Them vs. Unzism by Ron Unz
The American Conservative, March 4, 2010
With four long replies to my Hispanic crime article having appeared just in the last 24 hours, I obviously cannot respond to every point raised, but I’ll try to address the key issues.
First, Jason Richwine’s article in AlternativeRight (Model Minority?) raises some perfectly valid points.
He devotes the first half of his article to highlighting the substantial discrepancy between the BJS numbers in Table 2005-13 and those in Table 2005-14 and mentions that he took the commendable step of actually calling the BJS in hopes of clearing up the mystery, though to no avail. He argues that since the two tables seem to disagree, we should probably disregard both of them, and abandon the BJS statistics in general. This is not entirely unreasonable, but as VDare’s Ed Rubenstein has previously noted, the BJS data constitutes almost our only solid national data on ethnic criminality, and has been used by all previous researchers for this reason. If we throw it away, we’re left with virtually nothing at all so we should try to avoid this.
I’ve already pointed out that since the ethnic incarceration data is submitted by each individual state, the state-by-state numbers in Table 2005-14 are probably closest to the primary source data and hence most reliable. Furthermore, the BJS report explains that all missing data in that table was filled by interpolation on an individual state basis, which would seem the best approach. I’d speculate that since Table 2005-13 presented aggregate national numbers, some other, perhaps less reliable method of interpolation was used, probably accounting for the difference.
Now choosing between the conflicting government data tables based on properly deciphering the opaque methodological notes written by federal statisticians should make us very nervous. But fortunately we have a much better means available of independently validating this BJS data.
Taken by itself, California contains 30% of all the Hispanics in America, and since very few of these are from the Caribbean — namely the Cubans, Puerto Ricans, and Dominicans who tend to be Hispanic outliers — the state can serve as a good proxy for the Mexicans and Central Americans who represent the overwhelming majority of American Hispanics. In 2006, the Public Policy Institute of California (PPIC), one of California’s premier research organizations, published Who’s in Prison?, a detailed analysis of the California prison population, based upon the specific state inmate and ethnic data. Since the PPIC ethnic results were calculated relative to the adult rather than the total population, this exclusion of children served as a partial age-adjustment.
If you look on p. 5 of the PPIC report, you’ll find ethnic incarceration charts that look remarkably similar to the ones contained in my own article. According to the numerical results provided, adult Hispanics in California are about 40% more likely to be imprisoned than adult whites, and this is generally consistent with the BJS numbers in Table 2005-14 (the BJS numbers were normalized to the total population rather than just adults). This seems to confirm the accuracy of the BJS data which I had used as the basis for my national analysis.
Furthermore, almost 50% of California’s adult whites are over 45 compared to just 20% of adult Hispanics. Since very few individuals over 45 commit crimes, if we adjust for this population difference we get ratios quite close to the age-adjusted California estimates which I had provided in my original article.
We should also remember that California has been America’s most Hispanic large state for decades and also has had a long history of severe racial violence in the prison system; mandatory ethnic segregation has been the standard administrative response, although repeatedly challenged in the courts. Therefore, the ethnic classification of California inmates is probably more accurate than almost anywhere else in the country. So our California ethnic incarceration statistics are probably as solid as any we’re ever likely to see.
The rough agreement between the California PPIC numbers, the California numbers from the national BJS Table 2005-14, and my own age-adjusted California numbers would seem to strengthen the likelihood that all these consistent results are approximately correct.
Richwine’s second major point is less strong. In keeping with some of his previous work, he argues that there has been a very sharp rise in social pathology and criminality among the American-born children of Hispanic immigrants. Although there is certainly an element of truth in this, I already pointed out in my original article that some of the claims frequently made have been wildly exaggerated and unrealistic. Rubenstein has seconded my arguments about the very poor quality of immigration-status imprisonment data.
Richwine correctly notes that a large fraction of Hispanic adults are still of the immigrant generation, but fails to realize this is no longer true for Hispanic adults in their peak crime years, who are mostly American-born these days. Furthermore, the rapid shift of highest-crime-age Hispanics from the immigrant to the American-born generations occurred exactly at the time that Hispanic imprisonment rates and crime rates in heavily Hispanic cities were rapidly declining. This raises serious doubts about whether the crime difference is nearly as large as he is suggesting.
