The upward swinging, exponential models used to curve fit the data from selected counties in Georgia, Alabama and Virginia produce good predictions for the murder rate in the District of Columbia, suggesting that as the proportion of Blacks in the population of a region increases, only a part of the rise in "Black violence" is due directly to the increased concentration of Blacks. An extra criminal factor seems to enter the picture as Blacks acquire the political clout necessary for determining the laws and culture of the region; i.e., as Whites lose that same power. In the affairs of state as well as in ghetto life, it would appear that Blacks are their own worst enemy.

This hypothesis could likewise explain the sudden surge of violent crime rates in South Africa after the change to Black rule in that country, and likewise for other African countries that were once under White rule and relatively peaceful that afterward changed to Black rule and are now relatively quite violent.

On the other hand, there is a competing hypothesis. Whereas it can be demonstrated that the phenomenon of Black violence isn't accounted for by the degree of urbanization — i.e., the violent crime rates of cities are positively well-correlated with the percentage of Blacks in the resident population — urbanization might account for some degree of non-linearity in the graph of certain county-level data. To see that this idea is plausible, refer to the graph entitled Murder Rate versus Blacks in US Cities (it's in Part 13). If you interpolate the crime rate for a hypothetical trend-line city having 71% Blacks, you get a result of 62.6 murders per 100,000 population per year, which is nearer to the actual murder rate for Washington, DC, than the prediction from the trend line for US counties having the largest populations and having Blacks as their largest minority. The urbanization of the jurisdictions represented by the former trend line seems to account partially for the non-linearity sometimes observed in trends for jurisdictions for which urbanization is not controlled.

The non-linearity of the relationship between racial composition and crime rate does not always appear when a study is confined to regions having non-White fractions under 50%. For example, the data spread on the graph below is highly linearly correlated.

Hey Liberals! Do you think that I put any of these points in the wrong place? Check my work!

However, if one were to attempt to extrapolate the murder rate for the District of Columbia from the linear least-squares fit to those data, i.e.,

one would calculate that rate to be 48.2 murders per 100,000 population per year, which is too low. This further supports the hypothesis that White rule mitigates Black violent tendencies to some extent; however, that mitigating effect comes at the cost of exposing White people to the harmful consequences of living among Blacks.