W. Bradford Wilcox and Robin Fretwell Wilson argue in PostEverything that marriage is the best way to reduce violence against women — and that, by implication, unmarried women have only themselves to blame if they or their children are exposed to violence. This isn’t a surprising argument to hear from Wilcox, director of the National Marriage Project, which lists “divorce, cohabitation, and fragile unions” as “risks to children’s well-being and healthy development” on its homepage. But while their arguments appear to be substantiated by social science research, the most cursory examination of their data reveals a mess of logical fallacies and selective data use.
The most basic rejoinder to their argument is that both marriage rates and rates of intimate partner violence (along with violent crime in general) have been falling steadily over the past 20 years. If their basic claim that more marriage = less violence held up, one would expect to see a rise in violence against women as marriage rates plummeted. The data show exactly the opposite. Unless one is prepared to argue that marriage causes violence — at which point Wilcox and Wilson would surely be screaming “correlation does not equal causation!” — it’s logical to assume that something else must be going on.
Wilcox and Wilson rely heavily on data from the Bureau of Justice Statistics’ National Crime Victimization Survey (NCVS) Special Report on Intimate Partner Violence, 1993-2010. In particular, they focus on a chart that shows that single women with children experience markedly higher rates of violence than all other family categories, and that married women report the lowest rates, and use this to argue for a causal link between marriage and the likelihood of violence.
But as Shannon Catalano, the study’s author, has already noted, this chart measures only a single variable — conveniently, the one that supports Wilcox and Wilson’s argument. Once one starts controlling for other factors, the picture becomes more complicated. Victimization is significantly higher for young women, between ages 18 and 24. Women are, on average, getting married two years later in life than they were in the 1990s, meaning more married women are likely to be out of the highest danger zone for partner violence. What it doesn’t mean is that marriage itself reduced that incidence of violence.
And while the NCVS includes data from crimes — whether or not they were reported to the police — it is based on self-reporting from victims of violence. Nowhere in their analysis do Wilcox and Wilson acknowledge the possibility that reporting rates for violence may vary based on whether the victim is married to her abuser or not. Some studies have shown that married women are much more likely to report abuse (to anyone, not just the police) after they leave an abusive relationship, and that most of the abuse they report occurred during the relationship, not after it was terminated. (See page 37.)
Wilcox and Wilson also fail to acknowledge one notable caveat, which is that the NCVS does not record the marital status of victims of violence at the time of the incident of violence, but at the time they were interviewed for the NCVS (see footnote c to Table 1, NCVS SRIPV 1993-2010). While the status may be the same in plenty of cases, this hardly makes for a bulletproof argument about the protective power of marriage.
Of course, being a single mother correlates with some other socioeconomic factors, chiefly poverty. While marriage advocates are often quick to point this out, they confuse cause and effect. Studies show that both men and women who have financial assets are more likely to marry than those who don’t. These people are also more likely to live in safe neighborhoods and have the financial means to leave an abusive relationship, while poor women may remain trapped with a violent partner (married or not) because they don’t make a living wage capable of supporting themselves and their children independently. This is why other studies on domestic violence have pointed to developing poor women’s economic independence as the key tool in reducing exposure to violence.
The idea that economic insecurity might play a role in creating either perpetrators or victims of violence seems to be of no interest to Wilson and Wilcox. Instead of addressing the root causes that make women’s and girls’ lives dangerous — continued economic inequality for women and a culture that constantly devalues women’s bodies and lives — they prefer to offer reactionary victim-blaming arguments dressed up with statistical sleight-of-hand that does not meet the most basic standards of social science rigor.