On the racial disparity in incarceration rates

According to the Pew Research Center, relative to white men, black men are disproportionately imprisoned by a factor of almost 6.5. In my experience, American liberals believe that racism in the criminal justice system explains the bulk of that disparity or, in any case, most of them talk as if they did. But I think it’s very clear that, whatever anti-black racism exists in the criminal justice system, the effect it has on the black/white gap in incarceration rates is dwarfed by the consequences of the fact that blacks disproportionately commit most types of crime. When I say that to my liberal friends, it’s usually met with utter disbelief. In fact, in many cases, they even deny that blacks disproportionately commit most types of crimes, despite the fact that, in my opinion, the evidence for that claim is completely overwhelming. So I thought about a back-of-the-envelope calculation, which I think should convince anyone that, whatever bias exists in the criminal justice system, it can only account for a relatively small part of the racial disparity in incarceration rates.

Suppose that, as a result of bias against them and keeping all the relevant variables equal, blacks are 20% more likely than whites to be arrested, 20% more likely to be prosecuted if arrested, 20% more likely to be convicted if prosecuted and get sentences that are on average 20% longer. Now, I think it’s way more pessimistic than what is justified by the literature, even if you err on the side of caution. But even under those assumptions, in the absence of differences in offending rates between whites and blacks, you would only expect the black/white ratio of incarceration rates to be slightly above 2, when as I have noted above it’s actually close to 6.5. In other words, even under extremely pessimistic assumptions, bias against blacks in the criminal justice system would only account for approximately 19.8% of the black/white gap in incarceration rates. If we now assume that blacks are only 10% more likely to be disadvantaged at every step of the legal chain, which I suspect is much more realistic though probably still overly pessimistic, the black/white ratio of incarceration rates should be less than 1.5. In other words, under those assumptions, bias would only explain approximately 8.5% of the black/white gap in incarceration rates.

Of course, those are just back-of-the-envelope calculations, which neglect all sorts of complications. For instance, incarceration rates also depend on how likely people are to get parole, which is another point where bias may enter. It also depends, not just on how likely one is to be prosecuted if arrested, but also what one is likely to be prosecuted for exactly, yet another place where bias may play a role. If blacks are more likely, other things being equal, to be prosecuted for charges that carry mandatory sentences, things will less often be equal in court when sentence length is decided. These are things which, if they were taken into account, would probably increase the disadvantage suffered by blacks relative to whites in my back-of-the-envelope calculation. But there are also things my calculation doesn’t take into account that would have the opposite effect. In particular, if blacks are disadvantaged other things being equal at one stage of the legal chain, then things are less often equal at the next stage, which my back-of-the-envelope completely ignores. For instance, if blacks are more likely to be prosecuted, it will be harder for the prosecution to obtain a conviction against blacks than against whites, putting aside anti-black racism in court.

The simplistic model my calculation rests on also does not take into account complications whose effect is difficult to determine a priori. For instance, the only parameter in that model at each stage of the legal process is the mean bias at this stage, but there is no reason to suppose that whatever bias exists in the criminal justice system is equal for any type of crime. Depending on whether bias is stronger or weaker for the most serious crimes, which can send the defendant to prison for a long period of time, not disaggregating could result in underestimating or overestimating what fraction of the black/white gap in incarceration rates can be explained by racism in the criminal justice system. One may think that people have less room to act on their bias in the case of serious crimes, because the stakes are higher and there are probably more safeguards, but one can also think of reasons why bias would be stronger in the case of serious crimes. In order to figure this out, we have no choice but to look at the evidence.

Still, although this back-of-the-envelope calculation ignores a lot of complications, it should be quite sufficient to convince you that black/white differences in offending rates, not bias in law enforcement and the criminal justice system, are by far the major factor in the black/white gap in incarceration rates. This gap is so large that, in order for racism in the criminal justice system to explain the bulk of it, it would have to be totally out of control. Not only is such a level of racism completely implausible on its face, but as far as I can tell, the effect sizes one can find the criminological literature about bias in the criminal justice system are not even close to being as large as they would have to be in order to explain more than a small fraction of the black/white gap in incarceration rates. If you don’t think that’s true, I recommend that you read the review of the literature that Janet Lauritsen and Robert Sampson published about 20 years ago, which makes it clear that, once you control for all the relevant variables, it’s not easy to find bias in the criminal justice system.

