Abstract
We examined the association between rape myth acceptance (RMA) and reporting rapes to the police. Situational characteristics of the rape (e.g., stranger attack, injury) are known predictors of reporting, but no existing studies have examined the association between beliefs about rape and reporting. In addition, most studies of RMA do not assess victimization history. Incarcerated women experience high rates of sexual assaults prior to incarceration. We recruited 74 rape survivors from a northwestern state prison. Results suggest that women who endorsed higher levels of RMA were less likely to report their rapes to police; however, participants endorsed few rape myths.
Roughly 1 in every 6, or 17.7 million American women, are the victims of attempted or completed rape (Tjaden & Thoennes, 1998). However, the majority of rapes go unreported (Wolitzky-Taylor et al., 2011). The National Crime Victimization Survey (NCVS) revealed that until very recently, rape and sexual assault were the least reported of all violent crimes (Rennison & Rand, 2003). Specifically, this study indicates that less than 54% of all rapes are brought to the attention of law enforcement officials. Other psychological research suggests even lower rates of reporting, stating that only between 5% and 31% of rape victims inform the authorities (Fisher, Daigle, Cullen, & Turner, 2003; McGregor, Wiebe, Marion, & Livingstone, 2000; Ullman, 1996; Wolitzky-Taylor et al., 2011).
Research has shown that situational characteristics of the rape event are correlated with the likelihood that a victim will report. Nearly three decades ago, Williams (1984) introduced the concept of “the classic rape” as the most likely circumstances in which a woman reports a rape to the police. This scenario includes abduction, the perpetrator being a stranger, severe force, and serious injury. These situational characteristics of rape have been consistently replicated as correlates of the likelihood that a victim will report to the police (Bachman, 1998; DuMont, Miller, & Myhr, 2003; McGregor et al., 2000; Menard, 2005). Williams suggested victims of this specific type of rape are more likely to report, because they see themselves as real crime victims. Notably, Fisher and colleagues (2003) found that over 42% of rape victims who did not report to the police stated their primary reason for remaining silent was that they were not sure a crime had been committed. Finally, victims who were raped by an acquaintance are more reluctant to involve the police out of fear that the legal authorities will not believe them (McGregor et al., 2000; Sudderth, 1998). This is clearly problematic, however, given that 69% to 80% of all sexual assault victims know their assailants (Rennison & Rand, 2003; Sudderth, 1998).
Defining a rape based on its resemblance to the “classic rape” scenario, as well as other beliefs that negatively impact women’s decisions to report to the police (e.g., victim blame), are examples of a larger set of cultural beliefs known as rape myths. Rape myths are specific beliefs about rape that are widespread and persistently held, despite the fact that they are largely false (Lonsway & Fitzgerald, 1994). These authors note that rape myths include beliefs such as: It is the woman’s fault, she “deserves” it based on what she was wearing, partners cannot commit rape, she did not say “no” or fight back and thus it was consensual, or the victim is lying (i.e., “crying rape”).
The extent to which one believes rape myths to be true (rape myth acceptance [RMA]) has been shown to be influenced by perpetrator type. Studies have demonstrated that people who have not been raped tend to consider stranger assault to be more of a “real” crime and more distressing than acquaintance rape (Whatley, 1996) and that they assign more responsibility (i.e., blame) to victims of acquaintance rape than to victims of stranger rape (Abrams, Viki, Masser, & Bohner, 2003; Monson, Langhinrichsen-Rohling, & Binderup, 2000). This is especially true when rape victims are women of color. Specifically, researchers found that African American and Latina rape victims are less likely than White women to receive sympathy from others, disclose experiences of sexual assault, seek help, or report this to the police (Bryant-Davis, Chung, Tillman, & Belcourt, 2009; Tillman, Bryant-Davis, Smith, & Marks, 2010). However, little research has assessed levels of RMA among rape victims. Notably, most studies that assess RMA make no mention of previous victimization history (Lonsway & Fitzgerald, 1995; Monson et al., 2000).
