The 2010 NISVS study and sexual victimization: redux

[mellowness]

Previously, I have written about the 2010 NISVS study and its citation by MRAs, "egalitarians", and other anti-feminist folks to show that the incidence of rape is not gendered. In particular, they cite the study to show that, contrary to what those nasty man-hating feminists say, 40% of all rapists are female, and so rape is not primarily a male-on-female crime. In light of some information I accidentally left out in my earlier analysis of the study's figures, I want to reexamine the figures in this entry.

Typhonblue from GendErratic, an anti-feminist blog that I am only going to link to indirectly because I despise Typhonblue and her colleagues, wrote a post about the NISVS study in yet another effort to demonize feminists. The only interesting part of the article that seems to strongly support Typhonblue's claim is this quote from a 1997 study:

16% of men with documented cases of sexual abuse considered their early childhood experiences sexual abuse, compared with 64% of women with documented cases of sexual abuse. These gender differences may reflect inadequate measurement techniques or an unwillingness on the part of men to disclose this information (Widom and Morris 1997).

This is cited as an explanation for why there is a gender difference for lifetime rates and not for annual rates: male victims are simply less likely to report child sexual abuse in a self-report survey. However, this citation is silly for several reasons. First, the study includes only people who were sexually abused in their childhood about 20 years ago.

Second, the 16% figure applies to all kinds of sexual abuse, not just being made to penetrate. At most this study can be cited to show that men are less likely to disclose a history of being sexually abused in their childhood; it does not necessarily suggest that men with documented cases of being raped by women via envelopment are less likely to report being raped in a self-report survey. In fact, because the Widom and Morris study excludes victims who were victimized in their adolescence or adulthood, it would be ridiculous to use it alongside the NISVS report because the annual figures that Typhonblue was looking at actually exclude incidents of child sexual abuse for many victims. As the NISVS report itself says:

The National Intimate Partner and Sexual Violence Survey is a national random digit dial (RDD) telephone survey of the non-institutionalized English and/or Spanish-speaking U.S. population aged 18 or older.

In addition, here is a snapshot of the page that lists the age distribution of the survey sample:



This is insightful because 18-year-olds are borderline cases here for the annual figure - if they reported sexual victimization that happened within a 12-month period before they were 18, then legally such victimization could be considered child abuse since 17-year-olds are deemed minors under US law. But since the majority of respondents were not 18, as you can see in the above image, the number of victims of child abuse that occurred within the last 12-months is narrowed to a great degree, thereby limiting the application of the Widom and Morris study even further.

Another thing I should point out in the above excerpt: even the authors of the study say that the gender difference found may be a result of either inaccurate measurement or the relative unwillingness of men to report sexual abuse in a survey. That means that the study isn't as conclusive as the folks citing it would like to believe.

I wasn't satisfied with simply reading Typhonblue's excerpt, so I tried to find the study itself (for free). I was only able to find an abstract and two articles that reference the study. Here is an excerpt from the abstract:

Accuracy of adult recollections of childhood sexual abuse was assessed using 4 different measures, completed in the context of a 2-hr in-person interview in young adulthood (N = 1,196).

Emphasis mine. This is the methodological design of the study. And it raises an interesting question: what would the disclosure rate for men be if there wasn't a stranger in the presence of each respondent? It could very well be that the reason men were much less likely to disclose their history of child sexual abuse in the survey is that many of them felt uncomfortable providing affirmative answers in front of another person. In other words, it was perhaps the design of the study that led to the lower disclosure rate. An alternative methodology, specifically one in which survey respondents are not participating in an in-person interview devoid of respondent anonymity, could have led to a different disclosure rate for men.

I found some other papers referencing the Widom and Morris study. The one that is most relevant to this entry is this one, which provides a criticism of both the study in question and a similar study. Here's an interesting excerpt:

Goodman et al. (2003) deem Widom and Morris's (1997) prospective study as "uninterpretable" because it did not verify whether the documented case was disclosed (as opposed to some other instance of abuse).

