Concerning the prison system/crime in general... so many people simply love the idea that criminals are less intelligent than good upstanding citizens, and this assumption seems to be widely held. The media, Hollywood (see: The Shawshank Redemption... although it's a great flick everyone in there with the exception of Andy Dufrene is a dumb fuck, esp. Haywood), etc. cram this image down our throats.

And then, racial supremacists, ignorant mf's, whoever take this (flawed) assumption and look at the nation's jails and say, "Oh look, how convenient! Our prisons are black. Prisoners are by definition inferior mentally. Therefore, from premise A and B, we arrive at conclusion C: Blacks are mentally inferior. The fact that the prisons are filled with black inmates simply reinforces our assumptions." Aside from being flawed inductive logic, the premises are wrong as well.

Any serious academic study (I think University of Minnesota did one in 1991) shows that, when compared to the general population, median IQ among inmates is just as high as the median IQ of people on the outside. Now, where am I?
1) Most prisoners are black. This is a simple observation.
2) IQ tests have an inherent bias against black test-takers. M2 stated this and I happen to believe it.
3) The only people in prisons are the criminals who were dumb enough to get caught. No one really knows, but pessimistic estimates say that law-enforcement agencies only solve at most 12% of all crimes, and that only 40-45% of all crimes are even reported. This means that there is the possibility of a huge criminal population outside the prison system, and if you believe intelligence is positively connected with rate of capture, they could be substantially smarter than people think.

Let's talk a little more about #1 up there. Most inmates are black. But how many are inmates and shouldn't be? Unfortunately, the answer is, most likely, a lot. So. Inmate IQ tests report equal scores to the general population. The inmate population will be almost exclusively black, while (for reasons M2 elaborated on above) the general population sample will be almost exclusively white. That's not a random sample, and messes up a statistical comparison. The dependent variable here (being a criminal) is all fucked up due to the fact that, in the prison sample there are probably a lot of people who aren't criminals and in the general population sample there are probably a lot of people who are criminals, but it's indeterminate who is and isn't one. So all these variables are completely fucked up, but the same test returns the same results for each group. I take this to mean that there is something in the test itself that affects the outcome, and consequently, the IQ test probably has no fucking statistical relationship to intelligence whatsoever.

Short answer - I agree with M2