A study by DuMont and colleagues (2003) was one of the first to look at specific rape myths (and identify them as “rape myths”) as they relate to women’s reporting to police. They found that victims’ reporting to the police was significantly associated with the use of physical force and injury to the victim (which are, in essence, components of the “classic rape,” that is, myths). However, this study did not directly measure RMA of the rape victims themselves as the authors obtained their data from hospital incident reports and correlated characteristics of the assault with reporting behavior. At this time, there is no present research that examines the actual association between the attitudinal variable of RMA and reporting to the police.
Incarcerated Women and the Present Research
Within the prison system, women comprise a small, but rapidly increasing percentage of the total prison population. The U.S. Department of Justice reported that the rate of incarceration of women is double that of men (Sabol, Minton, & Harrison, 2007), and other research has shown that between 1990 and 2000, the number of women in prison has increased by 108% (Bloom, Owen, & Covington, 2003). Further, researchers have consistently found that women in prison experience substantially higher lifetime rates of physical and sexual abuse compared with women in the general population. Conservative estimates suggest that 43% to 50% of female inmates have experienced some form of abuse prior to incarceration (Bloom et al., 2003; Browne, Miller, & Maguin, 1999; Hills, Siegfried, & Ickowitz, 2004), although some studies show that 80% to 90% of incarcerated women have experienced at least one form of interpersonal trauma (Covington, 1998; Green, Miranda, Daroowalla, & Siddique, 2005).
Incarcerated women differ from the general population, not just in prevalence of experiencing sexual violence, but also due to their criminal involvement. Studies have shown that approximately 80% of women in prison have substance use problems and that these problems are highly related to their criminal behavior in ways such as using illegal drugs, being under the influence of drugs during the commission of the crime, or committing crimes to sustain a drug habit (Bloom et al., 2003; Bloom, Owen, & Covington, 2004; Browne et al., 1999; Green et al., 2005). However, as is true for all women regardless of incarceration status, women using substances are at greater risk for experiencing sexual violence and often use substances as a means to cope with negative affect and psychological symptoms following a sexual assault (Brady, Back, & Coffey, 2004; Chilcoat & Breslau, 1998). In this regard, women in prison demonstrate similar patterns to the general population in coping with sexual assault. Browne and colleagues (1999) highlight the strong links between interpersonal trauma and substance use and conclude that such high rates of pre-incarceration trauma among this specific population is likely to contribute to the increasing imprisonment of women.
Despite the high prevalence of sexual violence, this population is starkly understudied and little is known about variables related to the sexual assault experiences of incarcerated women. Understanding these women’s reporting behaviors and levels of RMA may not only highlight links between these specific variables, but also provide a greater opportunity to understand whether RMA is a true barrier for incarcerated women, provide a potential target for intervention given the high rates of violence and the likelihood of revictimization, as well as contributing to our general understanding of an under-researched population.
The goal of the present study was to assess RMA in incarcerated women and to examine the extent to which it contributes to women’s likelihood to report an assault to the police. We hypothesized that incarcerated women, similar to women in the general population, would be more likely to report stranger rapes to the police than they would rapes that were committed by someone known to the victim (Bachman, 1998; DuMont et al., 2003; McGregor et al., 2000). In addition, we expected that RMA will be a significant, unique predictor of reporting beyond that of the relationship to the perpetrator. Finally, we expected to find an interaction between RMA and perpetrator type, such that when a woman endorses high levels of RMA in combination with having been assaulted by a known perpetrator, this will significantly decrease her likelihood of reporting to the police.
