Arie Peliger is a charlatan

Out of our most renowned, prestigious military academy's Combating Terrorism Center comes a report entitled "Challengers from the Sidelines: Understanding America's Violent Far-Right". At nearly 150 pages, it was difficult enough to hastily skim through, let alone read it from top to bottom. Sample sentence: "The far right has become more vibrant and more ideologically and structurally diverse than ever before." (Wait, is that a good thing?)

I'd like, however, to highlight the jaw-dropping statistical analysis the distinguished author, one Arie Perliger, offers. Jumping to page 96, we're presented with a table ranking states by the number of "far-right attacks" that have occurred in each over the last two decades. Subsequent correlations are then reported. The four variables for which the relationship with the number of attacks in a state are the strongest are Jewish population size (r = .900), state population size (r = .888), Hispanic population size (r = .849), and African American population size (r = .598). By "size", Perliger is simply referring to population and is thus not talking about attacks per capita. The reason California has suffered so many more attacks than Wyoming has  suffered (782 versus 6) is because California has so many more people--Jews, blacks, and Hispanics among them--than Wyoming has! I'm not making this up. Take a look for yourself.

Employing this 'methodology' (that would embarrass a freshman in the second week of Statistics 101), Perliger notes that "the birthplace of groups such as the KKK ... is no longer that natural habitat of the far right. ... The two states at the top of the list are California and New York, which are considered liberal--or blue--in terms of their ideological and political orientation. ... It can be determined that during the last twenty years the violence has shifted from the center/South to the coasts and the North (with the exception of Texas)."

To legitimately talk about a shift over time, Perliger would need to include a dataset from the past to compare to the present, which he doesn't do. That is parenthetical to the fatal flaw underlying his analysis, though. He is essentially putting together a ranking of states by population size, using it as a proxy for far-right violence, and then cobbling together something of a historical narrative about geographic shifts and trends in said far-right violence over time. By Perliger's thinking, if North Dakota and South Dakota combined into a single state, the new state, Greater Dakota, would be twice as susceptible to far-right violence as either state was before the merger! Mind boggling.

Perliger subsequently describes the use of "two-stage hierarchical regression analysis" which is "intended for controlling both state population size and density". I'm not familiar with what exactly that means, but whatever "controlling" entails, it doesn't appear to have much to do with controlling for variables. The correlations between violent attacks and black and Jewish total populations remained robust (.47 and .69, respectively), but for black and Jewish population proportions, the correlations were much weaker (.16 and .11). The relationship with the size of a state's Hispanic population and proportion apparently disappears entirely. From this, Perliger determines that "anti-Semitic and anti-African American sentiments and narratives are still emphasized and dominant ... hence there is a delay in the identification of the Hispanic minority as a threat by far-right groups."

Doing a little legwork with Perliger's data, the positive correlation between a state's hate rate and its total population is a moderate .33 (p = .02). While it's notable that such a relationship exists at all, it renders the commentary Perliger offers meaningless. Here's a state ranking of the annual rate of "far right attacks" per 100,000 people over the period covered:

State	Rate
1. District of Columbia	.303
2. Oregon	.141
3. Maine	.115
4. New York	.110
5. Vermont	.104
6. Massachusetts	.104
7. New Hampshire	.096
8. Washington	.092
9. California	.090
10. Idaho	.088
11. Montana	.087
12. Connecticut	.084
13. Maryland	.073
14. New Jersey	.068
15. Rhode Island	.066
16. Nevada	.065
17. Iowa	.064
18. Arizona	.062
19. Delaware	.062
20. Louisiana	.062
21. New Mexico	.061
22. Wisconsin	.060
23. Illinois	.058
24. Colorado	.058
25. Florida	.056
26. Pennsylvania	.054
27. Minnesota	.051
28. North Dakota	.051
29. Indiana	.051
30. West Virginia	.049
31. Missouri	.048
32. Alaska	.048
33. South Dakota	.047
34. Wyoming	.046
35. Nebraska	.045
36. Kentucky	.041
37. Tennessee	.040
38. South Carolina	.040
39. Arkansas	.038
40. North Carolina	.038
41. Virginia	.038
42. Kansas	.035
43. Oklahoma	.034
44. Texas	.031
45. Michigan	.031
46. Mississippi	.029
47. Alabama	.029
48. Georgia	.028
49. Ohio	.027
50. Hawaii	.009
51. Utah	.005