When he argues that the low crime rate in 50% Hispanic Los Angeles is due to immigration status, I think he is missing the local geographical skew. For example, murder rates and violent crime have been enormously higher in Pico Union and the whole South Central region, where Hispanics tend to be immigrants (and also in violent conflict with the local blacks) than on the traditionally Hispanic Eastside, where the adult population is much more likely to be second or third generation. Thus, although Los Angeles crime rates are currently at historic lows, the rates would be much lower still if not for the remaining high crime hotspots in the most heavily immigrant areas. Although I do agree that immigrant Hispanics generally have lower crime rates, I think the difference is small enough to easily be trumped by other local factors.
I have other reasons for being skeptical of Richwine’s generational claims. I’ve personally been arguing this same topic of Hispanic crime rates for over twenty years now, and I remember how at the beginning of the 1990s, I frequently heard exactly the same claim that although immigrant Hispanics were well-behaved, their American-born children would create a massive crime-wave a bit down the road. But twenty years on, those very children are now in their highest-crime years, and meanwhile crime rates have fallen to much lower levels, whether in Silicon Valley, in Los Angeles, or in most of America’s other heavily Hispanic urban centers. Therefore, I would tend to discount this argument until I actually see some hard evidence to the contrary.
Finally, Richwine suggests estimating Hispanic incarceration rates by using Census data and assuming that Hispanics who report living in institutional quarters are imprisoned felons. But as Rubenstein has already pointed out, this category would also include residents of college dorms and military bases, and the numbers actually obtained by this method are frequently so erratic that they seem to make no sense. By contrast, our PPIC ethnic incarceration data for California appears very solid, and perhaps could be used to calibrate the incarceration figures which Richwine is attempting to tease out of the Census data.
I have much less positive things to say about the lengthy response that Steve Burton published on his website (Reply to Unz). Once again, he argues at great length that federal inmates should be included in the incarceration estimate, disputing that the large number of immigration violators or border drug-smugglers might skew the ethnic statistics, and pointing out that a good number of federal inmates have committed “regular” crimes such as robbery or murder. However, anyone who looks at the figures will see that 70% of federal inmates are being held for immigration or drug violations, while just 30% committed regular crimes. These 70,000 federal felons constitute merely 3% of America’s total inmate population of over two million, so excluding them should hardly make a huge difference in our calculations.
Burton is so adamant about including the federal numbers because I had primarily relied on Table 2005-14, which is limited to state+local inmates, and he is firmly convinced that the numbers in that table are simply wrong, although he has never provided any hard evidence to that effect other than pointing out that they seemingly conflict with the Table 2005-13 numbers, which he prefers to use. I’ve already addressed this issue above, pointing out above that the PPIC California numbers are consistent with the BJS numbers I’m using, so perhaps Burton should now seek to explain why the California numbers must also be wrong.
Burton’s next arguments, regarding individual cities, are even weaker.
I’d casually contrasted the crime rates between San Jose and Seattle, arguing that the cities were roughly comparable but noting that although the former has nearly six times the Hispanic population it also has much lower crime rates. Burton devotes several paragraphs to refuting this one example, performing complex crime calculations based on the ethnicity of the two cities, and emphasizes that although I had claimed Seattle was 8.4% black, the correct figure is actually the 11.5% number that he found on Wikipedia. Now this hardly seems like a huge difference to me, but since Burton made so much of it and I dislike having someone uncover an error in my numbers, I went back and checked the official US Census 2000 numbers and the US Census ACS reports, and sure enough my numbers were exactly correct, and the Wikipedia numbers that Burton used in his complex calculations were completely wrong. Given that Wikipedia is unsourced, any Wikipedia number might have come from the practical joke of a drunken teenager, so should be treated with extreme caution rather than blindly accepted. Automatically assuming that a Wikipedia number disproves the figure previously quoted by a quantitative researcher is hardly the mark of a serious analyst.
One fortunate by-product of Burton’s incorrect Wikipedia-derived calculations is that he formulates the explicit hypothesis that Hispanics commit twice as many crimes as whites of the same age. This is a strong claim which can be tested empirically.