Method
Participants
The participants in this study were 74 women incarcerated at a northwestern U.S. women’s correctional facility who participated in a larger study of female offenders’ (N = 102) trauma exposure and treatment needs. The 74 women who experienced sexual assaults comprise the sample for this study. Many of these women (76%) indicated multiple rape incidents; they identified a total of 157 sexual assaults. Of these women, 66.2% were Caucasian, 11.8% were Hispanic, 4.9% were Native American, 2.9% were African American, and 16.7% identified as multiethnic. This distribution of ethnic diversity is similar to the overall corrections population in the northwestern region where the data were collected. However, ethnic minority women were highly overrepresented in this prison sample compared with the general population in this region (6.7% Hispanic, 3.2% Native American, 0.8% African American, and 2.7% multiethnic; U.S. Census Bureau, 2010). The participants reported a mean age of 32.61 (SD = 8.96). The majority of the women were incarcerated for nonviolent crimes. Specifically, 54.05% were convicted of a drug-related crime, 35.14% were charged with property damage (e.g., burglary, forgery), 14.86% were convicted of a parole or probation violation, and 8.11% were charged with driving under the influence. Only 9.5% committed a violent crime such as manslaughter or assault. Some inmates were convicted of multiple crimes simultaneously; thus, these percentages are not orthogonal. All women in this sample were eligible for release within 2 years from the time of the interview, and 97.3% of participants had been incarcerated for 48 months or less at the time of the interview.
Measures
Demographics questionnaire
Basic demographic information was collected from every participant. Specifically, age, ethnicity, education, employment history, income, marital status, number of children under age 18, present charges, and sentence length were assessed.
Illinois Rape Myth Acceptance Scale–Short Form (IRMA-SF)
The IRMA-SF (Payne, Lonsway, & Fitzgerald, 1999) contains 20 of the items that are presented on the full version of the IRMA Scale. The IRMA-SF (α = .87) assesses general rape myth acceptance via the participant’s endorsement of items on a Likert-type scale from 1 (not at all agree) to 7 (very much agree). Payne and colleagues found that the IRMA-SF also demonstrated good construct validity, in that participants’ scores were significantly associated with known correlates (e.g., adversarial sexual beliefs, hostility toward women, acceptance of interpersonal violence). In this study, the obtained Cronbach’s α for this measure was .80.
Trauma History Questionnaire (THQ)
The THQ (Green, 1996) is a 24-item measure that is designed to screen for exposure to various types of trauma in the content areas of crime-related events, general disaster and trauma, and physical and sexual experiences. Survivors of sexual assault were identified by answering “yes” to the item which asks: “Has anyone ever made you have intercourse, oral or anal sex against your will?” Participants who endorsed this item were also asked to indicate the nature of relationship(s) to perpetrator, age(s), and frequency of assaults. An internal consistency statistic was not calculated for this scale, as one traumatic event does not necessarily correlate with, or presume the presence of, another trauma.
Semistructured interview questions regarding reporting
A series of open-ended questions and prompts was created for the larger treatment needs study to assess various aspects of the disclosure of sexual assault. Women’s responses about reporting to the police for each sexual assault experienced were coded as affirmative or negative to examine the association between reporting and RMA.
Procedures
Inmates were invited to participate in this study as part of an IRB-approved longitudinal study of incarcerated women’s trauma exposure and treatment needs. Inclusion criteria were English proficiency, age 18 or older, and an expected release date within 2 years. The study was announced on each prison unit. Women were invited to participate in interviews about “stressful life events and how they are feeling and coping now.” Roughly 300 women reside in the prison at any given time. The average sentence length is about 2 years; thus, the majority of the prison inmates were eligible to participate, and approximately one third chose to take part in the study. Inmates who volunteered met individually in private rooms with clinically trained graduate student interviewers to complete the informed consent and the questionnaire packets. The interviewers read the directions, question items, and answer options aloud, while the participants read along with their own copy. This ensured that women of all reading abilities could be included in the study.
Results
Descriptive Statistics
Due to the higher than expected number of assaults identified by the participants, the data were recoded such that each case in the data set represented separate experiences of sexual assault. As mentioned earlier, there were a total of 157 rapes experienced by 74 participants. Data were organized per assault, because many of the participants indicated that they were assaulted by people of differing relationships (e.g., strangers, known assailants) and had differentially chosen to report (or not) these events to the police.
Given that multiple observations were taken from some participants, we utilized an intraclass correlation to test whether we were violating the assumption of the independence of observations required for logistic regression analyses. An intraclass correlation is defined as the ratio of the between-cluster variance (in this case, the between-subject variance) on the perpetrator variable and the reporting variable divided by the sum of the within- and between-cluster variance on these same variables (Muthén, 2000). For the present study, the intraclass correlation for the perpetrator variable was equal to .446 and the intraclass correlation for the reporting variable was .128. The overall design effect for the perpetrator variable was then equal to 1.50, while the design effect for the reporting variable is 1.14. Both of these values are not considered to be significantly large. Thus, we were able to treat each assault within the data set as an independent observation (Muthén, 2000).
