Lynn Hecht Schafran and Jillian Weinberger of Legal Momentum (a women's legal defense and education fund) say that recent reports from the U.S. Department of Justice, Bureau of Justice Statistics (BJS) underestimate the number of rapes among persons with disabilities and women.
Schafran and Weinberger point specifically to two reports, Crime Against Persons with Disabilities (2007) and Female Victims of Violence (2008), arguing that the methodology for both were flawed.
Crime Against Persons with Disabilities, for example, excluded institutionalized people with disabilities - a huge omissions considering that sexual assault and abuse happen at extremely high rates in institutional settings. Schafran and Weinberger also note that the statistic in the report related to reporting abuse to the police is only "based on 10 or fewer sample cases."
Female Victims of Violence - which showed that rape rates have decreased significantly recently, has similar methodological problems.
The report states that between 1993 and 2008 "the rate of rape or sexual assault against females declined by 70%" (p. 6). This conclusion is "based on the NCVS [National Crime Victimization Survey] between 1993 and 2008" (p. 6). Yet a 2007 study by some of the most highly regarded researchers in this field sharply disagrees. Drug-facilitated, Incapacitated and Forcible Rape: A National Study was conducted by Dr. Dean Kilpatrick, Director of the National Crime Victims Research and Treatment Center, along with colleagues from the Medical University of South Carolina....The new BJS report estimates that there were 182,000 rapes/sexual assaults against women age 12 and older in 2008 and that 47% of these rapes were reported to police (p. 5). In contrast, the Kilpatrick et al. study concludes that: "During the past year alone [2006], over 1 million women in the U.S. have been raped... Our estimates do not appear to support the widely held belief that rape has significantly declined in recent decades...One of the more striking findings of this study was that only 16% of all rapes were reported to law enforcement" (p. 2).
Quite a difference in numbers! BJS says 182,000 rapes were committed in 2008 while the Kilpatrick report says it was more than 1,000,000.
Schafran and Weinberger says the answer is in the the NCVS methodology.
For example, the NCVS asks directly whether the respondent has been subjected to "[a]ny rape, attempted or other type of sexual attack" rather than asking behaviorally-based questions that do not label the victim's experience. The National Women's Study, in contrast, asks behaviorally-based questions like, "Has anyone ever made you have anal sex by using force or threat of harm? Just so there is no mistake, by anal sex we mean that a man or boy put his penis in your anus." It is essential to ask behaviorally-based questions because victims often do not put the label "rape" or "sexual assault" on their experience, especially when the perpetrator is someone they know, as is the case in the significant majority of rapes. Women raped by their husbands may not even know that there is such a thing as marital rape and that it is against the law.
As Schafran and Weinberger point out, this flawed methodology is big news - because studies published by the BLS get a lot of media coverage and are considered authoritative sources. So please get the word out - and check out this list of questions Legal Momentum put together that one should ask when reading studies about rape.
0 TrackBacks
Listed below are links to blogs that reference this entry: Gov't reports underestimate rape rates of women and people with disabilities .
TrackBack URL for this entry: http://www.feministing.com/cgi-bin/movabletype/mt-tb.fcgi/16966
Thanks for the link to this analysis. I was aware of the problems in the way questions were asked in the crime victimization survey and how that leads to a much lower number than when people are asked descriptively, but I wasn't aware that those in institutions were not surveyed.
From attending sexual violence prevention meetings in Minnesota and listening to the experts I believe the national crime victimization survey questions won't change since any change in the question eliminates the usefulness of trend data. What is needed is for the 2007 study done by Dr. Dean Kilpatrick to be repeated regularly and to have these results included or linked to in any summary documents.
Something not mentioned but which is related are the deep flaws in the methodology for measuring the number of reported rapes. Earlier this year it was disclosed that New Orleans classified over half of reported rapes as non-criminal complaints.
From a statistical point of view, behaviourally-based questions aren't exactly better than plain old "have you been raped" surveys, though - they measure different things. With behaviourally-based surveys, the survey designers rather than the women surveyed get to decide what counts as "rape", and that gives them a huge amount of latitude - not generally a good thing.
(For example, one widely-reported survey classed any women who had sex despite not being sure whether she wanted to as rape. Whether you think this should count or not, the way it was reported was misleading. In theory, such surveys could be fudged in the opposite direction too, but I'm not aware of any such case - probably because this form of survey is mainly used by feminists.)
