Final proof there is no human tragedy Brad Wilcox will not exploit in order to promote marriage

I’m not going to dignify this with a thorough debunking, but here’s a quick note to highlight the evil that walks among us in academic robes.

Brad Wilcox and Robin Fretwell Wilson wrote a piece for “Post Everything” at the Washington Post that was originally titled like this:


The post didn’t specifically say what’s in the headline, but in this case I have to give credit to the overreaching headline writer for accurately capturing the basic message of the piece. What Brad wants to do is make people think that without exactly saying it. Erin Gloria Ryan on Jezebel wrote a good alternate headline for it, too: “Violence Against Women Will End When You Sluts Get Married, Says WaPo.”

Their audience is married people who feel superior to women who aren’t married, who want to coerce women into marriage — or cast them out. The friendly side of this is paternalistic shaming, the unfriendly side is violent shaming; both are expressions of patriarchal outlook. Their conclusion:

And, most fundamentally, for the girls and women in their lives, married fathers provide direct protection by watching out for the physical welfare of their wives and daughters, and indirect protection by increasing the odds they live in safe homes and are not exposed to men likely to pose a threat. So, women: if you’re the product of a good marriage, and feel safer as a consequence, lift a glass to dear old dad this Sunday.

I can’t help reading this without hearing a voice that says, “We can do this the easy way, or the hard way.”

After the interwebs’ head exploded over the headline, Brad tweeted, “Working to match title w text,” and then a new headline appeared:


The new headline is supposed to be less offensive, I suppose, but it amounts to the same thing. And it’s based on the same correlations in the post. There is still nothing in the post to show that adding marriage to a random relationship would reduce the odds or level of intimate partner violence. So the implication is the same: shame on you.

On Twitter, Marina Adshade pointed out that marriage rates and violence rates have both been falling for several decades. Brad’s response was, “Fair enough. But the question is this: Would they have fallen even more if marriage was stronger?” That’s a question he should probably have asked before writing the piece.

Can you imagine what he would do if he had the opposite result to work with — an increase in violence during a period of decreasing marriage?

We don’t have to imagine, actually, because he and his marriage-promoting compatriots at the National Marriage Project were all over that in the 1990s. To choose one example I have handy, William Galston, who sits on Brad’s board of advisors at NMP, wrote in 1991 in the New Republic (12/2/91) that, “The American family has changed dramatically in the past generation, and it is children who have paid the price.” We needed, he said, to “relegitimate the discussion of the links between family structure and a range of social ills.” Indeed, “theft, violence, and the use of illicit drugs are far more prevalent among teenagers than they were thirty years ago.” Now, as “revolution in the American family” has reached unprecedented levels, crime has fallen for two decades. <Crickets>

As a spoof — but with real data — I illustrated Adshade’s point. Here is the relationship between marriage prevalence and intimate partner violence rates:


That curvilinear statistical relationship explains 84% of the variance in intimate partner violence rates. If you add the linear time trend, the variance explained jumps to 92% and the effects of marriage remain highly significant.



If I were like Brad on the other side of this debate, the news story would read like this:

“We had reason to believe marriage was harmful, on average,” said Prof. Cohen. “But I was surprised by the strength of the relationship, especially the fact that the effect seems to accelerate at higher levels of marriage, as if marriage feeds off itself in a violence loop.” Although further research will be needed to confirm the findings, he added, the statistical association is very strong. “The bottom line is that intimate partner violence is much less common in years when marriage is more rare.”

However, I am not seriously suggesting that the decline in marriage has caused the decline in violence (although reduced exposure of women to men in general may be one factor). In fact, if you add the curvilinear effect of time, the variance explained rises to 95% — and marriage effects disappear. But the fact that violence has dropped so much while marriage has plummeted means Brad has a steeper hill to climb to make his case. It’s not enough to say, maybe violence would have declined even more. This is not one of those random spurious correlations, these are two large social trends affecting whole swaths of the population, and the correlation directly contradicts his theory. When there is a plausible connection, or the trends at least affect the same people, the burden is on the one going beyond the existing evidence to reconcile the hypothesis with the available circumstantial evidence.