The participants noted the following types of relationships with the perpetrators: 27.4% were partners, 22.3% were family members, 16.6% were strangers, 15.3% were acquaintances, 12.1% were friends, and 6.4% were gang rapes. To sum up, over three quarters of all the perpetrators were someone who was known to the victim. Thus, perpetrator type was dichotomized in all analyses as either known (coded as 0) or unknown (coded as 1) to the participant.
Slightly over half of all rapes (50.3%) were committed when the participant was less than 18 years of age, 44.6% when the participant was above 18, 3.8% during both life periods (i.e., she experienced a chronic event, usually by a partner who assaulted her multiple times, that spanned from late adolescence through adulthood), and 1.3% of cases had missing information as to when the assault occurred.
In general, the participants endorsed relatively few rape myths (M = 30.54, SD = 12.52; where the range of possible total scores for this measure is 17-119). This low rate of endorsement resulted in significantly skewed RMA scores (skewness = 1.499, kurtosis = 2.007). Due to this non-normal distribution, a logarithmic transformation of the rape myth acceptance total score was included in all subsequent regression analyses (Tabachnick & Fidell, 2007). Finally, only 23.7% of the rapes were brought to the attention of the authorities.
Logistic Regression Analyses
We used a sequential logistic regression to test whether relationship to the perpetrator and RMA were associated with the likelihood of reporting to the police. Ethnicity and level of education were found to be significantly related to RMA, thus they were included in the analyses as covariates. Given that the number of women who identified with ethnic groups other than Caucasian was relatively small (n = 2-12 per ethnic category), ethnicity is represented in this study with two groups: women of color and Caucasian women. Specifically, women of color scored significantly higher on the logarithmic transformation of the rape myth total score (M = 1.510, SD = .186) than did Caucasian women (M = 1.425, SD = .126), t(99) = 2.74, p = .007. For these analyses, minority women were coded as 0, while Caucasian women were coded as 1. In addition, level of education (a continuous variable) and endorsed RMA levels were negatively correlated, such that women with more education were less likely to accept rape myths, r = −.389, p< .001.
Due to ethnicity and education being significantly related to RMA, the primary variable of interest in this study, these demographic variables were entered in the first block of the sequential logistic regression model as control variables; perpetrator type was then added as a predictor in the second block. As expected, when controlling for education and ethnicity, perpetrator type was a significant predictor of reporting to the police, Wald χ2 = 9.856, OR = 3.704, p = .002. This indicated that the odds that a victim reported her rape to the police was 3.7 times more likely when the perpetrator was a stranger than when the perpetrator was known to the victim. Overall, this model correctly classified 78.1% of all cases, but only correctly classified 10.8% of rapes reported to the police.
Subsequently, the logarithmic transformation of RMA was entered into the next block of the logistic regression analysis. This was also a significant predictor of reporting when controlling for education, ethnicity, and perpetrator type, Wald χ2 = 5.012, OR = .019, p = .025. The odds ratio indicates that reporting to the police is 98.1 times less likely to occur for every unit increase in the logarithmic transformation of RMA. Thus, greater endorsement of RMA is associated with a lower likelihood of reporting a rape to the police. This increased the correct classification of rapes reported to the police to 13.5%. Finally, we computed an interaction between perpetrator type and RMA to be included in the final block of this analysis and found that this was also a significant predictor of reporting, Wald χ2 = 8.542, OR = 36096.75, p = .003 (see Table 1). This suggests that stranger rapes are more likely to be reported to the police, regardless of the level of RMA. However, the probability of reporting a rape to the police decreases dramatically when the perpetrator is an acquaintance and the victim holds higher levels of RMA. For instance, as the logarithmic transformation of RMA increases from 1 to 3, the probability of reporting decreases from .35 to .04, when the victim knows her assailant (see Figure 1). With the interaction term added, the correct classification of rapes reported to the police increased to 18.9%.
