Since the equation of the form

1 = A (mestizos) + B (Blacks) + C (everybody else)

is the equation of a plane, and since this plane must intersect each axis at the point where the racial group it measures is 100% of the total population, a more natural way of showing the crime rates as a function of race would be to use the triangular plane segment having those three vertices. This triangle includes all the mathematically valid proportions of the two groups specifically measured and the group into which all others are lumped together, and furthermore every point in the triangle, including those on the borders, must represent a valid proportion of those groups. So we can simply take this triangle and stand it upright, like so:

When we fill in such a triangle with the government's data for murder rates in US cities, we get the following picture:

And when we do the same thing for the fifty US states, we get this picture:

Essentially, the same pattern occurs regardless of the degree of urbanization involved. The least murder occurs where the Blacks and mestizos are proportionately the scarcest, and the murder rate rises somewhat faster with increasing percentages of Blacks than it does with increasing percentages of mestizos. The overall decrease in the rates, moving from the graph with the city murder rates to the graph with the state murder rates, is most likely the result of the fact that the cities have a higher fraction of Blacks and mestizos in their populations than the rural areas do. In other words, the phenomenon of urban crime isn't (primarily) the result of crowding, per se, but rather of the races of the people who are being crowded. If the countryside had as high a percentage of non-Whites as the cities did, there's some reason to expect that the murder rates would be roughly the same, perhaps with only a small increase due to the higher population density of urban areas.

Anyway, out of the 74 cities in my sample, I selected the 31 cities having at least four times as many Blacks as Hispanics. I did that to focus on the murdering tendencies of Blacks and minimize the contribution of the mestizo group as much as possible. The included cities were Philadelphia PA, Detroit MI, Indianapolis IN, Baltimore MD, Jacksonville FL, Columbus OH, Milwaukee WI, Memphis TN, Washington DC, Cleveland OH, Nashville TN, New Orleans LA, Kansas City MO, Virginia Beach VA, Charlotte NC, Atlanta GA, St. Louis MO, Tulsa OK, Pittsburg PA, Cincinnati OH, Minneapolis MN, Toledo OH, Buffalo NY, Louisville KY, Birmingham AL, Norfolk VA, St. Petersburg FL, Lexington KY, Baton Rouge LA, Akron OH, and Raleigh NC. While each of these cities has at least four times as many Blacks as Hispanics, the Black percentage of the population runs from 13.0% (Minneapolis) to 75.7% (Detroit). Graphing the murder rates per 100,000 population versus the percentage of Blacks gives a well-correlated run of points up the chart.

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

The next time a Liberal tells you that Black crime is the result of "the degree of urbanization," you can prove him wrong with that graph. When the urbanization factor is controlled (held constant), more murders still happen where more Blacks are. The least-squares fit to the data is

Y = 0.97388 X - 6.59434

where Y is the murder rate per 100,000 residents per year, and X is the percentage of the city population that is Black. An all-Black city would probably have an annual murder rate around 91 murders for each 100,000 residents. Obviously, no city can have a negative murder rate, so all that can be said of an all-White city is that it would probably have very few murders indeed.

Possibly, the negative intercept could be removed if the model used for the data were non-linear to some extent. For example, the least-squares parabola through all of the data excepting Washington, New Orleans, St. Louis, Cincinnati, and Kansas City is:

Y = +0.006816492 X² + 0.1080996 X + 8.225503

For the cities most distant from the trend curves, there is most likely some factor other than purely population composition has affecting the crime rates. That factor might be local official corruption, laxitude in the curbing of Black crime by the police, or a political disempowerment of Whites in the area. I don't have the data necessary for a determination.

When all serious violent crimes are included in the study (murder, aggravated assault, rape, and robbery), the least-squares trend line is

where Y is the annual rate of serious violent crimes per 100,000 residents, and X is once again the percentage of the city's population that is Black. (Yes, I used a linear fit this time, instead of an exponential function model.) A typical all-White city would probably have about 590 serious violent crimes per 100,000 residents per year. A typical all-Black city would probably have around 4000 serious violent crimes per 100,000 residents per year. The Black-to-White ratio of the per capita rates for violent crime perpetration for urban areas, thus estimated, is about 6.8, which is commensurate with the ratio when the same rates are estimated across the whole United States. The famous "degree of urbanization" hypothesis to explain Black violence is another liberal myth.

(Added 29 October 2002.) After further examining the dependence of per capita crime rates on both racial demographics and on the degree of urbanization, I find that I must amend what I said above as follows:

THE EFFECT OF URBANIZATION.

Urbanization does have an effect on the per capita murder rates for both Blacks and Whites.

Whites in the countryside commit about 25 murders per million Whites per year. Whites in the cities typically commit about 82 murders per million Whites per year.

Blacks in the countryside commit around 230 murders per million Blacks per year. Blacks in the cities commit about 872 murders per million Blacks per year.

An average Black is 9.5 times more likely to commit murder than an average White.

An average city dweller is 3.5 times more likely to commit murder than an average resident of the countryside.

The effect of race is thus found to be 2.7 times greater than the effect of urbanization.

I've written a program in GWbasic that can calculate my estimate of the expected per capita murder rate in an area, given the overall population density and the Black percentage of the population.