES defendants to be guiltier than high-SES defendants (Freeman, 2006).

Although it has been established that classism can influence perceptions of defendants (perpetrators) in the courtroom, it has not been established whether there is a similar effect for plaintiffs (survivors). Even though the role of classism in the courtroom as it relates to the survivor has not been widely studied, I contend that participants’ level of classism will play a moderating role in how they judge low-SES survivors of sexual assault. Given that people who score high in classism are endorsing commonly held stereotypes of low-SES people and that one of the dominant stereotypes about low-SES women is that they are promiscuous, classist participants may feel that a low-SES rape survivor is more blameworthy than her high-SES counterpart.

These theories translate into a series of hypotheses that were tested in an experimental study in which participants read about a rape where the survivor was either low or high-SES and then completed measure of victim blame, attitudes toward rape victims, and sexual stereotyping. Participants also completed measures of classism. Additionally, because individuals develop and hold attitudes about others that can be explicit or implicit (Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997; Greenwald & Banaji, 1995), I deemed it important to go beyond self-reported attitudes and included an implicit attitude measure consisting of linguistic analysis of one open-ended question. It was this variable that I predicted may be moderated by classism because it was a more subtle, indirect measure, although this hypothesis was exploratory due to the lack of previous research on classism. Taken together, there are five specific hypotheses:

Hypothesis 1: Participants who read about the low-SES woman will assign more blame to the survivor than participants who read about the high-SES woman.

Hypothesis 2: Participants who read about the low-SES woman will show more negative attitudes toward rape survivors in general than participants who read about the high-SES woman.

Hypothesis 3: Participants who read about the low-SES woman will rate the survivor as more promiscuous.

Hypothesis 4: Participants who read about the low-SES woman will show more linguistic minimization in their description of the rape as compared to participants who read about the high-SES woman and this effect will be moderated by participants’ level of classism.

Method

Participants

The study was run online and participants either responded to advertisements placed on a popular advertisement website or were contacted through an online data collection service. A total of 131 respondents with valid entries were obtained, who varied with respect to ethnicity (2% American Indian, 14% Asian, 6% Black, 9% Latina/o, 59% White, 4% Multiracial, 5% preferred not to answer, and 2% were of another ethnicity that was not specified) and approximately two thirds were women.

Materials

Police file

Participants read one of two paragraph-long police files that described a heterosexual rape of either a low-SES or a high-SES woman. The vignettes were modified from George and Martinez (2002) and depicted a situation in which the survivor knew her attacker. They were manipulated to subtly convey the target’s SES by changing the target’s job (cashier vs. accountant) and housing (apartment vs. house).

On the evening of April 28, 2006, “C.S.” returned home from the store where she works as a cashier (business where she works as an accountant). Upon entering her apartment (house), she noticed that her cat was nowhere around. C.S. went outside to look for the cat. While outside a man, “D.M.,” whom she had met the previous weekend approached her. She invited him in for a coffee, where they spoke for about half an hour. At the end of the conversation, C.S. excused herself from the table and informed D.M. that she had to go to bed soon since the following day she was scheduled for a 10-hour work shift (she was scheduled for a long day full of meetings). At this point D.M. became insistent that they talk a little longer, and pushed her towards the bedroom. C.S. resisted and tried to hit him, and as a result D.M. pulled her and physically forced her in to the bedroom. C.S. states that D.M. continued to physically and sexually assault her for 30 minutes, culminating in rape. D.M. fled C.S.’ apartment (house) when the landlord (neighbor) started knocking on the door, responding to C.S.’ cries. On April 30 C.S. reported the incident to the police.

Victim blame

The Case Reaction Questionnaire (CRQ; Schult & Schneider, 1991) included nine questions that measured participants’ perceptions of both survivor (“To what extent do YOU think [the survivor’s] behavior contributed to the rape?”) and perpetrator blame (“If [the survivor] had come to talk to you, how much would YOU want [the survivor] to blame what happened on the assailant?”). Participants responded on a 1–7 Likert-type scale (1 = not at all, 7 = very much), thus scores could range from 9 to 63. A factor analysis on the CRQ Scale yielded one main factor accounting for 46% of the variance, on which 6 items of victim blame loaded strongly. A composite score, called blame, with these 6 items was thus created (α = .92) and used as the main blame measure. As its distribution was highly skewed (1.77), I applied a logarithmic transformation, which led to a less skewed distribution (.82). Subsequent analyses use this transformed variable, called t_Blame.

