May 30, 2009

Logical priority

As I mentioned, Tyler Cowen of Marginal Review spent some time recently tap dancing around his reader's impassioned question: "Why is Steve Sailer wrong?"

However, wouldn't the more fundamental question be:
"Is Steve Sailer wrong?"

In other words, the reader was asking not for objectivity but for rationalization to quell his Doubts. I suspect Tyler's answer (basically, it's all very complicated, and, besides, Steve Sailer is a bad person) and the pummeling Cowen took in his own Comments section by much better informed people won't serve to drive too many Doubts away.

My published articles are archived at -- Steve Sailer

Andrew Gelman on the Hispanic vote in 2008

Voting data wizard Andrew Gelman has a very good post up: "Where does the Hispanic vote really matter?" reviewing the 2008 election. His conclusion: not too many places.

I said much the same things immediately after the 2000 and 2004 election. Hispanics are not a crucial "swing" vote. They're more of a "flow" vote in that they tend to go with the flow of their white neighbors, just consistently farther to the left. (For example, the high point for the GOP in share of Hispanic vote in House elections was Newt Gingrich's 1994.)

We hear much obsessing over the Hispanic vote, but some of that was a smoke screen made up by Karl Rove. Rove's two big successes -- 2002 and 2004 -- stemmed from mobilizing heavy turnout among white voters and winning a high share of whites. But you aren't supposed to talk about appealing to whites, so Rove did a lot of hand-waving about how Republicans were going to win via Hispanics, and a lot of innumerate journalists bought it.

McCain, the chief Republican spokesman for amnesty in 2006, did poorly in motivating whites to show up and vote, and did mediocre in attracting whites votes, so he lost. Having an economic crash right before the election and the pointlessness of his own campaign other than as a celebration of his vanity no doubt doomed him anyway, but one obvious lesson is that being a famous amnesty enthusiast is a net loser for a Republican candidate -- it doesn't motivate Hispanics (who aren't very excited about making illegal immigration easier, and the ones who are are going to vote Democratic anyway) and it depresses non-Hispanic whites.

In the long run, of course, due to immigration and affirmative action, it's hard to see any successful GOP strategy other than a national version of their success in the South, which is based on carrying 75 percent of the white vote. I think the country would be better off with a competitive two party system in which whites were widely distributed among the two parties, but in the long run, that's unlikely to happen due to immigration. We'll either end up with a competitive system with most whites in one party, or we'll end up with non-competitive, corrupt one-party dominance by the Democrats on the model of the Chicago Machine writ large.

My published articles are archived at -- Steve Sailer

May 29, 2009

"Why Steve Sailer is wrong"

One of the odd side effects of the Ricci case is that it continues to inspire a lot of watery commentary about me.

For example, Tyler Cowen blogs on Marginal Revolution:
"Why Steve Sailer is wrong"

That's a request I received and probably the reader is referring to IQ and race.Let me first say that I am not the Steve Sailer oracle. On such a sensitive matter I don't wish to misrepresent anyone, so I'll simply tell you what I think of the issues, without suggesting that he or anyone else necessarily disagrees.

In other words, this post is going to be about a straw man.

If anybody is actually interested in what I have had to say about IQ and race, well, in 2007 I published FAQs on IQ and race.
There is a belief that progress in genetics will resurrect old, now-unpopular claims about race and IQ, namely that some races are intrinsically inferior in terms of IQ. I very much expect that we will instead learn more about the importance of the individual genome and that variations within "groups" (whether defined in terms of race or not) are where the traction lies. So I don't expect "old style eugenics views" to make a comeback as applied to race, quite the contrary. On that point, here is more.

I also think that IQ will be shown to be more multi-dimensional than we now think. If you wish to understand the role of IQ in human affairs, you would do better to study autism and ADHD than race (by the way, I discuss the importance of neurodiversity in much greater detail in my forthcoming book Create Your Own Economy.)

You may know that some nations -- basically the wealthy ones -- have higher IQs than the poor nations. But IQ is endogenous to environment, as evidenced by the Flynn Effect, namely the general rise in IQ scores with each generation. It is sometimes noted that some racial IQ gaps are not closing but I find it more significant that scores can continue to rise. For instance it is quite possible that groups with higher measured IQs simply have been on an "improvement track" for a longer period of time. More generally I think we should consider the Flynn Effect a bit of a mystery and that suggests an overall tone of caution on these issues rather than polemicism.