For comparative purposes, the national violent crime rate per 100,000 people in 2011 was 386.3. DC--home to so many militia types, as is as well-known as it is well established!--has highest far-right hate rate in the country. That alarmingly high rate is 1/1,277th of the national violent crime rate. Mercifully, the national hate rate is only 1/6,895th of the national violent crime rate. Yes, one of America's most pressing problems is indeed its violent, far-right extremists!

This is all really just a distraction, which I suppose is the point, much like designated "hate" crimes in general are (similarly to what is presented in the preceding table, DC has the second highest hate crime rate in the country, while Mississippi has the lowest--the correlation between Perliger's far-right hate rate and hate crime rates at the state level is a statistically significant .43, and with the South Dakota outlier removed, it jumps to .56). But the fact that tripe like this is being commissioned at the highest reaches of the nation's military establishment strikes me as almost indisputable evidence that we are doomed.

Thanks to an email from a reader who prefers to remain anonymous for pointing out James Bowery pointing out Perliger's report. The reader suggests those scandalized by this shoddy work write to the current commandment of cadets to express their displeasure:

Attn: Brig. Gen'l. Richard Clark
W. Pt Mil. Academy
606 Thayer Road
W Pt., NY 10996

Amnesty redux

Today, in tandem with the legislative push for open borders, NPR ran a segment on the Dillingham Commission that, in 1911, found that immigration into the US from northwestern Europe was preferable toimmigration from southern and eastern Europe. It served as a segue into how it is illegitimate to empirically and clinically track and measure differences in tendencies and behaviors of various population groups. Reporter Audie Cornish quoted a professor of sociology who "studies the immigrant experience". Says one professor Alba:

If we could, if you will, rank groups in terms of their desirability, in terms of their ability to assimilate--that kind of thinking is still present. ... [The Dillingham Commission] was overtly racist. There can be little question about the importance of scientific racism in the early twentieth century and the degree to which it shaped the thinking that went into the Dillingham commission report. And we are not as racist today, but that doesn't mean that we are today altogether free of this thinking that some immigrant groups are superior to other immigrant groups."

Census data make it quite easy to rank immigrant groups into the US (see here and here). That Alba and others like him would rather plug their ears, cover their eyes, and reminisce about the stories their grandparents told them about Ellis Island than face reality doesn't change this.

Incidentally, isn't it just wonderful how there is always a bipartisan gang of congress critters glowing eagerly in the limelight as they wait to introduce amnesty X.0? Let's take a look at the lifetime immigration grade cards that NumbersUSA gives members of the sensible, even-keeled, putatively middle-of-the-road party of amnesty-pushers this time around:

John McCain, AZ (R) -- D
Marco Rubio, FL (R) -- C-
Jeff Flake, AZ (R) -- C
Lindsey Graham, SC (R) -- C
Michael Bennet, CO (D) -- F-
Dick Durbin, IL (D) -- F
Robert Menendez, NJ (D) -- F-
Charles Schumer, NY (D) -- F

The group gets an F+. Congress as a whole earns a C. Not one of the eight are to the restrictionist side of the Congressional center, while six of eight are on the open borders side of the spectrum. When Lindsey Graham and Jeff Flake are your immigration hard-liners, you know you have trouble.

Having grown wary of this seemingly perennial amnesty push, I was at risk of apathy this time around, but tripe like the NPR segment has prodded me into action. I've contacted both my senators and my house member expressing my opposition to 'comprehensive' immigration reform. Pithily, I stated that unemployment is high, the need for low-skilled labor has never been lower, and a path to citizenship is a path to more Democratic voters (all three of my representatives are Republicans) with bastard children in one hand while the other hand stretches out to Uncle Sam.

If you're of a similar mind, please do the same. We've risen up and body-slammed the Establishment before. Let's do it again.

