Abuse of Numbers

The image is of a new ad put out by Ernestine’s Women’s Shelter. It jumped out at me on the Toronto subways this week.

The message is simple, and probably accurate: that there appears to be an association between observing domestic abuse as a child and perpetrating similar abuse later in life. The implication is also clear: that we should use this information to push for appropriate policy change that will address this phenomenon and hopefully reduce the instances of domestic violence.

So what’s the problem?

The problem is that I’m a numerically anal epidemiologist and I cringe when I see numbers presented without context. Regardless of how honourable the intent, statistics without context are essentially propaganda.

The ad states: “approximately 3-5 children in every Canadian classroom have witness their mother being assaulted.” Okay, that sounds bad. But what we should be asking ourselves is… is that a lot? 3-5 kids per classroom is about 10-20% of the child population. Are 10-20% of the adult population perpetrators of domestic violence? I don’t know the answer. What proportion of kids are from low income families? What percentage come from families with substance abuse issues? What percentange are immigrants? The list of other putative covariates potentially associated with perpetrating domestic abuse goes on.

Without those bits of additional information, the 3-5 kids statistics is valueless and presented only for emotional impact.

The ad further states: “70% of men in court-ordered treatment for domestic violence witness it as a child.” The first question one needs to ask is: what percentage of men not in court-ordered treatment witnessed abuse as a child? If the number is comparatively high, then the statistic is meaningless. Without comparator statistics, this kind of context-free presentation of numbers –however well intentioned– is manipulative and propagandist. Tsk tsk.

It reminds me of a dinner party I attended once, during which a woman piped up: “Did you know that 100% of sex offenders use porn?” To which I assholishly responded, “Yes, and 100% of sex offenders also eat bread.” Context is key.

Someone out there is no doubt furious with me for making these arguments. They will shout at the screen: “Does this fool really believe that watching your mother abused would not fuck you up?”

Of course that’s not what I’m saying. What I’m saying is this… It doesn’t matter what I believe; all that matters is what the evidence suggests. There might very well be strong evidence that shows that witnessing abuse as a child compels you to become an abuser as an adult; it seems like a reasonable correlation. But that is not the information that has been presented in this ad. Instead, what we are presented with are statistics without comparators: an attempt to slyly let big numbers appeal to our emotions and not to our rational forebrains. And I’m a little offended by that.

If we, as a society, are going to bring statistics into policy debates about matters as important as domestic abuse, they’d better be meaningful numbers rationally presented.