Summary of Sequential Logistic Regression Analyses.
Note. Outcome: Reporting to the police (0 = no, 1 = yes). Relationship between the probability of reporting and logRMA as it varies by perpetrator type. log RMA = logarithmic transformation of RMA, perp. = perpetrator type.
p< .05. **p< .01.
Relationship between the probability of reporting and logRMA as it varies by perpetrator type.
Discussion
This study attempted to address existing deficits in the literature by examining the association between attitudes reflecting rape myth acceptance and reporting to the police among incarcerated survivors of sexual assault. First, it is important to note that the frequency of sexual assault reported by the participants in this study is high. In all, 74 of 102 women interviewed about stress and coping indicated experiencing at least one rape in their lifetime. In addition, 76.1% of these assaulted women experienced multiple assaults, such that there was a total of 157 rape events among these 74 women. This obtained prevalence of sexual assault is similar to national averages reported for incarcerated women in other studies (e.g., Bloom et al., 2003). These results provide unique additional information suggesting that many female offenders not only experience high rates of interpersonal violence, but also experience multiple assaults that increase the risk of negative psychosocial outcomes.
This study also replicated past findings that one’s relationship to the perpetrator is a significant predictor of reporting to the police (Bachman, 1998; DuMont et al., 2003; McGregor et al., 2000). Logistic regression analyses indicated that women who were raped by a stranger were approximately 3.7 times more likely to report to the police than were survivors of sexual assaults committed by someone known to the participant. Next, when analyzing the unique relationship between RMA and reporting, interpretation of the odds ratio showed that for every unit increase in the logarithmic transformation of RMA, reporting to the police becomes 98.1 times less likely. Although women were generally endorsing very low levels of rape myths, these attitudes still appeared to be relevant to the decision to report (or not) to the police. The interaction between RMA and perpetrator type suggests that beliefs about what constitutes a rape or sexual crime impacts women’s decision making in regard to reporting. In cases of high RMA and a known perpetrator, the probability of reporting to the police appears even further diminished. There are no previous studies of survivors’ RMA and the association between RMA and reporting.
The finding that attitudinal variables are associated with a rape victim’s decision about reporting to the police has implications for the development of advocacy and psychoeducation programs to enhance victims’ willingness to report to the police. Although the situational characteristics of a rape cannot be changed, it is possible to impact one’s attitudes toward rape in general. Increasing the number of rapes that are reported to the police (and subsequently the number of cases brought to court) has the potential to deter the incidence of rape more generally.
At the individual level, understanding factors influencing a decision to report to the police can also have significant effects on rape victims’ mental health. The literature on the effects of disclosing a sexual assault to others more generally (e.g., to friends or family) suggests that the act of telling others about having been raped as soon as possible after the event can result in higher self-esteem and fewer posttraumatic stress disorder (PTSD) symptoms (Filipas & Ullman, 2001; Ruggiero et al., 2004). Thus, if RMA is a potential barrier to disclosure of rape to the police (i.e., reporting), it could then also be a barrier to psychological recovery and healing.
This added barrier to recovery also has implications for incarcerated women, as a high percentage of them have been sexually assaulted and may experience psychological sequelae as a result. Gender-responsive treatment for incarcerated women calls for attending to interpersonal trauma and violence as a means to not only help women recover, but also to prevent them from reoffending post-release (Bloom et al., 2003). If this trauma-informed care included psychoeducation to decrease levels of RMA, incarcerated women may also feel more willing to perceive legal authorities as resources rather than adversaries.
Women of color comprised over one third of this sample and reported higher levels of RMA than White women. Although ethnicity was not a significant predictor of reporting to the police in the logistic regression analysis, this raises important questions about the potential role of ethnicity in relationships between RMA and other rape-related variables. Previous studies have shown that women of color are less likely to report a rape to the police or discuss it with friends or family (Bryant-Davis et al., 2009; Tillman et al., 2010). Cultural beliefs and the importance of the community may then be critical variables in the assessment of RMA as well. These potential relationships warrant further investigation to better understand ways RMA may differ across ethnic groups.