Rates of unsolved murder by state

News reports based on FBI statistics show that somewhere between 35%-40% of homicides in the US go unsolved in the US today. Uniform Crime Reporting data, available at the state level (excluding Florida), show that right at one-third of homicide offenders remain racially unidentifiable. These unidentified killers are the ones who haven't been caught, which is why law enforcement is uncertain of their racial compositions.

This means we have access to rates of unsolved murders at the state level staring us in the face. Parenthetically, there are presumably a small number of homicides for which a perpetrator's race is reported even though he is never positively identified, which would explain why the UCR data on the racially unidentifiable comes in at 33% while the FBI reports an unsolved rate in the 35%-40% range. Still, the following estimates must be pretty damn close to the real thing.

The following table ranks states (with the exclusion of Florida for which no data are available) by their rates of unsolved murders. It's like golf here--the higher a state's percentage, the worse it's doing:

State	Unknown
1. District of Columbia	56.1%
2. Illinois	55.4%
3. Maryland	46.1%
4. New York	44.0%
5. California	43.9%
6. Massachusetts	43.8%
7. Rhode Island	42.0%
8. New Jersey	41.8%
9. Michigan	38.8%
10. Connecticut	37.1%
11. Missouri	36.7%
12. Indiana	36.3%
13. Arizona	34.7%
14. Nevada	34.6%
15. Ohio	31.4%
16. Louisiana	31.2%
17. Alabama	27.8%
18. Hawaii	26.5%
19. Georgia	26.3%
20. Kentucky	26.1%
21. Texas	26.0%
22. North Carolina	25.5%
23. Pennsylvania	25.1%
24. New Mexico	24.5%
25. Arkansas	21.7%
26. Kansas	21.1%
27. Colorado	20.8%
28. Virginia	19.8%
29. Oklahoma	19.0%
30. Nebraska	19.0%
31. Mississippi	18.7%
32. Washington	17.9%
33. Minnesota	17.8%
34. Wisconsin	17.5%
35. Alaska	17.4%
36. Delaware	16.7%
37. New Hampshire	15.8%
38. Oregon	15.7%
39. Utah	15.6%
40. Tennessee	14.6%
41. West Virginia	12.1%
42. South Carolina	10.6%
43. Maine	10.4%
44. Iowa	9.8%
45. South Dakota	9.2%
46. Montana	8.2%
47. Vermont	5.6%
48. North Dakota	4.5%
49. Wyoming	4.5%
50. Idaho	3.9%

An accompanying visualization is available here.

I wouldn't fault someone for imagining that rural murders must be the easiest ones to get away with. If I leave my house and head west, it's 600 miles of sparsely populated farmland and prairie from here to Denver. How would anyone find a body buried under the great blue sky out there in the middle of nowhere? Yet rural states dominate the bottom of the list, while DC, by far the most urban 'state' listed, has the inglorious distinction of being the worst of the worst when it comes to positively identifying killers. This article on the subject of unsolved homicides from the Scripps Howard News Service gives an indication as to why. It requires a little reading between the lines, of course:

Despite dramatic improvements in DNA analysis and forensic science, police fail to make an arrest in more than one-third of all homicides. National clearance rates for murder and manslaughter have fallen from about 90 percent in the 1960s to below 65 percent in recent years.

The majority of homicides now go unsolved at dozens of big-city police departments, according to a Scripps Howard News Service study of crime records provided by the FBI.

...

Experts say that homicides are tougher to solve now because crimes of passion, where assailants are easier to identify, have been replaced by drug- and gang-related killings. Many police chiefs -- especially in areas with rising numbers of unsolved crimes -- blame a lack of witness cooperation.

---

++Addition++A sarcastic Steve Sailer comments.

Not serving bologna here

The data, methods, and estimations utilized in the previous posts on white and black homicide rates by state suggest a national black offender homicide rate of 20.6 and a national white (including most Hispanics) offender homicide rate of 3.1. The FBI reports a black rate of 26.5 and a white rate of 3.5, both from 2005. Take that as you will, but I read it as a pretty good vindication of the methodology employed here.

