That makes a lot more sense. I tried to figure out what you were doing math wise to make that happen and I couldn't come up with anything coherent.
Here is a secondary source, with info to link to the primary.
Interesting, nothing I'm really surprised by (the criminal justice system in general fucks you over if you are poor, black or male and the effects of all three compound) but it doesn't hit the million dollar question, which is how often wrongfully convicted people are convicted in place of an actual perp of a different race? I don't know if the data even exists to answer that one, though I suspect the set of cases where both the wrongly convicted killer and the actual killer are both known is probably too small to do anything useful with.
If you don't see what I'm getting at, if a black man is wrongfully convicted of a murder committed by another black man, that doesn't impact the racial breakdown because it doesn't change what group the perpetrator belonged in. Is the implication that there is a number of black men wrongfully convicted of homicide where the actual killer was white large enough that it would have a meaningful impact on the stats?