THE BEST DEFENSE AGAINST RACIST LIES IS A STATISTICS COURSE

So I wasn’t planning on posting today. I was perfectly happy with the traffic my most recent post received. I was planning on vegetating after a long week at work, watch a movie with my wife, eat some pizza. I came home, clicked on Facebook to check up on my friends and the first thing I saw was this piece of blather…

Dammit! Now I have to post something.

You see, statistics really don’t lie. People lie…especially racist people. Sometimes they like to use statistics as a way of substantiating their lies. It could be that most racist, since they are not particularly dedicated to the whole “fact” thing, simply do not know or understand statistics. Many, like my friend above, are rather skeptical about educated folks to begin with. My friend often tells me that I’m educated beyond my intelligence. That may be true, but my education does include a couple statistics courses that come in handy when presented with nonsense like this.

So here’s how this works. A racist trolls through some data tables. He’s either looking for a data point he can distort, or he has no idea what the hell he’s looking at. When he finds something that confirms his prejudice, like the data above, he turns it to a meme so that the rest of his small minded cadre can click “like” and spread the word without any of that fancy, shmancy thinking getting in the way. When someone challenges the data, that person is accused of being politically correct (ooh, devastating!) or being educated beyond his intelligence.

Now responding to such ignorance in and of itself is a waste of time and energy. After all, I’m not going to convince my friend that he shouldn’t be racist. Hell, I won’t even be able to convince him that he is a racist. After all, I’m sure he has a black friend somewhere who validates his sense of open-mindedness.

However, there are plenty of folks who may not be racist, but may not be conversant in statistics and data analysis. They can be swayed by what appears to be a legitimate statistical claim. It’s for these people that I spend my time writing responses to mindless hate rather than relaxing on a Friday evening. I hope they appreciate it.

So let’s get to the educating.

This particular meme, to the best of my knowledge, comes from a table that looks like this…

So I’m a racist, and I see this table and I think, ‘Wow!’ This table is great for confirming my mindless prejudices astute, politically incorrect, observations about the world.’ Look, there are 117,640 white women being raped (actually, the number represents rape and sexual assault, but…you know) every year (this table does not specify gender but…you know). Of those rapes, 16.4% are committed by black men. Let’s see, 16.4% of 117,640 is 19,292.96…let’s round up…white women are being raped by black men. However, 0% of black women are raped by white men. Score! this would make an awesome racist politically incorrect meme!

Now for the education part. This data comes from the National Crime Victimization Survey. It’s a very reliable source of information, albeit with some weaknesses. The idea behind the survey is to get a profile of crime from a victim’s perspective. This is intended to compliment the information that we can get from the Uniform Crime Report, which is based on reported crimes. The NCVS is interested in collecting data on estimates of unreported crimes. The researches do a great job, and the NCVS is an invaluable resource for understanding crime in the United States.

What it is not, is a precise measure of the actual number of crimes committed in the nation. It’s a profile. Here’s how it works, more or less. The researchers who work for Department of Justice conduct surveys of a select number of individuals over twelve years of age in a select number of households. The NCVS does not call every household in the country. That would be daunting. Instead, researchers collect a random sample of households representative of the nation as a whole. In 2008, the year for which the statistics were gathered for this report, the researchers interviewed 67,090 individuals in 38,060 households throughout the nation. This is pretty good.

Wait a minute! if they only talked to 67,000 people, how did they figure that over 160,000 people were raped? The magic of statistics…which isn’t at all magical. They used the most recent census data to extrapolate their sample size relative to the population as a whole to get their numbers. So yes, that means that the number 117,640 white rape and sexual assault victims is not the actual number of white rape and sexual assault victims. It’s a model that, combined with data collected over time using the same methodology, allows us to analyze trends and consistent characteristics relevant to the subject.

This isn’t statistical trickery. Anyone and everyone who is interested in understanding how the Department of Justice produces these models can simply look up the methodology. Racists don’t care much about methodology.

Well, okay, but what about the numbers? Clearly, the numbers in your so-called model still confirm the conclusions of the meme.

Not so much. There’s a big difference between the data that is presented in the meme, and the data in the table. That big difference is encapsulated in this  (*). The hard working little asterisk! We always hear that the truth is in the details. Well, in statistics, the truth is often in the asterisk.

Here’s how it works. of the 68,000 people surveyed for this research, a certain number of them claimed to have been raped or sexual assaulted. This table does not give us that number.  Of that number of white women who reported having been raped or sexually assaulted, 16% identified their attacker as being black. The presence of that magic asterisk, however, indicates that this constituted fewer than ten cases in the sample. This is problematic to the study, because the researchers cannot get a significant result from such a low sample size. They have to report their findings, but they must make note in their results.

That magic asterisk is really important. Since African Americans constitute thirteen percent of the population, this data suggests that African Americans are over-represented among this group, but since the sample size is so low, there are limits to how much we can rely on this data for this purpose.

The same magic asterisk applies to the 0% of white offenders raping black victims. This data suggest that in this sample, there were no, or perhaps less than .5% of white offenders against black women. But since the sample size is so low, the data is not reliable for this purpose. We can also see in about a quarter of rapes or sexual assaults against black women the race of the offender was not known. We could speculate that at least a good percentage of those unknowns were white, but again, the speculation would be unreliable because of that blasted asterisk.

You’ll notice that the little asterisk is missing from the meme. Racists hate asterisks!

So you might ask yourself, with all of these asterisks flying around, what exactly is this data good for?

Well, it’s clear that the sexual victimization of black women is, according to this data, a largely black on black crime. This is consistent. If you go back through previous years’ data you see similar trends. This is important to know when formulating policy responses to rape and sexual assault. Why is black on black rape so disproportionately higher than one would expect by chance? I would wager it is more a reflection of our historically segregated society moreso than it is an indicator of any particular virtue of white men. After all, the only number on the table without an asterisk is the percentage of white men raping white women.

What this data does not support is the paranoid assertion that black men are targeting white women while white men are no threat at all to black women. That is the message being presented by this and similar memes, and it is a message that must be confronted.