Most importantly, there is a critical distinction between hypocritical discourse on race and racism itself. Hypocritical discourse on race is harmful and often Sailer does a very good job skewering it. But racism itself is far, far more harmful, whether in the course of previous history or still today. It is fine if a given individual, for reasons of division of labor, spends his or her time attacking hypocritical discourse about race rather than attacking racism itself. (For instance we shouldn't all focus on condemning Hitler and Stalin, simply because they were among the most evil men; there are other battles to fight.) But I still wish that specified individual to ardently believe that racism is the far greater problem. Insofar as that individual holds such a belief about racism, I am much happier than if not.

The comments section is for discussion of the issues in a mature way; if you want to attack any particular individual, that is for elsewhere.

Addendum: If you are looking for another perspective, here is William Saletan on Steve Sailer.


My published articles are archived at -- Steve Sailer

May 27, 2009

Slate: Well, actually, it isn't a mystery why Sotomayor voted against Ricci

Yesterday's Slate article on Sonia Sotomayor and the Ricci case was so clueless that Slate is back today with a better informed article on the case by a Stanford law prof:
Bad Test
Sonia Sotomayor rejected the New Haven firefighters' claim because it threatened to burn down civil rights law.
By Richard Thompson Ford

New Haven's decision may sound like blatant racial favoritism, but in fact the city rejected the firefighter exam because the test violated Title VII, the federal civil rights law that prevents discrimination in employment. Title VII requires employers to consider the racial impact of their hiring and promotion procedures in order to prevent discrimination that's inadvertent as well as intentional. Ricci's claim is that the city's effort to comply with Title VII is itself race discrimination (under the 14th Amendment to the Constitution and under Title VII itself).

This argument would undermine an important part of modern civil rights law. Some of Sotomayor's critics argue that, in the era of Obama, we no longer need such proactive policies to promote racial equality. But racism isn't a thing of the past yet. In fact, we haven't corrected the lingering effects of racism that is in the past. It's precisely because overt racism is no longer the main impediment to racial equality that the law against inadvertent discrimination is arguably now the most important part of civil rights law.

Well, that's one way of putting it.

Another way of putting it is that there isn't much actual discrimination going on anymore. What there is a lot of these days, however, is below average performance by non-Asian minorities.

There are two ways an employer can discriminate according to Title VII. He can intentionally discriminate by making race a factor in employment decisions—choosing a black candidate over a white candidate because he is black. Frank Ricci claims the city intentionally discriminated when it threw out the exam results because most of the people who scored high were white. An employer can also discriminate by using a selection process that has a disparate impact—in other words, that screens out a particular group for no good reason.

Yes, but what if there is disparate impact for a good reason that is unmentionable: that blacks, on average, aren't as smart as whites? We are supposed to constantly act as if the racial gaps seen on the New Haven firefighters' written test were surprising when they are exactly the same as those seen on, say, graduate and professional school exams. As I wrote in

On the Lieutenants’ exam, the mean black score would have fallen at the 20th percentile among whites. (Hispanics scored the same as blacks). On the harder Captain’s exam, the mean black score fell at the 10th percentile for whites, while the average Hispanic scored at the 18th percentile. (You can see the individual test results at

And, as I wrote on earlier, on the five major grad and professional school tests (GRE, LSAT, MCAT, GMAT, and DCAT), blacks score at the 10th to 18th percentile of white scores.

New Haven claims that the test it tossed out had a disparate impact. Eight black, 25 white, and eight Hispanic firefighters took New Haven's test for promotion to captain; three black, 16 white, and three Hispanic candidates passed. Nineteen black, 43 white, and 15 Hispanic firefighters took the test to become lieutenant; six black, 25 white, and three Hispanic candidates passed. This result counts as discriminatory under the rules of the Equal Employment Opportunity Commission.

I.e., the EEOC's (should-be, but isn't) notorious Four-Fifths Rule. If the highest scoring group passes at, say, a 50% rate, then the lowest scoring group better pass at a 40% or higher rate or the federal government will want to know the reason why? Since there is a one standard deviation difference on average in cognitive ability, only about one-third of blacks score above the white median, not four-fifths. The Four-Fifths Rule assumes the average black IQ is about 96 (when the average white's IQ is assumed to be 100), when it's more like 85.

By the way, have you noticed how ignorant the Four-Fifths rule is simply in terms of simple statistical numeracy? Personnel selection differences should be measured in standard deviations, not in percentages. Using a simple-minded fraction like Four-Fifths invites catastrophic game-playing. For example, the city of Chicago has recently gotten around the Four-Fifths Rule by setting the pass rate so low that 85% of applicants passed the fireman's entrance exam and policeman's promotion exams. This allows the pass rate for blacks to be Four-Fifths of the white rate, but it means that really stupid whites and blacks get important jobs.

