http://www.timwise.org/2013/08/nazis-cant-do-math-reflections-on-racism-crime-and-the-illiteracy-of-right-wing-statistical-analysis/
After all, whites and blacks don’t encounter each other at rates that are strictly commensurate with our relative population percentages. So because of residential isolation from one another, for instance, whites are not 69 percent of the persons encountered by blacks, and therefore available to be the victims of black murderers, and likewise, blacks are not 13 percent of persons encountered by whites, and thus available to be
our victims either. The
best estimates we have for real-world encounter rates are that about 57 percent of persons encountered by blacks will be white and
only about 3 percent of persons encountered by whites will be black. If we apply these numbers to the interracial homicide data, the results are truly startling and completely eviscerate the white supremacist and racist argument about how “disproportionately dangerous” blacks are to whites.
So, using 2010 data, and applying the rates of interracial encounter noted above (3 percent and 57 percent), this means:
Of 4,765 victims of black killers, random chance would have predicted 2,716 white victims (4,765 x .57); and,
Of 3,896 victims of white killers, random chance would have predicted 117 black victims (3,896 x .03).
In other words, given relative racial offending rates and rates of interracial encounter — the two factors that determine the likelihood of interracial victimization — random chance would predict that there should have been about 2000
more black-on-white homicides in 2010 than there actually were, and about 300
fewer white-on-black murders than actually occurred. In other words, whites killed more blacks than should have been expected and blacks killed fewer whites than should have been expected: exactly the opposite of the white nationalist argument. In terms of ratios, because of relative offending rates and rates of interracial encounter, there should have been 23 times more black-on-white homicides than white-on-black homicides in 2010, but in fact, there were only 1.7 times more of the former than the latter. Meaning that in 2010,
black-on-white homicide happened only 26 percent as often as random chance would have predicted, and the ratio of black-on-white to white-on-black homicide was only one-fourteenth as large as should have been expected!