But none of this matters to Brad*, or, apparently, Robin Fretwell Wilson. Their conclusion is predetermined. There is nothing that would lead them to conclude that society would not be improved by more marriage. It’s just a case of picking a subject in the news, picking some facts, and repeating their conclusions. And I think it’s appalling.

* If you’re wondering why I seem to be picking on Brad individually, please rest assured it’s nothing personal. If there was any other sociologist who behaved as poorly as he consistently does I would pick on them, too. For endless details, follow the National Marriage Project tag.

28 thoughts on “Final proof there is no human tragedy Brad Wilcox will not exploit in order to promote marriage

  1. So, in a dysfunctional relationship, marriage has a way of cementing and formalizing a person’s emotional investment in that relationship such that if the situation becomes abusive / violent, it’s that much harder to leave.

    Sounds to me like the data shows sensible people keeping their defenses solidly in place until they know they can trust a person with the (hopefully) lifelong commitment that is marriage. This would explain both the decrease in marriage rate, the decrease in violence rate, and the fact that in an era where divorce is possible, most of the remaining marriages are a self-selected population of healthy relationships.


  2. Fretwell Lewis isn’t a sociologist – she is a law professor, who along with other religous conservative law professors, has lobbied for anti-gay exemptions in existing and proposed state non-discrimination/marriage equality laws. She was so dishonest in her testimony before the DC city council that David Catania sent a blistering letter to her law school Dean, as well as her state Bar president.
    Why am I not surprised she is associated with Brad?


    1. The author of the BJS study criticized Wilcox for basing his conclusions on a bivariate relationship she presented. That seems to be SOP for Wilcox. E.g. this report where he was arguing for the importance of fathers by showing the association b/t father’s involvement and children’s outcomes. Seems to me the obvious control would be mother’s involvement.


  3. The exposure mechanism sounds plausible, but the “feedback loop of violence does not,” nor does overfitting a polynomial to 17 observations seem reasonable and then reporting an R^2.

    But you’re absolutely right about the trend in the cloud of observations.

    I don’t think this association is very meaningful in response to Brad’s hypothesis though, because he didn’t make a claim about bad husbands – he made a claim about husbands protecting their wives and daughters from other men.

    That doesn’t make your claim on its own any less meaningful, because I think it’s significant to note the trend you do (additionally – that like another poster has said – married people may have stayed in crappy marriages and beat each other up more in the past).

    You’d want a model that incorporated a measure of women’s exposure to men in and outside of marriages (women may just be less exposed overall, ie more single, rather than substituting out marriages for cohabitation etc) and a measure of the rate of violent marriages against all marriages (per-marriage violence, because marriages themselves may be becoming less violent as a result, say of the advocacy of Catholic therapeutic intervention or lefty pop-psychology marriage books – regardless the rate of exposure to marriage and exposure to men through other means).

    That would be an interesting study and say the things that your polynomial trend can’t — specifically, whether Brad is wrong in his claim on fathers and husbands being good protectors (absent whether assholes beat up their wives — the issue you’re scratching at w the data above — which I suspect Brad is also deeply concerned about and would love to collaborate with smart people over).

    You guys are both perfectly decent human beings in my experience and have perfectly reasonable feelings in light of your starting assumptions about the world.

    Hope I didn’t get anything too badly wrong in the stats discussion. Best regards.


  4. All-time low for WaPo. What is it with newspapers and having the Wilcoxes and Rosins of the world spread falsehoods about sexism, marriage, and sociology?


  5. The graph plotted is essentially an example of how not to plot data.

    1. The X axis is percentage married among women 18-54; why plot that as the independent parameter?
    2. The Y axis is intimate partner violence 12+; why 12+? why not violence rate among married families of the X axis? why is the X and Y axis is even related to each other?
    3. then, the data is for each year; the year could have been the x axis parameter.