We'd get better firemen and policemen with outright quotas.

New Haven was right to worry about the possibility of a lawsuit from black firefighters if it accepted the results of the tests.

The city was also in a bind because its agreement with the firefighters union required that the [written] exam count for 60 percent of the decision about whether to promote each candidate and because a city charter rule required that every promotion go to one of the three top-scoring candidates.

In other words, collective bargaining agreements and Civil Service laws designed to prevent politicians from rigging the test are such a nuisance when the goal is to rig the test.

These rules magnified the disparate racial impact of the exam—no black candidate and only one Hispanic candidate was eligible for promotion, even though several of them passed the test. More reason for the city to worry about a lawsuit by the African-Americans who were to be passed over. ...

But, properly applied, disparate impact law doesn't excuse poor performance, nor does it require quotas. Instead it smokes out hidden bigotry and requires employers to avoid unnecessary segregation of the work force. Suppose an employer wants to keep women out. Knowing that he can't just put a "women need not apply" sign in his window, he might use a proxy, such as a weightlifting test, knowing that women on average have less upper body strength than men.

In other words, the law says that we should be suspicious of any fire department that looks for upper body strength in firemen, and we should force each and every fire department in America to prove, over and over, that being weak is a genuine detriment in fire fighting.

The law against disparate impact discrimination is designed to reveal such hidden bias. Now, suppose an employer has no desire to discriminate against women but uses a weightlifting test simply because he thinks, all other things equal, stronger employees are better than weaker ones. Disparate impact law also prohibits this: It requires the employer to reconsider job qualifications that favor one race or sex, unjustifiably.

Of course, there might be a good reason to prefer people who are physically stronger—or who score higher on a written exam. The law gives employers a chance to prove that the discriminatory criteria are job-related. The idea, then, isn't to make an employer hire less qualified women or minorities over more qualified men or whites. It's to make sure the employer is testing for job qualifications, not unrelated ones.

The theory is that just as it's better to let a guilty man go than to convict an innocent man (so the burden of proof is on the prosecution), it's better to discriminate against a man than against a women and better to discriminate against a white man than a black man.

That's arguable, but let's accept it for now.

In practice, however, the way it actually works out is to encourage permanent de facto quotas for NAMs. If vast mechanisms exists for searching out and punishing those who unwittingly discriminate against NAMs but the punishment for discriminating against whites is haphazard and rare then, well, what do you know!, lots of employers will wittingly (if surreptitiously) discriminate for NAMs and against whites.

But Ricci isn't attacking the timing of New Haven's decision; he's attacking the city for considering the racial impact of the exam. And that's exactly what disparate impact requires an employer to consider. Ricci's position threatens to burn down one of the nation's most important civil rights laws. Even in the improved racial climate of the Obama era, that should set off alarms.

Perhaps in the unimproved economic climate of the Obama era, the fact that one of the nation's most important civil rights laws promotes the hiring of less competent employees should set off alarms.

My prediction for Ricci is that Anthony Kennedy will uphold Disparate Impact in general, but send the Ricci case back to the district court for retrial on the facts on the grounds that the city refused its testing company's offer to perform the validation study included in its $100k contract. Frank Ricci will get the promotion he earned in 2003 around 2012, but the overall system across America will continue more or less the same.

Also, Sonia Sotomayor isn't going to go down over the Ricci case. But it just shows how intentionally ignorant the press is about the realities of civil rights law that Slate could blunder into this like it has.

Sonia Sotomayor v. Frank Ricci

Emily Bazelon writes in Slate:
Judge Sonia Sotomayor is smart and sharp, and her formidable track record on the bench should put to rest any lingering doubts that she isn't. (Speaking of which: Why was the left, or at least the center, criticizing one of its own?) But there is a mystery in Sotomayor's recent history: a brief, unsigned opinion in the difficult race case now before the Supreme Court, Ricci v. DeStefano. Sotomayor punted when Ricci came before her, to such a degree that she raised more questions than she answered.

Ricci is a hard case with bad facts—a case that could do serious damage to Title VII, one of Congress' landmark civil rights laws.

Actually, Ricci is a easy race case with simple facts -- a representative example of how Title VII routinely works. It's only a hard case if your goal is to somehow, someway, preserve the dominant "disparate impact" concept.
In 2003, the city of New Haven, Conn., decided to base future promotions in its firefighting force—there were seven for captain and eight for lieutenant—primarily on a written test. The city paid an outside consultant to design the test so that it would be job-related. Firefighters studied for months. Of the 41 applicants who took the captain exam, eight were black; of the 77 who took the lieutenant exam, 19 were black. None of the African-American candidates scored high enough to be promoted. For both positions, only two of 29 Hispanics qualified for promotion.