A few plausible reasons my estimates come in a little lower than the FBI's do:

- My estimates excluded negligent homicide but the FBI figures do not. According to the UCR, about 1.2% of homicides are negligent. Consequently, my estimates are marginally understated, but presumably uniformly so across states.

- The data I used only included the "first" (primary?) offender in a homicide. In the case of multiple offenders being charged, my data only included the first of them while the FBI's national figure presumably included multiple offenders.

- Florida--a state with a homicide rate above the national average--isn't included in my data but I assume it is in the FBI's national numbers.

Inspiration

Who owns guns?

Since its inception, the GSS has asked respondents whether or not they have a gun in the house. Seems like a germane topic for consideration at the moment. For contemporary relevance, everything we'll look at here comes from 2000 onward. Fewer than 1% of respondents refused to answer the question and are consequently excluded from all analyses.

John Derbyshire contends that the gun control debate is a contest of moral posturing between two very culturally distinct groups of whites:

This is really about is good old American sectionalism—two big groups of white people who can’t stand the sight of each other. We are eternally re-fighting the Civil War. 

(Yes, white people. Blacks are hors de combat here, as they mostly were in the Civil War, neither side of which liked or trusted them. Sherman would not let colored troops march armed in the Grand Parade. Some black pioneer units marched with picks and shovels, but they were regarded as comic relief by the spectators and newspapers. Nobody cares what the generality of blacks think, no more now than in 1865. This war, like that one, is an intra-white affair.)

While NAM opinion may not feature prominently in pronouncements from either white faction, Steve Sailer argues that SWPLs favor strict control as a means of disarming NAMs in the urban areas the two groups share. A direct assault on NAM culture being of course unthinkable, SWPLs divert their contempt for gun rights from NAMs and towards white rustic bumpkins. To the extent that NAMs factor into the thoughts of said bumpkins, it's that in the bucolic places they call home, low population density means police manpower and response times are too low to be relied upon. Consequently, each man need be responsible for his own self defense.

In the midst of this fray, it may come as a surprise to many to find that whites are significantly more likely to own firearms than non-whites--including Asians--are. The percentages of Americans who have at least one gun in the home, by mutually exclusive racial/ethnic identity (n = 7,214):

Race	Has gun
White	42.1%
Native American	38.4%
Black	20.0%
Hispanic	11.0%
Asian	10.5%

While gun advocates often invoke the weapon as the great equalizer for allowing a petite woman to defend herself from a larger, stronger male assailant, men are more likely to own firearms than women are. The percentages of people who have at least one gun in the home, by sex (n = 8,164):

Sex	Has gun
Male	41.7%
Female	29.9%

Getting back to the racial angle, as previously mentioned, whites aren't monolithic. There is an ideological divide on the issue of gun rights among whites that is reflected in varying rates of gun ownership. The percentages of whites who have at least one gun in the home, by political orientation (n = 3,998):

Politics	Has gun
Conservative	55.4%
Moderate	41.9%
Liberal	26.9%

Gun ownership rates are more than twice as high among white conservatives as they are among white liberals. Circling back to the previous table, though, note that even white liberals are more apt to have guns in the home than blacks, Hispanics, or Asians are.

There is a geographic component at play here as well. The GSS has only sporadically inquired about the type of community respondents currently live in, but from the beginning has asked where they lived at the age of sixteen. Understanding that the following measurement is imperfect not only because people move but also because geographic distinctions recalled from decades in the past are inevitably somewhoat murky, the percentages of whites who currently have at least one gun in the home, by the type of community they were living in at the age of 16 (n = 5,526):

Community	Has gun
Rural	56.6%
Suburban	40.9%
Urban	31.6%

There is a good deal of overlap between political orientation and community type among whites, but they obviously don't map identically. If we distill the white camps down to their most essential elements, we get  single urban SWPLs on one side and married God-and-country farmboys on the other. The proportion of unmarried urban white liberals with at least one gun in the home is 13.6%; the corresponding figure for married rural white conservatives is 72.1%. It's here that the battle lines are most conspicuously drawn.

