i crunched the numbers and found a way to get the 19:1 ratio using DOJ data.

according to the 2005 numbers whites make up 80% of the population, and blacks about 13%. if you take the DOJ figures and do some funny math you get 337/0.8 = 419 (337 = the number of white-on black crimes). if you do the same math with the number of black-on white crimes and their proportion of the population, 934/0.13 = 7302. if you divide 7302/419 you get approximately 17 (they could have done the math using slightly different numbers, but this is pretty close considering). what's funny is this is totally ass-backwards, and basically the most inflated and meaningless way you could process these numbers.

the meaningful data is such. given 237,854,954 white people in america, and 37,909,341, the probability of being a white victim of a black murderer is 934/237,854,954 = 0.00039%, the probability of being a black victim of a white murderer is 337/37,909,341 = 0.00089%. what does this mean? you are about twice as likely to be a victim of white-on-black crime in this country, and that doesn't even include death row.

alright, i found a way that makes a little more sense to come to the 19 times more likely to commit a crime. using the same numbers as before, the % of the black population commiting a black-on-white murder is 934/37,909,341 = 0.0025%. The % of the white population committing a white-on-black murder is 337/237,854,954 = 0.00014%. Thus, 0.0025/0.00014 = 17, which i guess checks out with the way you word that statistic "times more likely to commit black-on-white murder than white-on-black murder" all this approach does is amplify the effect of black people being a minority, and white people being the majority. so black people are 17 times as likely to murder a white person as the reverse, but are twice as likely to be the victim of a white person. ain't that a bitch.