In my experience, when I make this point to my liberal friends, they react in one of two ways. First, some of them deny that liberals believe that racism in the criminal justice system explains the bulk of the racial disparity in incarceration rates, but I think it’s only because they now realize it’s untenable. I read a lot of liberal publications, probably more than any liberal I know. I also listen to Democracy Now almost every day and most of my friends are liberals. So I don’t think I’m deluded when I say that most of them would expect racism in the criminal justice system to explain a much larger part of the racial disparity in incarceration rates than what my back-of-the-envelope calculation suggests. I also don’t believe for a second that, if I were able to do a rigorous survey and I offered people who react in that way to bet a substantial amount of money with me that, when you ask self-identified liberals, a majority of them say that most of the racial disparity in incarceration rates is a result of racism in the criminal justice system, any of them would take that bet.

Other people just bite the bullet and say that, in their opinion, such a level of racism in the criminal justice system as would be necessary to explain most of the black/white gap in incarceration rates is not implausible at all. Which is fine with me, but if you’re going to say that, then my back-of-the-envelope calculation still gives you a rough idea of what kind of evidence you need to provide and, if you ask me, you won’t be able to do so. Frankly, as I noted previously, this obsession about racism among liberals, especially in the US, strikes me as completely irrational. It’s not that racism doesn’t exist in the criminal justice system or that it doesn’t explain part of the racial disparity in incarceration rates. But this is a completely trivial claim that nobody in his right mind would deny. Strictly speaking, if you can just find one racist cop who once made a false arrest that led to a wrongful conviction, that’s enough to make it true. The interesting question is how much of the racial disparity can be explained by racism in the criminal justice system and I don’t think it’s very much.

Of course, the fact that blacks commit most type of crimes at much higher rates than whites is only the proximate cause of the racial disparity in incarceration rates, it doesn’t mean that injustice has nothing to do with that disparity. I think ultimately injustice probably has a lot to do with it, but it’s not the kind of injustice most liberals tend to focus on. Indeed, as I have argued, whatever racism exists in the criminal justice system, its effect on the black/white gap in incarceration rates is probably quite small. But I think housing segregation ultimately has a substantial effect on that gap and nobody in his right mind would say that injustice has nothing to do with housing segregation. In fact, it could even be that racism plays a substantial role in the racial disparity in incarceration rates, but it doesn’t mean that racism in the criminal justice system does. At the very least, racism in the past certainly plays a substantial role even today, if only because it produced socio-economic conditions which explain that disparity today.

Indeed, given the socio-economic conditions that so many black people experience in the US, it’s just weird to focus that much on racism in the criminal justice system. When someone lives in a neighborhood where people sell crack on every corner, where 80% of kids are raised by a single mother and 50% of households are below the federal poverty level, it’s really not that surprising that he’s likely to engage in criminal behavior and end up in prison even if there were no racism at all in the criminal justice system. (To be clear, it doesn’t mean that he isn’t morally responsible for the choices he made and that he shouldn’t be punished, but it does mean that you can’t just punish him without trying to address the underlying causes.) If American liberals had not been so thoroughly brainwashed with identity politics, they would spend a lot more time talking about that kind of things. But instead most of them prefer to talk about implicit bias, the number of black actors at the Oscars, etc. If they spent more time discussing how to address those socio-economic issues, I’m sure I would still disagree with them most of the time, but at least I’d feel like they’re addressing the real issues instead of wasting their time on irrelevant stuff.

EDIT: I have added a paragraph in which I give another example of a complication that my back-of-the-envelope calculation doesn’t take into account.

ANOTHER EDIT: I have written a follow-up to this post, where I use the recent literature to estimate the value of the parameters of the model I use for my calculation.