Saturday, 27 December 2014

Damn Statistics

     Mark Twain (crediting, perhaps mistakenly, Benjamin Disraeli) famously said there are "lies, damn lies, and statistics". Fox News appears to have warranted the addition of a new category: damn statistics. The following image has come across my Facebook feed several times in the last little while:

     I think this is one of the more egregiously irresponsible things I've seen this year. The numbers themselves are accurately reported from the FBI statistics, which you can confirm for yourself. But it's appallingly unhelpful to pick a statistic like that without giving appropriate background into how to interpret it. Reading it only as it appears above, you could be forgiven for concluding that white people are peaceful law-abiding decent folks, and black people are dangerous violent beasts, which is almost certainly what the people circulating this screen capture want you to believe. (Is that what Bill O'Reilly and Fox News want you to believe? I'll stop short of claiming that; all we have here is a screen capture taken out of context, and I'm sure a fair and balanced reporter would have taken great pains to explain the limitations and significance of the data.)

     So the first thing I'd like to point out about the actual data is this: it's compiled by the FBI from the reports of the nearly 18,000 state and local law enforcement agencies operating the United States. Each of those 18,000 agencies has its own staff, and its own standards and procedures for investigating, reporting and categorizing crimes. Not all of them report on the race of the suspect, and if you look at the FBI table, you'll see that 4112 murders are attributed to an offender of "unknown" race (and 249 "other", which I'll ignore for the purposes of this essay). That's almost as many murders as the total attributed to whites, a huge hole in the data if you're trying to establish a correlation between violence and race.
     How many of those 4112 murders were committed by whites and how many by blacks? There's no way to know from what's reported here. We probably shouldn't assume that it matches the proportions of the murders where race is reported, though, because the very decision to treat race as a relevant factor in reporting crime statistics may reveal institutional biases, either in the police agency or in its community or both. A police agency that tracks crime by race may be more prone to racial profiling, and thus more likely to catch black offenders than white ones. So, a fair number of those 4112 "unknown" murders may in fact be committed by whites who got away with it, and some of the 5375 murders attributed to blacks may well represent the wrongly accused.

     But, you might observe, even if all of the 4112 "unknown" murders were committed by whites, that still means that there were 8508 murders by whites and 5375 by blacks, which is still grossly disproportionate to their respective shares of the population at large. A race that makes up 13% of the population should not account for 39% of the murders!
     True enough, but there's some other important information missing from the FBI data: economics. We know (and have known for thousands of years) that there's a pretty strong correlation between poverty and violent crime; wealthier communities tend to be less violent than poorer ones. (I would say less criminal, but it's complicated; white collar crime is still crime, but it's not generally violent, and the worst white collar crime isn't even treated as crime at all.) We also know that in the United States, wealth disparities are huge, correlated with race. In 2009, the median income for white families was $62,525, and for black families it was $38,409. (The same table shows "Asian and Pacific Islander" median income at $75,027, which might explain the almost negligibly small 249 "other" murders I said I'd ignore.)
     So it would be interesting to see how the FBI stats would look if broken down by family income. I'd expect that once you correct for income, the murder rates of whites and blacks are much closer to the same. They're probably not quite equal, mind you, because many law enforcement agencies in the U.S. tend (intentionally or not) to be more zealous in pursuing black offenders than white. But they're equal enough that we should not be reinforcing these stupid and destructive prejudices.


  1. I don't think your two objections would impress a racist. For the first one, you are simply arguing that the evidence for blacks being more violent than whites is weaker than the screenshot suggests. Maybe you're right, but it's still pretty strong evidence! Your second objection doesn't make much sense at all to me, because you don't appear to be actually objecting to the thesis that blacks are more violent. Instead, you are trying to offer a somewhat palatable explanation for why the thesis is true--blacks are poorer, and poverty leads to violence. The problem is, even if you are correct about the explanation, it doesn't change the fact that blacks really are more violent!

  2. You're right, but impressing racists isn't actually my objective; generally speaking, committed racists (like committed ideologues of any stripe, including I suppose anti-racists) are highly adept at rationalizing ANY evidence so that it supports or at least doesn't contradict their prejudices. I am instead hoping to provide the critical thinking tools to allow an open-minded reader (whether they already lean towards or away from any particular ideology) to recognize that the "obvious" conclusions someone may try to lead you to through selective statistics like the above are not nearly so obvious when you probe a little deeper.

    In this case, I don't dispute "the fact that blacks really are more violent" at least in terms of the correlation between skin color and violence at this point in American history. What I dispute is the pernicious inference that this correlation means something more than just correlation; it does NOT follow that violence is an essential feature of blackness. Violence is higher in impoverished communities, and higher still in communities that do not enjoy the same relationship to police and courts that we take for granted in more well-to-do neighbourhoods. If my neighbour's stereo is too loud at night, I know I can complain to the police who will politely ask them to turn it down, and no one will get shot or go to jail. But there are communities where people quite justifiably do not trust the police and the courts to help resolve disputes, and that leaves people with self-help solutions.

    How do we fix this? Well, part of the problem is the self-reinforcing belief that blacks are inherently more violent, because it inclines us to act in ways that give predominantly black communities even more reason to avoid working with police to solve problems in constructive ways. And that's why I'm trying to attack that belief.

  3. You seem to want truth in order to make an argument. Truth is easily set aside when you have made your point. You show too much of your hand in this blog.

  4. I'm not sure you show enough of your hand in this comment. I don't have a clue what your point is.