    This horrible mess of a figure should have been plotted with the same age range for women, for both percent married and marriage inside the married family. The independent variable should have been year.

    As it stands now, we cannot tell if the violence to women happened within the married families are not. Actually I cannot even explain why the Y axis should even relate to X axis. The correlation over years is so spurious you could have plotted it against anything like,, bottles of wine sold over the year.

    My point is you jump to critique Chetty or Wilcox or Pilketty or Wade, or flavor of the day, making some spurious correlation.

    I recommend George Will next.


      1. Ah I see it now. I was trying to help; spent three hours reading up on this issue and thinking carefully about it. Joke’s on me. Thanks.


  6. To be fair, a similar argument could be made against another popular theory of sexual violence against women: Rape culture (or at least a version of it). Consider the famous feminist dictum: “pornography is the theory–rape is the practice” (Morgan, 1980). Obviously — and several review articles confirm this — there is a strong inverse relationship between pornography consumption and rape rates in the U.S.. In fact, porn acceptance and use is now normative among young adult men (Carroll et al., 2008). I would argue that what Giddens (1992) calls the “transformation of intimacy,” which is simultaneously tied to the proliferation of porn as well as women’s growing equality and increased family and sexual diversity, provides a more helpful theoretical starting point for thinking about recent trends in violence in women’s lives.


  7. I have no dog in the pornography-rape association debate; I will say that one commonly cited study (Todd Kendall’s) that purports to show that Internet access causes a decline in rape (purportedly because of easy access to pornography) has a serious multicollinearity problem.

    With two variables that are extremely difficult to measure (consumption of porn and sexual violence) I find it hard to believe that ANY solid research could be done on this, but I am willing to be proven wrong.


    1. We at least know the direction of bias on both of those variables — undermeasurement. And unless there’s reason to believe that those biases are not constant over time, they wash out in inter-temporal comparisons of rates.


  8. I was referring to rates of porn consumption and sexual violence over time. Is there really any debate about whether porn use has increased over time? Just look at the economic growth of the porn industry, the availability of porn, or, if you need to, responses to survey questions regarding porn use (e.g., compare findings from Laumann et al. in 1994 to those of Carroll et al. 2008). As Sarracino and Scott (2008) note, porn has become mainstream and the mainstream has become porned. And as Chris Uggen notes here (, “rates of sexual violence in the United States, whether measured by arrest or victimization, have declined by over 50 percent over the last twenty years.” Also, here is recent review article ( that reflects on these trends and concludes that “it is time to discard the hypothesis that pornography contributes to increased sexual assault behavior.”


    1. What I found surprising about Levitte’s comments on his former student’s paper cited above) is that he implied that only an economist would find it intuitive that people might substitute porn for sexual violence (rather than them being complements as the not-sex-positive feminists might claim). I think that’s wrong – people recognize in a variety of settings that one thing offers catharsis for another. It’s different language but the same idea.

      I’m not sure that sexual violence is propogated by men who can’t get laid though. The fact that most rape is date rape and committed by men who repeadetly victimize women suggests that the issue here is not the inability of rapists to charm women into the bedroom. We have to keep being vigilant about the “man in the street cuts screen with knife and violently rapes stranger” story. Conjecturing about that man’a motivations isn’t helpful, and I think we’re seeing that story creep in the back door here b/c we’re discussing the mechanism at the aggregate level where the statisitical story/mechanism often gets less clear.

      A straightforward explanation of the drop in sexual violence is that generations of men are learning to view women’s sexual rights differently, and that women feel more empowered and knowledgable about avoiding situations in which they’re in danger or stopping situations that are devolving into coercion. I don’t think we need to reach for cynical, counter intuitive, or thrilling explanations to explain social outcomes, and especially in the area of sex and gender research where those are SO tempting.


Comments welcome (may be moderated)

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s