Paul Elam's big mistake on domestic violence: A case study in MRA self-deception

I didn't think I was going to reply to Paul Elam's latest post in our abortive "debate" on domestic violence  -- see here for the details on why it fell apart, and here for details on his childish and unethical behavior since and here for the rest of my debate posts -- but he's really outdone himself this time, with an utterly spectacular misreading of an important research report on violence against women. Indeed, I've read over the relevant portion of his post several times, because I can't quite believe he's saying what he seems to be saying. If he is, and I have no other explanation for his remarks, his post becomes something of a case study in the way in which antifeminist dogma can distort even the most basic analysis of empirical data.

In the context of my debate with Elam, it's not an insignificant error. Indeed, Elam sees his erroneous conclusion on this research as a sort of trump card in our debate, the grand finale to his final post in the debate. The only problem is that he's completely wrong.

You don't have to take my word for it. To make sure there was absolutely no doubt that Elam was misinterpreting the report, I contacted one of the report's authors. She indeed confirmed that Elam's interpretation was flat out wrong. I'll get to that in a minute.

Let's get into the details, shall we?

The report in question is one I cited in my initial post, titled Prevalence, Incidence, and Consequences of Violence Against Women: Findings From the National Violence Against Women Survey. (EDIT: You can find a pdf of it here.) The paper, co-written by Patricia Tjaden and Nancy Thoennes, summarizes the findings of a massive survey on violence jointly undertaken by the National Institute of Justice and Centers for Disease Control and Prevention, which, despite the title, also dealt with violence against men. The researchers surveyed 16,000 people, divided equally between men and women, about the violence they had experienced over their lifetimes -- specifically, whether or not they had been raped, physically assaulted, or stalked.

Elam's ideologically driven misreading of the report starts with a misreading of the opening paragraph of the report, a brief historical summary of how the rise of feminism led researchers to start to seriously pay attention to violence against women:

Violence against women first came to be viewed as a serious social problem in the early 1970s, in part because of the reemergence of the Women’s Movement. In unprecedented numbers, scholars trained in such diverse disciplines as philosophy, literature, law, and sociology began to examine violence against women in the context of a feminist ideology.

All of this is a pretty straightforward accounting of what actually happened. But, the researchers continue:

Despite the resulting outpouring of research on violence against women, particularly in the areas of rape and intimate partner violence, many gaps remain in our understanding of violence against women. Until now, empirical data on the relationship between certain types of violence against women, such as childhood victimization and subsequent adult victimization, have been limited. Reliable information on minority women’s experiences with violence and on the consequences of violence against women, including rates of injury and use of medical services, is also limited.

So far, the meaning of these remarks is crystal clear: Though feminism inspired a great outpouring of research on violence against women, there was still insufficient reliable empirical data to measure the true extent of the problem.

The researchers then go on to present the details of the National Violence Against Women Survey, a study designed to provide precisely what they said was lacking: reliable empirical data on the various forms of violence against women.  (In order to provide more context for this data, and to provide a basis for comparison, the study also asked the same questions to an equal number of men.)

Elam, though, reads this relatively straightforward introduction to the report as a sinister statement of purpose. Highlighting the phrases "Women's Movement" and "in the context of a feminist ideology," he declares:

Yes, in this the very first paragraph of the study, they identify not as academicians, but feminist ideologues. With a profound lack of erudition that can only be rooted in hubristic hegemony, they inform readers from the beginning that this is a political action. Straight from jump.

Not a promising start for Elam. But we haven't gotten to Elam's biggest error. 

Elam's Great Misunderstanding starts off innocently enough: he cites data from the report on rape and physical assault that shows that, with the exception of the category of rape, men report suffering more violence than women. This is a fairly unsurprising result; numerous studies have found the same thing.

Note that this data measures violence overall, NOT intimate partner violence by itself. Most of the violence against men is in fact perpetrated by other men.

Elam then shows a chart from the study that looks at the incidence of intimate partner violence, broken down into various categories of violence; it shows women more than three times as likely to report being victimized by IPV than men.

It's what Elam does next that truly boggles the mind. After noting that the data did indeed seem to suggest that women are the primary victims of IPV, he firmly declares this conclusion "wrong." No, he says, what the dastardly feminist researchers did was to "factor weigh for under reporting [but] to their disgrace they did not figure it in to the graphs."

