Monday, January 7, 2013

Rape is Worse than Bad Charts But Harder to Stop with Blogging

At the risk of turning this into the, "Hey, here's a marginally more misogynistic way to interpret data!" blog, I have some questions about this chart currently making the blogger rounds:

They say their data comes from the national crime victimization survey, but I have some problems with that.

1. The national crime victimization survey doesn't measure the number of falsely accused rapists, as near as I can tell.  The FBI measures "unfounded reports" of forcible rape, but 1. "unfounded reports" is a dumb statistic with definitions varying from police department to police department, 2. as near as I can tell they just measure forcible rape, not all rape (I probably just haven't dug enough), and 3. it comes to 8 out of a 100 reports being false.  I don't get how you go from there to a graph showing that there are two falsely accused rapist for every thousand rapists.

There are lots of studies of false rape accusations, and that is a remarkably low number.  It looks like the way they got there was by creating a vast universe of unreported rapists: "sure, 5 out of 100 rape reports are false, but there are a thousand unreported rapes for every reported one" gets you to a very low number of falsely accused people per rapist (numbers in that example were made up).  But there's a problem with that:

2. The national crime survey has found that two out of three rapes aren't reported, which is obviously terrible.  But this chart looks much more terrible, showing 90 percent of rapists going unreported.  Now it's possible for both statistics to be true: if the reported rapists were committing lots of rapes per person  and the unreported rapists were committing less rapes per person (IE, you could have 66 unreported rapes committed by 66 different people, but 33 reported rapes committed by, say, 3 people who committed 11 rapes each).  But it isn't immediately obvious to me how you can conclude that reported rapists are serial rapists and unreported rapists only rape once.  I just can't find that data in the survey.

So yeah, this chart could be absolutely right and I just haven't dug through the numbers enough.  But at first glance, it looks weird.

I'd add that "number of people falsely accused of rape against the number of people who have raped" is a weird statistic.  For what I think they are trying to show, "you should tend to believe people who accuse other people of rape," the statistic you want is the number of accusations to number of false accusations.  Or maybe they are trying to show that we should prioritize rape victims over victims falsely accused of rape when making policy decisions, in which case they should focus on number of raped people to number of people falsely accused of rape.  Or maybe they are trying to say, "there are way more rapist bad guys than false reporting bad guys, when making policy trade offs we should go after the rapist bad guys," in which case we want to know the number of rapists to the number of false accusers.  But this metric -- comparing victims of one crime to perpetrators of a completely different crime -- doesn't make much sense to me.  It would be like comparing "number of burglars to number of identity theft victims" to show that burgling is a bigger problem.

Edit: The enliven project has posted an explanation of where they got their "data."

No comments:

Post a Comment