Limitations and Future Directions
There are some important limitations to this study. First, incarcerated women are a unique population for several reasons as discussed earlier. Perhaps the most relevant reason here is the fact that they experience substantially higher levels of interpersonal violence than the general population; thus, these findings may not generalize to the broader population. In addition, incarcerated women are also likely to have tenuous relationships with law enforcement independent of their sexual assault history. Because this is a population that has engaged in criminal behavior, other reasons for not reporting rape to the police (e.g., drawing unwanted attention to their illicit drug use, parole violations, money crimes) could be potential covariates in this analysis. Future studies assessing the relationship between RMA and reporting in other populations are necessary. With further regard to sampling concerns, a self-selection bias may have occurred, as this project was advertised as attending to “stressful life events.” As in all studies investigating violence against women, there may be a significant difference between women who volunteer to participate and those who do not. However, it is important to recognize that one third of the women housed in the prison chose to participate.
Second, a sample of 74 women is relatively small and likely limited the number of people who fell within the reporting group of the dependent variable; however, by using event-level data, this allowed us to have a larger sample size (N = 157) without violating any assumptions of independent observations. Third, this study only examined one aspect of situational characteristics of rape (i.e., relationship to the perpetrator). Incorporating other rape characteristics (e.g., level of injury, perpetrator’s use of a weapon) as predictors or control variables may further clarify the role of RMA in disclosure or reporting processes.
Fourth, it should be noted that factors similar to RMA could also contribute to a woman’s decision to report to the police. Examining the impact of related variables such as rape acknowledgment and participants’ views of normative sexual experiences (i.e., sexual scripts; see Kahn & Andreoli-Mathie, 2000; Kahn, Andreoli-Mathie, & Torgler, 1994) in relation to RMA may prove fruitful. Fifth, accurate endorsement of RMA may be adversely impacted by social desirability as the RMA variable was highly positively skewed, requiring transformation. Though one study found no correlation between these two constructs (Spohn, 1993), some of the IRMA-SF item content may be evocative for some participants. Therefore, future research ought to continue to investigate this finding and attend to potential social desirability factors when assessing RMA. Although RMA may be non-normally distributed, its effects on other variables related to sexual assault are still present and clearly relevant to further investigation. Finally, the language on rape myth scales (e.g., IRMA-SF) may impact women’s responses. For example, women may be more likely to endorse items if the term “rape” was replaced with a less affectively laden term. Therefore, it may prove beneficial to assess if this modification has unique effects on RMA. In addition, the scale is written in the third person, possibly suggesting beliefs about the general “woman” rather than eliciting women’s beliefs about themselves.
To summarize, this study successfully generalized prior research regarding relationship to the perpetrator as a predictor of reporting among incarcerated women, thus expanding what is known about this largely understudied population. This project also contributed to the small but growing literature on RMA by addressing potential limitations of existing studies. First, we explored the extent of RMA in survivors of sexual assaults rather than utilizing the common practice of relying on general or college population samples reading vignettes about assaults and subsequently assessing RMA. We also assessed participants’ actual endorsement of rape myths rather than extrapolating RMA from other sources. Assessing RMA in a diverse sample of sexual assault survivors provides important, additional evidence about the nature of the relationship between RMA and reporting to the police. These findings suggest that attitudes about rape in general impact a woman’s decision to report. Further assessment of these attitudes in a variety of populations may have important implications for understanding how rape victims define their experiences, rationalize their decisions to bring these crimes to the attention of the authorities, and designing interventions to change the rate of reporting.
Footnotes
Acknowledgements
We express our appreciation to the women who participated in this project as well as the prison administrators and staff. We also recognize Lyn McArthur, Shilo Smith, Galatia Cepeda, Norma Boyd, Erin McGlade, Aaron Paugh, and Michelle Stewart for their contributions to the research team as well as Samantha Chesney for her editorial service.
Authors’ Note
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the Department of the Army or the Department of Defense.
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 funded by the Faculty Research Committee, College of Arts and Sciences, and Office of Research of Idaho State University.