As proof for this, Elam quotes a relatively straightforward passage in the text that discusses some of these results, and specifically refers back to the chart in question:

It is important to note that differences between women’s and men’s rates of physical assault by an intimate partner become greater as the seriousness of the assault increases. For example, women were two to three times more likely than men to report that an intimate partner threw something that could hurt or pushed, grabbed, or shoved them. However, they were 7 to 14 times more likely to report that an intimate partner beat them up, choked or tried to drown them, threatened them with a gun, or actually used a gun on them (see exhibit 8).

After quoting this text, Elam triumphantly declares victory:

And so there you have it.  A rough sketch of the math will lead you to a very familiar situation.

Domestic Violence- Women are half the problem.

Huh? The first time I read this I was simply baffled. Elam posts a chart showing that women report being the victim of IPV more often than men do, then a paragraph discussing the very same results, which says exactly the same thing, and which specifically refers back to that very same chart, and somehow concludes that ... women are responsible for half the problem?

It took several rereadings for me to even grasp how he might have come to that utterly erroneous conclusion. Apparently, as best as I can figure it, he has interpreted the word "report" in the text to mean "overreport" instead of, you know, "report." (Or that it indicated in some way that women overreported in comparison to men, who underreported, or something along these lines.) So that, as Elam figures it, the numbers in the text basically cancel out the numbers in the chart. In fact, the numbers in the text reflect the exact same data as the numbers in the chart.

Thus Elam transforms, in his mind at least, an empirical report of survey results that challenge his central claim -- that women are half the problem in domestic violence -- into one that proves his pet theory, and which reveals the perfidity of devious, cunning feminists.

Just so there would be absolutely no question that Elam is completely mistaken in his conclusion, I got in touch with Patricia Tjaden, one of the key researchers behind the survey, and the co-author of the summary Elam quoted from. She told me that, indeed, his interpretation of the figures in the paper is flat out wrong. As she put it in an email:

Yes, you are right in your interpretation of our results: Generally
speaking, in our study "reported" means respondents disclosed that they had
ever been a victim of a specific type of violent victimization. So, for
example, as presented in Exhibit 3 in our report on intimate partner
violence ... 8.5 % of women compared to 0.6% of men
disclosed that they had been beaten up by an intimate partner at some time
in their lifetime.  It should be noted that some were beaten up more than
once, but these estimates reflect only if they "ever had."  Thus, (surveyed)
women were 14 times more likely than (surveyed) men to report ever being
beaten up by an intimate partner [8.5/0.6 = 14.17.]

I have no idea what your [debate] opponent means when he said our
estimates reflect over-reporting.  Perhaps he meant that women are more
likely than men to report victimization to an interviewer?  There is little
research on what influences women and men to disclose victimization during
telephone surveys.  We conducted a small study during the course of the
NVAWS to see if interviewer gender impacted male respondents' responses to
survey questions.  (We didn't do it for women because all the women were
interviewed by female respondents.)  We found that male respondents were
more likely to disclose sensitive information, such as age, income, fear and
accommodation behavior, and recent victimization, to male interviewers.
This contradicts findings from previous research that shows respondents -
male and female alike - feel more comfortable disclosing sensitive
information to female interviews in face-to-face surveys.

The paper she is citing here is this one, available online here (pdf format):

Tjaden, P & Thoennes, N. (2000).  Extent, nature, and consequences of
intimate partner violence:  Findings from the National Violence Against
Women Survey.  Washington, DC: US Department of Justice NCJ 181867.

The only real question is whether Elam has distorted the results of the NVAWS deliberately. I don't actually think so. He is enough of an ideologue to believe that a report based on a massive government study and which has been exposed to an enormous amount of scrutiny over the years in fact secretly proves his pet theory.

One final note: Elam also makes a big deal of the fact that the NVAW used the Conflict Tactics Scale (CTS) in its surveys, a research tool which I have criticized in my previous posts in the debate. As is often the case with Elam, this is a half-truth. The survey, as Tjadan noted in her email to me, "used questions similar to those in the CTS, but framed them differently," and thus got very different results.

I will end with another comment from Tjaden, which helps to put this debate in a broader context:

I know this debate over whether men and women are equally likely to
perpetrate violence against their intimate partners is very confusing and I
have spent a good part of my career attempting to convince fellow
researchers and the federal government that we need to spend time and money
figuring out why different studies (i.e. different methodological approaches)
have yielded such disparate findings.  This would be far more fruitful than
pointing fingers at each other and calling each other names.

This is a topic I will take up further in future posts.