Pat Buchanan recently wrote the following:

Today, we Americans are a far more heavily armed people than half a century ago. Forty-seven percent of adult males own a firearm. There are 270 million rifles, shotguns and pistols in private hands.

The man knows his stuff, but Pat doesn't provide a point of reference from which to base the assertion made in that first sentence upon. The reflexive gun-buying spree that has occurred in response to calls for stricter gun control from nanny-staters like Obama and Michael Bloomberg notwithstanding, the American public appears to have been more heavily armed in the past than they are today. Making an exception to the 2000 onward rule established earlier, the following graph shows the percentage of Americans with at least one gun in the home from 1973 through 2010, the last year for which GSS data were available at the time of this posting (n = 32,863):

Pat comes to the following conclusion based on the claimed increase in private gun ownership:

Given the loosening of gun laws at the state level in recent years, the gun controllers no longer have the numbers to impose their will on the folks who have a love for, or feel a need for, guns.

While most American households don't contain firearms, many do. The trend has been away from gun ownership (given demographic trends, this really isn't surprising) over the last several decades, but I suspect that gun rights still have the upper hand over gun restrictions, at least at present. Presumably the vast majority of Americans who own guns want to keep them, while there are copious quantities of residents among the gun-free majority (like myself) who sympathize with the spirit of the second amendment, are wary of government intervention into the  private lives of citizens, and who like thinking that because a number of their neighbors own guns, would-be burglars are taking a gamble assuming that they do not.

GSS variables used: OWNGUN(1-2), RACECEN1(1)(2)(3)(4-10)(15-16), POLVIEWS(1-2)(4)(6-7), MARITAL(1)(2-5), RES16(1-2)(3-4)(5-6), YEAR(2000-2010)

Black homicide rates by state

++Addition++In response to Anthony's comment regarding the variances in the New Hampshire and Vermont black offender rates being the difference between one black murderer over five years in the former and five black murderers over the same period of time in the latter, I've noted by asterisk states in which there were fewer than five black murderers per year over the years considered.

Parenthetically, the correlation between a state's white and black offender rates is a statistically significant .52 (p = 0). Removing the low-end black offending states actually mitigates the relationship slightly to .48 (p = 0).

---

We've taken a look at white homicide rates by state. Now let's give black rates a gander. Using an online database out of the University of Michigan that utilizes the familiar SDA interface to pull figures from the Uniform Crime Reporting Data Series' homicide reports from 2009, 2007, 2006, 2005, and 2003 is how we'll go about it.

Murder rates among the total population are a lot easier to determine definitively than murder rates by subgroups within a population are. Murder is unlikely to go unreported--a corpse usually provides pretty good evidence of a homicide if it occurred. Take the number of murders by the overall population over a year and, voila, we have the homicide rate. However, the perpetrator(s) is sometimes unknown. Consequently, the sum rates of any number of non-overlapping subgroups is always going to fall short of the rate for the total population, even if the entire population falls into one of the various non-overlapping subgroups being considered.

To address this, for each state I figured the percentage of homicides perpetrated by blacks among those homicides for which the race of the killer was known and then assigned this percentage of the unknown perpetrator number to the black total. This assumes that the racial breakdown of unidentified murderers mirrors that of their identified brethren. Shows like CSI would have us believe that lots of the hard-to-catch killers are white. On the other hand, structural racism suggests that society often turns a blind eye to blacks killing blacks in the ghetto. Who knows? My guess is that this method understates the true black rates but that the effects of said understatements are pretty uniform across states.

Because 2006 is conveniently both the mean and median year employed, the black homicide rate per 100,000 people is calculated by averaging the number of murders in each state over the included five years and comparing it to a state's total black population in 2006. Both numerator and denominator include Hispanics who racially identify as black.

Only non-negligible homicides, which constitute the vast majority of all murders, are included.