Attitudes Toward Rape Victims Scale (ARVS)

The ARVS (Ward, 1988) was selected because it specifically measures attitudes toward survivors of rape, not attitudes toward rape in general. The scale consists of 25 statements that assess attitudes about the credibility and sympathy toward rape survivors (“a raped woman is a less desirable woman.”). Participants responded on a 1–7 Likert-type scale (1 = strongly disagree, 7 = strongly agree); thus scores could range from 25 to 175 and were summed to create a composite score (α = .89).

Perceived promiscuity

In order to assess the role of stereotyping, participants reported on a 0–100% scale the probability (see Deaux & Lewis, 1984) that the survivor was promiscuous.

Open-ended response

Participants completed one open-ended question. The question was meant to gauge their initial response to the police file and the way that they described the events. The question simply read, “Please write down what you remember about this incident. You can write anything that comes to mind about the case, the people, or the events.” Participants were not given any set length or time requirements. Coh-Metrix analysis, with software version 1.4, (McNamara, Louwerse, Cai, & Graesser, 2005) was used to analyze the responses. This software analyzes the text and provides each participant a score (proportion) of the relative use of a variety of linguistic categories, eight of which were considered here and can be distinguished into positive and negative connectives. Positive connectives include positive additive (e.g., also, moreover), positive temporal (e.g., again, even then), positive logical (actually, if), and positive causal (consequently, because), reflecting a linguistic extension of the situation, that is, a tendency to maximize the significance of the events. The four negative connectives, negative additive (e.g., however, but); negative temporal (e.g., then, until); negative logical (e.g., anyhow, despite); and negative causal (e.g., although, nevertheless), reflect a minimization of the situation. Proportions were heavily skewed and no transformation improved the situation, thus I decided to dichotomize the connectives, so that a proportion of 0 was left unchanged but proportions greater than 0 were assigned the value of 1. Additionally, I created two composite scores, one for minimization and one for maximization. This measure was considered an implicit measure.

Classism

To assess participants’ baseline level of classism, each participant completed the 13-item Classist Attitudes Scale (CAS; Hoyt, Doyon, & Dietz-Uhler, 1998). The CAS measures three factors of classism: stereotypes, attributions, and prejudice. Items include beliefs about perceived social mobility (e.g., “the reason some people are poor is because they do not work hard enough”) as well as stereotypes about the poor (e.g., “poor people often physically abuse their children”). Participants responded on a 1–7 Likert-type scale (1 = strongly disagree, 7 = strongly agree), thus scores could range from 25 to 175. The items comprising the classism scale were factor analyzed, and this revealed the presence of only one factor (explaining 33% of the variance) on which all the items of the scale loaded satisfactorily. They were thus averaged into a composite score (α = .87).

Manipulation check and demographic form

Participants’ background information was used to examine possible moderating effects. The demographic form included questions of ethnicity, gender, income, age, sexual orientation, and political affiliation. A manipulation check asked participants to report the target’s ethnicity, age (brackets were 18–23,24–29, 30–35, 36–41, 42–47, 48–53, 54–59, and above 59; recoded 1–8), income (brackets were under US$20,000; US$20,001–35,000; US$35,001–50,000; US$50,001–65,000; US$65,001–80,000; US$80,001–95,000; US$95,001–110,000; US$110,001–125,000; US$125,001–140,000; US$140,001–155,000; and more than US$155,000; recoded 1–11), and class bracket (social brackets were upper class, upper-middle class, middle class, lower-middle class, and working class).

Procedure

Male and female participants of varying SES completed two ostensibly unrelated tasks. In the first task, participants read a mock police file and then answered questions relating to the perceived innocence and guilt of the survivor (CRQ; Schult & Schneider, 1991). In addition to these measures, participants completed an open-ended question immediately after reading the scenario. The question asked them to write everything that they remembered about the case and served as a reminder of the manipulations as well as a more implicit and qualitative measure of participants’ reactions to the case. To examine the effect of perceived sexuality, participants rated the target’s probability of being promiscuous, followed by the ARVS (Ward, 1988). In the second task, participants completed an “attitude survey” which measured classist attitudes. The CAS (Hoyt et al., 1998) was embedded in a longer survey with filler questions. Participants’ levels of classism were examined for moderating effects on perceptions of the target woman. Finally, participants completed a manipulation check that asked them to report the survivor’s ethnicity and SES, and a demographic form in which they reported their own gender, ethnicity, class background, sexual orientation, and political affiliation.

