Monday, February 6, 2012

Correlation Data on Homicide and Race

Over at HBD Chick's place I posted some data on homicide rates by region.  These data are for murders per million residents, as per Wikipedia:

All the countries in Southern Africa – 373 (2004), 320 (2010)
All the countries in East Africa – 208 (2004), 230 (2010)
All the countries in Central America - 293 (2004), 250 (2010)
All the countries in West and Central Europe - 15 (2004), 12 (2010)

[Example of a particularly nice European country: France - 8 (2006)]

Data by US State came from an anti-death penalty site:

New Hampshire – 10 (2010), 8 (2009)
North Dakota – 15 (2010), 19 (2009)
West Virginia – 33 (2010), 46 (2009)
California – 49 (2010), 53 (2009)
Louisiana – 112 (2010), 118 (2009)
USA nationwide - 48 (2010)

So while the American totals are dramatically worse than West/Central Europe, it's impressive how much it varies from state to state.  Almost looks like it varies with IQ, doesn't it?

Of course psychometrics enthusiasts wouldn't find that too surprising, but neither would race realists find it too surprising if it was correlated with race.  The above data make it look like states that are racially equivalent to Europe are approximately as dangerous as Europe, with respect to murder/homicide.  Since there are no US states with as a high a percentage of blacks as most African countries, we shouldn't be surprised that there are no states topping 200 murders per million inhabitants (and indeed only Louisiana tops 100, but of course DC is at about 234).

More direct data, from victimization reports, indicates quite the same thing (though as Queen Latifah proved recently, noöne will call you out if you ignore victimization data and concentrated on arrest data, which are allegedly tainted by racism).

HBD Chick recommended I do a post on this, so I decided to do a little research and put 2010 data on homicides by state together with data from the census of the same year on race and ethnicity by state.  I used the state-by-state homicide data from the above site, while I found data on race and ethnicity at this US Census Bureau page.

Here are the data:



State Homicide White Black Hispanic
Alabama 62.5 68.5 26.2 3.9
Alaska 37.5 66.7 3.3 5.5
Arizona 61 73 4.1 29.6
Arkansas 54.5 77 15.4 6.4
California 51 57.6 6.2 37.6
Colorado 28 81.3 4 20.7
Connecticut 33 77.6 10.1 13.4
Delaware 49.5 68.9 21.4 8.2
Florida 53.5 75 16 22.5
Georgia 58 59.7 30.5 8.8
Hawaii 18 24.7 1.6 8.9
Idaho 14.5 89.1 0.6 11.2
Illinois 57.5 71.5 14.5 15.8
Indiana 47 84.3 9.1 6
Iowa 13 91.3 2.9 5
Kansas 39.5 83.8 5.9 10.5
Kentucky 43 87.8 7.8 3.1
Louisiana 115 62.6 32 4.2
Maine 19 95.2 1.2 1.3
Maryland 75.5 58.2 29.4 8.2
Massachusetts 29 80.4 6.6 9.6
Michigan 59.5 78.9 14.2 4.4
Minnesota 16 85.3 5.2 4.7
Mississippi 68 59.1 37 2.7
Missouri 67.5 82.8 11.6 3.5
Montana 29.5 89.4 0.4 2.9
Nebraska 26.5 86.1 9.2 7
Nevada 59 66.2 8.1 26.5
New Hampshire 9 93.9 1.1 2.8
New Jersey 39.5 68.6 13.7 17.7
New Mexico 84 68.4 2.1 46.3
New York 42.5 65.7 15.9 17.6
North Carolina 51 68.5 21.5 8.4
North Dakota 17 90 1.2 2
Ohio 43.5 82.7 12.2 3.1
Oklahoma 57.5 72.2 7.4 8.9
Oregon 23.5 83.6 1.8 11.7
Pennsylvania 52.5 81.9 10.8 5.7
Rhode Island 29 81.4 5.7 12.4
South Carolina 64 66.2 27.9 5.1
South Dakota 32.5 85.9 1.3 2.7
Tennessee 65 77.6 16.7 4.6
Texas 52 70.4 11.8 37.6
Utah 16.5 86.1 1.1 13
Vermont 12 95.3 1 1.5
Virginia 46.5 68.6 19.4 7.9
Washington 26 77.3 3.6 11.2
West Virginia 39.5 93.9 3.4 1.2
Wisconsin 26.5 86.2 6.3 5.9
Wyoming 17 90.7 0.8 8.9
DC 234 38.5 50.7 9.1