I disagree heartily. Operationalizing the concept is the researcher's prerogative. This simply means the definition of the concepts the researcher is trying to measure through survey questions is for him or her to determine. The risk of measurement error increases more with survey questions such as "have you been raped" than with behaviorally-based questions, precisely because the respondent interprets what rape means. Therefore, what the researcher may be taking rape to mean, may not be exactly what the respondent is thinking of when the word rape is mentioned, resulting in measurement error.
Indeed, even the armed forces use behaviorally-based questions to assess the prevalence of sexual assault amongst their cadets, precisely because voluminous research has documented that question wording effects determine the accuracy with which sexual assault rate are measured. Your are correct in your assertion that much depends on the researchers' definition. Another study, that the National Institute of Justice funded compared behaviorally-based questions in the National College Women Sexual Victimization Survey (NCWSVS) and the National Violence Against College Women Survey (NVACW). Both surveys employed the exact same methodology with respect to sampling and data collection; and both surveys were in the field during approximately the same time period for the same duration. The NVACW uses questions very similar to those in the National Crime Victimization Survey (NCVS) as screening questions about sexual assault. The NCVS questions are very nonspecific about what rape is, referring to it simply as "unwanted sexual activity." In contrast, the screening questions in the NCWSVS differentiated between unwanted penetration, touching, fondling, and the threats thereof in their screening questions. Not surprisingly, the NCWSVS found much higher prevalence rates of sexual violence, even though both used behaviorally-based questions. That is to say, that even the quality of the definition matters to the way respondents answer questions.
A final couple of points. A study of the prevalence of child sexual abuse history among college students directly tested the effect of question wording using labels for the incidents vs. using behavior definitions. The researchers found that respondents replied affirmatively to behaviorally-based questions significantly more often than to label questions. This has to do with the psychology of survey research. Survey respondents are cued with question wording, which is why good survey researchers know that little should be left to the respondent's imagination. The more clearly a concept is defined in the question, the more likely the researcher is to get an answer that captures what they actually seek to capture, and not merely the respondent's own perception of what the researcher means.
I disagree heartily. Operationalizing the concept is the researcher's prerogative. This simply means the definition of the concepts the researcher is trying to measure through survey questions is for him or her to determine. The risk of measurement error increases more with survey questions such as "have you been raped" than with behaviorally-based questions, precisely because the respondent interprets what rape means. Therefore, what the researcher may be taking rape to mean, may not be exactly what the respondent is thinking of when the word rape is mentioned, resulting in measurement error.
Indeed, even the armed forces use behaviorally-based questions to assess the prevalence of sexual assault amongst their cadets, precisely because voluminous research has documented that question wording effects determine the accuracy with which sexual assault rate are measured. Your are correct in your assertion that much depends on the researchers' definition. Another study, that the National Institute of Justice funded compared behaviorally-based questions in the National College Women Sexual Victimization Survey (NCWSVS) and the National Violence Against College Women Survey (NVACW). Both surveys employed the exact same methodology with respect to sampling and data collection; and both surveys were in the field during approximately the same time period for the same duration. The NVACW uses questions very similar to those in the National Crime Victimization Survey (NCVS) as screening questions about sexual assault. The NCVS questions are very nonspecific about what rape is, referring to it simply as "unwanted sexual activity." In contrast, the screening questions in the NCWSVS differentiated between unwanted penetration, touching, fondling, and the threats thereof in their screening questions. Not surprisingly, the NCWSVS found much higher prevalence rates of sexual violence, even though both used behaviorally-based questions. That is to say, that even the quality of the definition matters to the way respondents answer questions.
A final couple of points. A study of the prevalence of child sexual abuse history among college students directly tested the effect of question wording using labels for the incidents vs. using behavior definitions. The researchers found that respondents replied affirmatively to behaviorally-based questions significantly more often than to label questions. This has to do with the psychology of survey research. Survey respondents are cued with question wording, which is why good survey researchers know that little should be left to the respondent's imagination. The more clearly a concept is defined in the question, the more likely the researcher is to get an answer that captures what they actually seek to capture, and not merely the respondent's own perception of what the researcher means.
At the same time the NCVS and UCR are not simply used as an assessment of total crimes but also as a time series. A major problem with time series is that you need to maintain a constant methodology or at least have a reasonable lag period of collecting according to a parallel methodology when switching over.