Results

Manipulation Check

Because of the only partial overlap of the two main manipulation checks (perceived income and class of the survivor), and in order not to eliminate too many participants, the study relied exclusively on the perception of the social class of the survivor as a manipulation check, that is, the most important indicator of how participants perceived the survivor. Confirming the validity of the manipulation, the majority of participants in the high-SES condition perceived the survivor as belonging to the middle class or higher, and the majority of participants in the low-SES condition perceived the survivor as belonging to the lower-middle class or lower. Thirty participants were eliminated from further analyses because their perception of the survivor did not fall within the above range for the experimental condition in which they found themselves, the majority of whom (n = 25) were in the low-SES condition. This is admittedly a relatively high proportion, but given that I had not pretested the manipulation and that the final number of participants was still large (N = 101), I deemed it an acceptable procedure.

Before computing the analyses relative to the hypotheses, a χ² was computed that confirmed there were no significant differences in participants’ perceptions of the race of the survivor depending on the condition. Among participants who did indicate an ethnicity, χ²(4) = 7.43, ns. Most participants in both conditions reported that they thought the target was White (58%) or that they did not think of any particular race when reading about her (34%).

There was a significant difference in perceived age, F(1, 98) = 23.25, p < .001, with the high-SES target being perceived as older than the low-SES target (Ms = 2.92 and 2.14, SD = .84 and .77). Given this difference in the perception of the survivor’s age, and given that other variables such as participants’ gender and social class could have influenced some of the dependent variables of interest, all of these variables were used as covariates in a first set of analyses similar to those presented below. As the inclusion of these covariates did not alter the significance of the results, I report below analyses without covariates.

Victim blame, attitudes toward rape victims, and perceived promiscuity

Because victim blame, attitudes, and perceived promiscuity were moderately correlated (see Table 1), a multivariate analysis of variance was conducted to examine the effect of condition on the three attitude variables. This analysis revealed a significant effect of condition, Wilks’s λ = .87, F(3, 75) = 3.50, p = .01. Specifically, in regard to victim blame, participants saw the target as more blameworthy when she was of low-SES (M = .29) rather than high-SES (M = .21), F(1, 77) = 2.46, p < .05. Participants also held more derogatory attitudes toward rape survivors on the ARVS after reading about a low-SES (M = 2.79) survivor rather than a high-SES survivor (M = 2.14), F(1, 77) = 8.94, p < .01. Important to the theory, there was a significant effect of condition, such that low-SES (M = 4.22) women were perceived as being significantly more promiscuous than high-SES (M = 3.25) women, F(1, 77) = 3.96, p < .05 (see Figure 1).

OPEN IN VIEWER

Table 1.

Correlations Between Explicit Measures and Perceived Promiscuity.

Dependent Variables	t_Blame	ARVS	Promiscuity
t_Blame	1
ARVS	0.57	1
Promiscuity	0.43	0.34	1

Note. All correlations are significant at p < .05 or less. ARVS = Attitudes Toward Rape Victims Scale.

OPEN IN VIEWER

Figure 1.

Means for promiscuity stereotyping.

Linguistic analysis

Because positive connectives suggest maximization of the significance of the events and negative connectives suggest minimization of the same, I analyzed the data via a moderated regression with strategy (minimization vs. maximization) as a within-participant factor and target and classism as between-participant factors. The interaction between strategy, classism, and condition was significant, F(1, 97) = 6.23, p < .01. The effect of target was then computed for low and high levels of classism, revealing that participants who scored low in classism made use of greater maximization in the low (M = .96) compared to the high-SES target condition (M = .81), t(97) = 1.92, p < .05, but displayed no difference in the use of minimization strategies (Ms = .10 and .12). That is, participants who were low in classism were more likely to maximize the low-SES target’s experience. High-classism individuals, on the contrary, made use of greater minimization in the low (M = .32), compared to the high-SES (M = .10) target condition, t(97) = 3.34, p < .005, but displayed no difference in the use of maximization strategies (Ms = .76 and .72); namely, participants who were high in classism were more likely to minimize the low-SES target’s experience (for the overall pattern of means, see Figures 2 and 3).

OPEN IN VIEWER

Figure 2.

Maximization strategies as a function of condition and level of classism. Participants who scored low in classism made use of greater maximization in the low compared to the high-SES target condition. High classism participants did not differ in their use of maximization strategies.

OPEN IN VIEWER

Figure 3.

Minimization strategies as a function of condition and level of classism. Participants who scored high in classism showed more minimization in the low compared to the high-SES target condition. Low classism participants did not differ in their use of minimization.