Data are available from all states for each of the five years under consideration with the exception of DC (2009 data only) and Florida, which apparently doesn't participate in the UCR. Estimated black murder rates during the aughts per 100,000 blacks, by state:

State	Rate
1. District of Columbia	38.68
2. Pennsylvania	34.16
3. Wisconsin	30.93
4. Michigan	30.89
5. Indiana	30.74
6. Arizona	29.90
7. Louisiana	29.28
8. Nevada	27.71
9. Oklahoma	27.06
10. Missouri	27.01
11. Tennessee	25.20
12. California	25.13
13. Kansas	24.10
14. Maryland	23.73
15. Arkansas	22.79
16. Ohio	22.34
17. Minnesota	21.58
18. Vermont*	21.24
19. West Virginia	19.70
20. Alabama	19.61
21. Illinois	18.62
22. New Jersey	18.19
23. Texas	18.00
24. Colorado	17.89
25. South Carolina	16.64
26. Virginia	16.26
27. Kentucky	16.16
28. New Mexico	16.00
29. North Carolina	15.40
30. Alaska*	15.39
31. Utah*	15.02
32. New York	14.76
33. Washington	14.53
34. Iowa	14.29
35. Delaware	14.00
36. Oregon	13.94
37. Georgia	13.54
38. Massachusetts	13.45
39. Wyoming*	12.57
40. Rhode Island*	11.77
41. Mississippi	11.20
42. Connecticut	10.57
43. Hawaii*	9.14
44. Montana*	8.78
45. South Dakota*	8.69
46. North Dakota*	8.41
47. Idaho*	8.28
48. Maine*	4.42
49. Nebraska*	3.68
50. New Hampshire*	2.23

An accompanying visualization is available here.

States without areas of high black population density generally appear to do the best. Mississippi is an impressive exception to that rule. If asked to take a stab at which American cities have the most dangerous black populations, I bet people answering candidly would heavily include Philadelphia, Milwaukee, and Detroit in their short-lists. Well, there you have it. I hear echoes of the unmentionable here, as well.

Iowa has been a safe place to live for a long time

Hail found CDC data on white homicide rates by state from 1960. Comparing those with rates from the 2000s* yields a statistically significant positive correlation of .74. More than half of a state's modern white murder rate is 'explained' by what its white murder rate was a couple of generations ago. Everyday things change but basically they stay the same.

* Excluding DC, for which contemporary data appears to be erroneous, and Florida, for which figures on homicide by race aren't publicly available.

White murder rates by state

++Addition++Hail looks at white homicide rates 45 years prior and compares them to the contemporary figures in a post that should be read in full. To his discussion of "Hispanic inflation", I'll note that 'whites' in the now heavily Hispanic states of Arizona, New Mexico, Nevada, and California were more murderous in 1960 than they are today, though Texas has become a bit less violent, as Steve explains in the comments of Hail's post.

---

Steve Sailer is curious about white murder rates by state. One of Steve's commenters pointed to an online database out of the University of Michigan that uses the SDA interface. Because of my familiarity with said interface, I'll give it my best.

The Uniform Crime Reporting Data Series provides detailed homicide data by state with accessible reports from 2009, 2007, 2006, 2005, and 2003. Steve wants a decade of data to account for year-to-year randomness in small states. For now, we'll have to settle for half of that.

There are a few technical issues to address before diving in. Most importantly, murder rates among the total population are a lot easier to determine definitively than murder rates by subgroups within a population are. Murder is unlikely to go unreported--a corpse usually provides pretty good evidence of a homicide if it occurred. Take the number of murders by the overall population over a year and, voila, we have the homicide rate. However, the perpetrator(s) is sometimes unknown. Consequently, the sum rates of any number of non-overlapping subgroups is always going to fall short of the rate for the total population, even if the entire population falls into one of the various non-overlapping subgroups being considered.

To address this, for each state I figured the percentage of homicides perpetrated by whites among those homicides for which the race of the killer was known and then assigned this percentage of the unknown perpetrator number to the white total. This assumes that the racial breakdown of unidentified murderers mirrors that of their identified brethren. Shows like CSI would have us believe that lots of the hard-to-catch killers are white. On the other hand, structural racism suggests that society often turns a blind eye to blacks killing blacks in the ghetto. Who knows? My guess is that this method overstates the true white rate but that the effect is overstated pretty uniformly across states.