The first column of numbers is the 2009-2010 average homicides per million residents.  After that are statistics from the 2010 census for two racial groups (whites and black) and one ethnic group (Hispanics).


Here are some correlation coëfficients:
Between homicide rate and fraction of whites ... - 0.584
Between homicide rate and fraction of blacks ... 0.784
Between homicide rate and fraction of Hispanics ... 0.136
Between homicide rate and sum of blacks & Hispanics fractions ... 0.716
Between homicide rate and sum of blacks & adjusted Hispanics fractions ... 0.816.


I wasn't satisfied with just correlating homicide and number of blacks, but adding in Hispanics detracts from the correlation (basically because Hispanics aren't nearly as homicidal as blacks).  So I decided to try to sum blacks plus a fraction of Hispanics, with the fraction chosen to maximize the correlation.  The fraction I eventually chose was 0.3125.  (The latter is certainly not a formal multiple regression, but I've forgotten how to do that so for the time being it will have to suffice.)


In conclusion, the magnitudes of these correlations tend to support the thesis that blacks and Hispanics really are more prone to violent crime than whites, and that arrests of innocent blacks and Hispanics by racist police officers are not responsible for differing arrest rates.  


(Coming up: I found a different site with the interesting "White persons not Hispanic" statistic, which I add into the data mix for comments below.  I hypothesize that homicide and WNH will be even more strongly negatively correlated than homicide and whites.  Rates for Europe broken out by country are available here.)

6 comments:

Audacious Epigone said...

Well, perhaps we should be happy to find that in half a decade, the state-level correlation between NAM % of the population and violent crime decreased from .80 to .72. And you focus exclusively on homicide, which, if memory serves, is more heavily skewed towards blacks than rape and assault are (though less skewed towards blacks than burglary is). Or maybe whites are just behaving worse...

Olave d'Estienne said...

I was wondering if anyone had any opinions on the adjustment I did to the Hispanic population. Is there a name for this?

I can add in data on gringos and Amerindians, if it seems worthwhile.

Olave d'Estienne said...

I did put in additional data on WNH and Amerindian populations. I surprised by both results: the correlation between homicide and total blacks, Hispanics, Amerindians, and Native Alaskans (no Islanders) was 0.715. Also, the correlation between homicide and the number of white non-Hispanics was -0.548.

M.G. said...

Between homicide rate and number of blacks ... 0.784

Thanks for posting these. As a U.S. expat in France, I field a lot of B.S. about 'super-violent America', little do they realize it's actually super-violent Afro-America, although in the last 15 years or so their African and Arab immigrants have been giving them a little wake-up.

Don't know if you've seen this data or are interested, but awhile back I went on a quest for historical black-white crime rates in the U.S. I found imprisonment rates by region for 1910, global U.S. incarceration rates for 1929 to 1935, as well as black crime/incarceration stats from Philadelphia in the mid-1800s (courtesy of W.E.B. DuBois). I naively thought they'd be lower, but even with all the societal strictures back then, Afro crime seems to have been as disproportional as it is today.

Olave d'Estienne said...

Thanks for stopping by. I'll read your site some time.

Olave d'Estienne said...

Those are interesting data M.G. I haven't seen anything like that.

Thus you can see that while the civil right movement may have increased crime, by leaving more NAM criminals on the street, it didn't fundamentally change black criminality.