I'll agree that neither are particularly good instruments (the NCVS e.g. only records incidents of opposite gender sexual assault, the UCR only records incidents of male on female sexual assault) but as far as a time series is concerned it should be consistent for the purposes of informing whether crimes have gone up or down and the evidence as presented does not suggest that its estimates in this regard are incorrect.
I find it hard to believe the usual statistic that only one in six women are subjected to sexual assault. I don't know a single female acquaintance or friend who has *not* been molested in some way. They may not call it "rape" but it is, according to all known legal definitions. It's unwanted sexual touching. And many of these friends and acquaintances have been molested several times. This doesn't even begin to cover things like street harassment or sexual harassment at work.
Unwanted sexual touching is both common and bad, but virtually no "legal" definition classifies it as "rape" unless there is penetration.
The company I work for has many, many assisted facilities for people with mental retardation/developmental disabilities. Of the people who receive our services, roughly 80% (men and women) have been sexually abused at some point in their lives.
The definition of a disability in the report is very, very broad. I am inclined to believe that people with lower cognitive functioning are more susceptible to abuse, but I don't really know.
From what I recall, Deaf people also have higher rates of abuse than the rest of the population. The issue seems to be the environment they're forced into:
1. There's not a lot of accessible sexuality education, and people often assume it's not needed for people with disabilities.
2. In institutions like residential schools for the Deaf, many adults (and older students) have access to people when they are vulnerable--asleep, in the shower, or similar.
3. Hotlines and other protective resources can be harder to get to from an institution, and harder to access because social workers and police rarely know American Sign Language.
My company educates its consumers about sexuality, but only if they are in - or about to get into - a relationship.
Institutions, residential or otherwise, all carry risks when one considers the time and place an assault might occurr: at night, when fewer staff members are on duty, or in the shower, in a private or secluded place. Our consumers know to call 911 if they are ever in trouble, but it is ultimately up to the staff members to keep them safe (from dangerous employees or other consumers).
I am surprised that the TDD system does not have a 911 option, and think you are correct to say that protective resources are limited.
I'm glad that your company does do some sex education. I work in HIV prevention, and the lack of *any* sex education makes things quite difficult for us.
I expect there is a TDD 911 option. I hadn't thought of that.
There can be many issues with access, though: written English is based on spoken English, so reading and writing can be very difficult for Deaf people. And when someone from the 911 system responds to the emergency, they are not likely to know Sign Language, so communication can be very difficult.
The company I work for has many, many assisted facilities for people with mental retardation/developmental disabilities. Of the people who receive our services, roughly 80% (men and women) have been sexually abused at some point in their lives.
The definition of a disability in the report is very, very broad. I am inclined to believe that people with lower cognitive functioning are more susceptible to abuse, but I don't really know.
I'm unable to find the evidence that the female victims of violence study is wrong about there having been a decline in the frequency of rape. The "executive summary" of the Kilpatrick study only asserts this without mentioning what data or methodology they rely on. In order for this to be the case, the BJS would have to be underestimating rape rates in 2008 much more than they underestimated rape rates in 1993; Kilpatrick's executive summary doesn't mention anything about rape rates in 1993 from which one might draw conclusions as to whether this is the case. The executive summary does say that rates of rape reporting haven't gone up, but again without mentioning the evidence, and more importantly (since absence of change perhaps doesn't require as much evidence as change) it would be especially surprising that their "no decline" hypothesis would turn out to be right if there wasn't a drop in rape reporting rather than a constant rate.
I realize this is a tiny point; the important issue is that there is a huge problem at present, and whether it was or was not slightly more huge in the past doesn't affect how huge of a problem it is now. And I suppose I don't find it that implausible that rates of reporting might have gone down after years of conservative rule which could have made women less inclined to expect the government to be sympathetic, or that there could be other factors operating such that the BJS under Clinton might have underestimated rape rates to a lesser extent than the BJS under George W. Bush. But I always prefer to get the most detailed evidence possible, so if anyone knows more about how Kilpatrick gets that result, I'm very curious.
Well, I could have told you this. Beyond the psychical disabilities and delays people are thinking add to your chance of getting raped, women and men with mental illness are abused at a high rate.
I was abused as a child in a mental hospital and even my abuser knew that no one would believe a crazy 8 year old.