Something that hasn't been mentioned is that the the liberal complaints about the 2003 test stem from innumeracy about the effects of a small sample size. In the 1999 New Haven firefighter's promotion exam, which didn't cause major protests by the black minister who is the white mayor of New Haven's chief vote-gatherer in the black community, the racial gap in average scores was the same as on the 2003 test. However, in 1999, two blacks scored far enough out toward the right edge of the distribution that they were promoted. In 2003, however, although the average distribution of scores by race was the same, there didn't happen to be any blacks who scored particularly highly relative to the black average. The difference in whether 2 blacks passed in 1999 or 0 in 2003 is just a matter of small sample sizes.

The politicians didn't raise a stink in 1999, but did in 2003, because their statistical sophistication is at the black and white level.
In other situations like this, minority candidates have successfully sued based on the long-recognized legal theory that a test that has a disparate impact—it affects one racial group more than others—must truly be job-related in order to be legal. You can see why New Haven's black firefighters might have done just that. Why promote firefighters based on a written test rather than their performance in the field? Why favor multiple-choice questions over evaluations of leadership and execution? It's like granting a driver's license based solely on the written test, only with much higher stakes. ...

First, the promotional exam wasn't "solely" written. Forty percent of the score was based on oral exam. And the city had attempted to rig the oral results by stacking the panels of out-of-state senior firefighters brought in to judge the oral results by putting two minorities to one white on almost every three man scoring panel.

However, the city's collective bargaining contract with the fireman's union mandated a 60% weighting for the written test? Why? Well, one reason is that the firemen wanted to be evaluated partly objectively. They didn't trust the politicians to be objective in whom they favored to give them orders in life or death situations, so they wanted at least a majority of the score on the promotional exam to be unbiased by racial prejudice.

The district court judge who heard Ricci's case ruled against him and his fellow plaintiffs. They appealed to the 2nd Circuit, the court on which Judge Sotomayor sits. In an unusual short and unsigned opinion, a panel of three judges, including Sotomayor, adopted the district court judge's ruling without adding their own analysis. As Judge Jose Cabranes put it, in protesting this ruling later in the appeals process, "Indeed, the opinion contains no reference whatsoever to the constitutional claims at the core of this case. … This perfunctory disposition rests uneasily with the weighty issues presented by this appeal."

If Sotomayor and her colleagues were trying to shield the case from Supreme Court review, her punt had the opposite effect. It drew Cabranes' ire, and he hung a big red flag on the case, which the Supreme Court grabbed. The court heard argument in Ricci in April. New Haven didn't fare well.

The high court's decision in the case will come in June, before Sotomayor's confirmation hearings. The problem for her will not be why she sided with New Haven over Frank Ricci. The four liberal-moderate justices currently on the court are likely to agree with her, in the name of preserving Title VII as a tool for fair hiring. There's even an outside chance that Justice Anthony Kennedy will follow along. The problem for Sotomayor, instead, is why she didn't grapple with the difficult constitutional issues, the ones Cabranes pointed to. Did she really have nothing to add to the district court opinion? In a case of this magnitude and intricacy, why would that be?

Well, Ms. Bazelon, I think you may find that you just answered your own question! Judge Sotomayor no doubt shares your goals on policy (preserve Disparate Impact) but is much more aware of the facts. If trying to hush up the Ricci case was the best she could come up with, then that's the best anybody on the left could come up with.

Very similarly, the Obama Administration doesn't want to make Ricci the Waterloo of Disparate Impact. The case is both so representative and so well prepared in details (e.g., notice choosing Ricci over his fellow plaintiffs as the lead plaintiff -- you know how the left loves to dig up personal scandal on sympathetic-sounding conservatives these days, so I suspect the firemen's lawyers carefully chose a guy with the few skeletons in the closet), that they would be happy if Anthony Kennedy just tersely sent it back to district court for retrial on the facts. (Years later, Frank Ricci would probably get his promotion, but the system would continue.) What Obama is terriffied of is the Supreme Court using Ricci as a precedent-setting case.

Hence, Sotomayor's attempt to bury the case is exactly in line with the Obama Administration's desires. Which is hardly surprising, since they nominated her for the Supreme Court.

My published articles are archived at -- Steve Sailer

May 25, 2009


I likely won't be posting new material or moderating comments on Monday evening or Tuesday. Please check back later in the week.

My published articles are archived at -- Steve Sailer