Discussion

The results yielded considerable support to the hypotheses. Overall, participants did assign more blame and show more negative attitudes toward low-SES rape survivors. Participants also rated the low-SES survivor as more promiscuous, and importantly, this stereotyping correlated with the blame and negative attitudes measures. Finally, participants’ level of classism moderated linguistic minimization and maximization. Interestingly, although I predicted that the main difference would occur in minimizations strategies, in fact, participants who scored low in classism were more likely to maximize the events of the low-SES condition, whereas participants who scored high in classism were more likely to minimize the events in the low-SES condition. This suggests that low and high classism work differently in shaping how we speak about such events. For example, low classism participants, when reflecting about the low-SES target, would write maximizing sentiments such as, “ … [a]nd since she trusted him, C.S. invited him inside her home. A little while later, the man showed his true, forceful personality and physically and sexually abused her.” In comparison, high classism participants, when reflecting about the low-SES target, were more likely to write minimizing opinions such as, “But why did she invite that person in the first place?” Although this is only a preliminary finding, it is important as it suggests that the very way in which individuals structure the events in their minds and turn them into a narrative is influenced by their level of classism and the SES of the survivor. Our results support Cozzarelli et al.’s (2002) finding that one of the main (and perhaps most influential) stereotypes about low-SES women is that they are promiscuous. Furthermore, our results support previous findings that sexual stereotyping leads to justifications of sexual assault (George & Martinez, 2002; Greene, 2000).

Taken together, these research findings can contribute to understanding of the plight of low-SES sexual assault survivors and serve as a starting point for examining the barriers they may face in the legal system. It is, however, important to expand upon this work in several directions. First, to see how sexual assaults that occur between different socioeconomic targets (e.g., high-SES perpetrator, low-SES survivor and vice versa) would be perceived both in cases of survivor and perpetrator culpability and perpetrator sentencing recommendations. Although the literature shows that low-SES perpetrators often receive harsher sentences than high-SES perpetrators (Mazzella & Feingold, 1994; Osborne & Rappaport, 1985), I do not yet know whether differences in the assailant’s SES influence perceptions of low and high-SES survivors. Although Yamawaki et al. (2007) found that low-status survivors of high-status attackers were more likely to be blamed for the rape, the manipulation of status did not actually reflect SES differences between the survivor and attacker. It would also be important to examine real-life conviction rates for perpetrators who attack low- rather than high-SES women. Second, these results may help us think about the jury selection process in cases of sexual assault. Interestingly, women tend to be more sympathetic to survivors of rape than men (Struckman-Johnson & Struckman-Johnson, 1992), but our predominately female sample was fairly negative toward the low-SES survivor. This suggests that there may be a distinction in regard to whom woman are sympathetic. Our results suggest that women may be less sympathetic toward other women when they perceive the survivor to be of low-SES and therefore more blameworthy. The predominately White participants in this study may have been engaging in an in-group distancing effect (Kunstman, Plant, & Deska, 2016), a tactic used specifically against low-income people, whereby the White female participants would distance themselves from White low-SES women in order to maintain their own sense of status. Overall, White people do dehumanize people who they consider “White trash” (Loughnan, Haslam, Sutton, & Spencer, 2013), and as a result, are less likely to be as sympathetic toward a survivor as they typically would be in cases of sexual assault. As such, one cannot assume that a female juror will be less lenient and more understanding in cases of sexual assault that involve low-SES survivors. In relation to this, we also need to better understand how a low-SES woman may be viewed and, thus, represented by her attorney. Although attorneys may be able to move beyond the explicit stereotyping, implicit stereotyping vis-à-vis language may shape perceptions of the assault.

Third, it is important to examine whether classist stereotyping may impact the behavior of low-SES women after cases of sexual assault. It is possible that low-SES women know that because of their SES they will be viewed negatively, thus preventing them from reporting incidences of sexual violence. This is admittedly a difficult research question to address, as by definition it is difficult to estimate and of course interview women who did not report a rape. Nevertheless, a combination of in-depth interview with a few women and analyses of some national statistics could help elucidate this point.

Fourth, and most importantly, if low-SES women are internalizing classist stereotypes, or avoiding reporting sexual assault because they know that others may be classist, it would be essential to examine intervention and therapeutic strategies for low-SES survivors and education strategies for emergency workers (e.g., nurses, police, doctors, etc.) who may encounter, and in many cases be first responders to, survivors of assault. Of course, the best methods for combating these effects would be to reduce or remove the sexual stereotypes associated with low-SES women. The easiest and fastest route would be through the media. Just as many people have learned stereotypes through television and film, they can then, too, easily learn more realistic information about low-SES people through positive representations of their lives. Films and television shows that portray low-SES people with respect and dignity, that show low-SES women as being well-rounded humans rather than caricatures, and that provide a multifaceted, nuanced portrait of low-SES people’s lived experiences are necessary and perhaps the last real unexplored territory in the media landscape.