The two whitest cities in America are Lexington, Kentucky and Lincoln, Nebraska, which average 81% white. The two most Hispanic cities in America are El Paso, Texas and Santa Ana, California, which average 80% Hispanic. Each of the two Hispanic cities has a crime rates lower than either of the two white cities. Since Burton seems to regard blacks as the source of all evil, and Lexington is 13% black, perhaps we should substitute one of the white urban runners-up such as Portland, Oregon or Colorado Springs, Colorado, each of which has a minuscule black population — but the results remain the same. Los Angeles is 50% Hispanic while Portland is less than 10% Hispanic, and LA is also much more heavily black; yet the two cities have very similar violent crime rates. How could any of these urban comparisons — or dozens of similar ones — produce the real world results they do if Burton’s theory is correct and Hispanics have twice the white crime rate?
Or consider the national cross-correlations I presented in my original article, which represent the polar opposite to “cherry-picking” a few cities to make a biased point. They are based on quantitatively examining the ethnic composition of all American cities across several different crime categories and several different years. And in nearly every case, the White+Asian and Hispanic results were very close or even indistinguishable. How could this be remotely possible if Burton’s theory is correct and Hispanics tend to commit twice as much crime as whites?
A theory absolutely resistant to any amount of conflicting evidence is best understood as constituting part of a secular religion, much like Marxism or the beliefs of those neocons who still continue to claim that Saddam did indeed have WMDs, which have been hidden away somewhere. Since Burton has seemingly chosen not to believe in reality, perhaps he should apply for a job at the Weekly Standard and spend his (well-paid) days tracking down those mysteriously missing WMDs.
As for Richard Spencer’s short piece Thoughts on Hispanic Crime, the first part of his focus is on the negative social indicators of American-born Hispanics compared to their immigrant parents. While some of this is certainly true, especially with regard to “conservative social values,” as I discussed above I think the claims of increased criminality have been greatly exaggerated. Furthermore, my strong impression is that nearly all American groups — immigrant or not, Hispanic or not — have recently shown a similar decline of social values from their parents’ generation, and this has long been a major topic of conservative complaint.
Meanwhile, as I briefly suggested in my article, the exceptionally positive news about Hispanic economic advancement has received almost no public attention. A couple of years ago Doug Besharov pointed out that just during the twelve years from 1994 to 2006, the poverty rate among Hispanics dropped by one-third, while the median individual income also rose by one-third. Most remarkably, this rapid rise occurred despite the continuing and massive influx of relatively impoverished new immigrants, which we might have assumed would produce a sharp decline in all these average numbers. All things considered, I’d think that this broad Hispanic economic rise may be as rapid as that of any group we’ve seen in modern American history.
Admittedly, whether these strong Hispanic gains represent a good thing or a bad thing depends upon your perspective. For example, the percentage of skilled blue-collar jobs held by Hispanics more than doubled during these twelve years, while the white percentage dropped by almost precisely the same amount. For many years, VDare.com has been running a regular “American worker displacement” column, highlighting exactly these sorts of statistics. But arguing that Hispanics are advancing economically and may be doing so at the expense of working-class whites is very different from arguing that Hispanics are failing economically. Although I’m no expert on this subject, my own impression is that the former — as endlessly documented by VDare — is much more likely the case than the latter.
Spencer’s final point is that even if we accept that Hispanics do indeed have approximately the same crime rates as whites, since Hispanics are disproportionately youthful, America’s crime will become disproportionately Hispanic. I certainly can’t dispute this simple logic, though the crime rates in the heavily Hispanic cities I examined hardly seem much cause for concern, since they tend to be the lowest in America.
The very heavy volume of the ongoing responses to my article has led Matthew Roberts to produce a web summarizing the current state of the debate, The Unzism Debate – Around the Web. Since Roberts has himself been a strong partisan in the debate, his perspective is hardly neutral, but his collection of articles should be useful to everyone.
And finally Austin Bramwell earlier this week published a piece on the American Conservative website making some sensible points. As a restrictionist and a conservative, he argues that since the data on Hispanic crime rates does remain somewhat murky while the consequences of mass immigration are irreversible, we should err on the side of caution by opposing it. Whether or not I agree with the actual conclusion, this seems like a very reasonable — and certainly a very “conservative” — position to take. By contrast, trying to pretend that numbers say something they don’t seems more a mark of simple innumeracy than of any respectable ideology.