Thursday, July 09, 2009

Race and violence: When you use arrest statistics as evidence for racial differences in crime, liberals claim that police data cannot be trusted because they're out to get blacks. Next, when you use victim-reported data collected by Bureau of Justice Statistics interviewers (NCVS) to identify the race off the offender, liberals will claim that white victims hate blacks so much, they will lie about the perp's race to the interviewer. They hate blacks more than they hate the person who actually victimized them.

So, we patiently move to the next step and look at what self-report data have to tell us. I looked at The National Longitudinal Study of Adolescent Health where adolescents were asked about their violent activities. Going more in-depth than I usually do, I estimated a logistic regression model with 22 predictors, including four dummies: Hispanic, black, Amerindian and Asian. Whites, then, are the comparison category. All models are based on well over 6,000 kids. Here are the racial coefficients for four different types of violent crime:

Carrying a gun to school in past year

Hispanic .43*
Black -.03
Amerindian .37
Asian .04

Pulling a knife or gun on someone in the past year

Hispanic .50*
Black .97*
Amerindian .24
Asian .03

Armed robbery in the past year

Hispanic .47*
Black .38*
Amerindian .08
Asian -.90*

Seriously injured someone in the past year

Hispanic .39*
Black .32*
Amerindian .19
Asian -.28

*p < .05, one-tail

Hispanics are more likely to be violent than whites for all types of behavior. Blacks are too, except for carrying a weapon to school. Amerindians do not differ significantly from whites. Asians are the same, except they are less likely than whites to commit a robbery. Keep in mind that I tried to eliminate the racial effects with many controls, but race proves to be powerful in this analysis. So, blacks admitted committing more of these crimes, even though delinquency research by Travis Hirschi and Michael Hindelang--two leading criminologists--has shown that black males underreport their criminal activities to interviewers.

Let me guess: whites hate blacks so much, they convince them that they have committed crimes that they didn't actually do.

And I won't mention anything more about Hispanics--I sound like a broken record.


  1. There can be no genetic predispositions that vary by race because if that were the case reality itself would be racist and the only way to be realistic would be to be a racist.

    Thank God reality isn't racist! God hates racists. Racism is a sin. Racism is worse than sodomy. Racism is worse than child molestation. Racism is of Hitler, er, I mean Satan himself!

    Jews suffered and died on Auschwitz for our racism.

  2. Interesting. But why one-tailed statistical tests? I would have thought that, in principle, differences could be in either direction.

  3. BGC: My approach was, based on previous research, we can hypothesize that NAMs have higher rate of violence than whites, and Asians lower.

    You're approach is perfectly valid as well. The only change if we do two-tail tests is that the lower rate of robbery among Asians becomes non-significant.

  4. Commodore2:48 PM

    Not really quibbling with your conclusions, as it seems reasonable, but does this take into account income differences? I've never actually heard anyone deny that NAMs commit more crime, but the (not entirely unreasonable) assumption is that is due to them being poorer (and poorer folks do commit more crime).

    Now, chickens and eggs being a little scrambled here, it does however seem obvious that in our current system, NAM poverty isn't due to The Man pushing them down.

  5. Good question. The survey didn't ask about social class, so the best I could do was control for several factors known to be associated with class: grades, skipping school, suspensions, repeated grades, trouble with teachers, trouble with other students, trouble finishing homework, trouble concentrating in class, frequent residential moves, and feeling safe at school. The racial effects are, not surprisingly, larger before entering all these controls. Genes associated with race could, of course, influence many of these factors. In my next post, I'll look at racial effects without controls.


Meta-analysis of clinical trials: Eat walnuts

I am always looking for easy eating choices that are good for you. This new meta-analysis of 26 clinical trials looked to see if walnuts ma...