Another issue is ethnicity, specifically with regards to the question of Hispanic origin. The data are broken down into five categories: White, Black, Asian or Pacific Islander, American Indian or Alaskan Native, and Unknown. Hispanics may be of any race, meaning most of them are included in the white numbers. This becomes obvious in the visualization subsequently linked to, as the states along the Southwest border are conspicuously more murderous than the rest of the country is (excepting, perhaps, Florida, which apparently does not participate in the UCR).

Because 2006 is conveniently both the mean and median year employed, the white homicide rate per 100,000 people is calculated by averaging the number of murders in each state over the included five years and comparing it to a state's total white population (including most Hispanics) in 2006.

Only non-negligent homicides, which constitute the vast majority of all murders, are included.

Finally, data are available from all states for each of the five years under consideration with the exception of DC (2009 data only) and the aforementioned Florida no-show. Estimated white murder rates during the aughts per 100,000 whites by state:

State	Rate
1. District of Columbia	12.43
2. Nevada	6.64
3. New Mexico	6.63
4. Arizona	6.43
5. California	5.81
6. Oklahoma	4.45
7. Texas	4.38
8. Alaska	4.12
9. Hawaii	3.64
10. Maryland	3.52
11. South Carolina	3.52
12. Tennessee	3.51
13. Louisiana	3.42
14. Missouri	3.42
15. West Virginia	3.05
16. Arkansas	3.00
17. Alabama	2.91
18. North Carolina	2.90
19. Colorado	2.86
20. Georgia	2.68
21. Kentucky	2.66
22. Kansas	2.51
23. Wyoming	2.49
24. Virginia	2.44
25. Washington	2.43
26. Indiana	2.41
27. New York	2.35
28. New Jersey	2.27
29. Mississippi	2.21
30. Michigan	2.19
31. Pennsylvania	2.16
32. Idaho	2.10
33. Connecticut	2.03
34. Oregon	1.98
35. Montana	1.84
36. Rhode Island	1.84
37. Massachusetts	1.74
38. Ohio	1.73
39. Delaware	1.69
40. Utah	1.63
41. Maine	1.60
42. Vermont	1.58
43. Wisconsin	1.34
44. Illinois	1.27
45. South Dakota	1.21
46. Nebraska	1.19
47. North Dakota	1.19
48. Iowa	1.14
49. Minnesota	0.91
50. New Hampshire	0.86

Here's a cartographic visualization (requires Java).

Live free and don't die!

In addition to the white Hispanic/non-white Hispanic factor, proximity to the Canadian border appears to be associated with pacifistic tendencies. The upper Midwest does best (Nazis these Nordics and Teutons are not!), followed by the Northeast.

DC's whites are conventionally thought to be a cut above the rest of the country. They're the elites, after all. Even the white kids with parents who don't get them into private schools are as sharp as tacks. As Steve sardonically asks:

Has anybody checked out what Ezra Klein, Chris Matthews, and Cokie Roberts are up to?

As noted previously, data on DC are only available for 2009, a year in which there were 12 identified white killers and another 87 who went unidentified. That's a small sample size to work with. Fewer than one-in-five DC whites are Hispanic, so that offers little in the way of explanation, either. While it is a completely urban 'state' and the jokes about the district of corruption practically write themselves, I suspect something else is up with the data from the capital.

Parenthetically, when compared to the rest of the developed world, the US does notoriously bad on measures of criminality (among other things). Race, of course, is a major reason why this is the case. Taken as nations of their own, the upper Midwest and the Northeast look just fine when measured against Europe.

Variables used: V2, V16(a), V25,

Wordsum score and age at first marriage

Respondents who are not currently nor have never been married are excluded. Not as stark a trend as I would've guessed, but it runs in the predictable direction. My guess as to why those of the most modest intelligence actually buck the it to some extent is due to a lack of prospects and very delayed emotional and psychological maturation. Only 1% of respondents failed to answer a single wordsum question correctly, so we're looking at the mentally handicapped here.

GSS variables used: AGEWED, WORDSUM, BORN(1)