In other words, according to this source, Widom and Morris didn't check whether the incidents of child sexual abuse disclosed by the respondents were the officially documented incidents of child sexual abuse. That means that the disclosure figures were misleading and unreliable from the start. For all of Typhonblue's talk about feminist intellectual dishonesty, she sure seems to be dishonest herself. Who could so easily overlook these glaring problems with the citation provided?

Now I want to mention a recent article posted on Man Boobz. David Futrelle emailed an NISVS representative about the 40% figure being thrown around by anti-feminists, and they responded by saying that the figure is based on flawed math and severe misinterpretations of the data found by the study. Much of the email is interesting, cogent, and definitive, but this quote really stuck out to me:

While the percentage of female rape victims and the percentage of male being-made-to-penetrate victims were inferred from the past 12-month estimates by combining two forms of violence, the percentage of perpetrator by sex was taken from reported estimates for males for lifetime (a misuse of the percentage of male victims who reported only female perpetrators in their lifetime being made to penetrate victimization). This mismatch of timeframes is incorrect because the past 12-month victimization cannot be stretched to equate with lifetime victimization.

That's a pretty big thing to overlook if you ask me. Fortunately, in the same GendErratic article, Typhonblue (kind of) addresses this point:

If we look at the more reliable statistic, the risk of rape in the last twelve months, and we fix the NIPSVS’s mistake in classifying forced envelopment as “other sexual assault” and not rape, we find that 80% of men report a female rapist and 98% of women report a male rapist. (This estimate is based on the sex of reported perpetrators for sexual assault over a lifetime. There is no reason to think the number of female perpetrators for ‘forced envelopment’ would decline between the lifetime and last year reports: if anything they would increase)

She says that there is no reason to assume any statistical difference in the sex of perpetrators across the 12-month sample and the lifetime sample. But data gathered from lifetime estimates can't really be used for a sample of people victimized within the last 12 months since the samples aren't comparable or representative of one another. And, as I have shown earlier, the 12-month figures are the least inclusive - they exclude a very large number of victimizations that did not happen within a past 12-month period. Moreover, the implied assertion that the 12-month figures would reveal more female perpetrators is baseless and nonsensical. It's as if she said that solely to impress her readers. Can you conceive of any reason the 12-month sample would include more men victimized by female perpetrators than the lifetime sample? I can't.

And so that concludes my ranting about the dishonest and unintelligent analysis of the NISVS study. I wonder how much could be done for male victims of sexual violence if anti-feminists decided to use this study to bring to light the issue of rape against men instead of using it as leverage for their misogynistic, anti-male, unapologetically-privileged discourse.

EDIT: I should also note that, even if the Widom and Morris citation was valid evidence for the claim that the lifetime figures are less reliable, it still wouldn't help support the 40% female perpetrator figure. That's because using the citation as evidence would imply that, because the lifetime figures for incidence are unreliable, the information (such as the sex of the perpetrator) gathered from the lifetime figures would face similar reliability issues and so cannot be applied to the figures considered to be more reliable: the 12-month figures. After all, why use information from a less reliable sample to make a conclusion about a more reliable sample?

Comments

Hmm

[mendokuse]

One thing about this study is that it was done by cold-calling people. I personally do not accept survey calls, I hang up on them. I am sure a great many other people are the same way. So the study right there is skewed towards people who are willing to answer phone surveys about intimate details of their lives which a great many people simply are not prepared to do.

But this is part of the issue with "science" all together. It's not objective at all, and the "data" can be twisted any way someone wants it to be.

Re: Hmm

[mellowness]

Yeah, RDD surveys have their flaws, although the researchers did try to make up for the noncompliance bias by offering incentives to the people who were called. In general, victimization surveys are difficult to conduct.

Regarding science, I think it has its uses on a basic level, but people have exaggerated them greatly. I really hate the supremacist attitudes towards science. Even hearing someone say "pseudoscience" makes me cringe.