I have long been fascinated by the fact that most conservatives support racial and ethnic profiling for national security and law enforcement purposes, yet are categorically opposed to the use of racial or ethnic classifications for affirmative action. Most liberals, by contrast, take exactly the opposite view. Both ideologies oppose racial and ethnic classifications as a matter of principle in one area, yet defend them on pragmatic grounds in another. Consider, for example, this recent Weekly Standard article by Philip Terzian defending ethnic profiling in airport security:

[T]here is no harm in acknowledging that the sort of person who is likely to be a terrorist is not just any citizen who happens to walk into an airport, but someone with specific, comprehensible characteristics of age, national origin, sex, religion, and behavior. So far as we are aware, no jihadist plots have been perpetrated against Americans by little old ladies from Dubuque, but several terrorist attacks--in particular, 9/11--have been carried out by young Muslim men of Middle Eastern origin. No, not all young men, not all Muslims, not all people from the Middle East, are jihadists or potential terrorists. Of course not. But common sense, and the overwhelming preponderance of evidence, should make it obvious to airport security personnel where to concentrate their energies.

Terzian is saying that ethnic profiling of airline passengers is justified because, on average, a young Middle Eastern Muslim male is more likely to be a terrorist than members of other groups. This, despite the fact that not all (or even most) Middle Eastern Muslims are terrorists, and there are of course some terrorists (Richard Reid, Tim McVeigh, etc.) who belong to other groups. The harm to innocent Middle Eastern Muslims affected by profiling is presumably outweighed by the benefits to national security.

Defenders of affirmative action, of course, make a very similar argument. On average, an African-American or Hispanic applicant to college is more likely to be a victim of racism and to suffer from the historical legacy of Jim Crow and slavery than a white applicant is. Thus, it makes sense to give preference to applicants from these groups, despite the fact that some of the beneficiaries will be people who haven't suffered much from racism, and some of the members of the non-preferred group may themselves be disadvantaged. Defenders of AA also claim that the average black or Hispanic applicant contributes more to campus diversity than the average white one, although there are of course many individual exceptions to this rule. Paraphrasing Terzian, an AA defender could say:

There is no harm in acknowledging that the sort of person who is likely to be a victim of prejudice is not just any citizen, but someone with specific, comprehensible characteristics of race, national origin, or ethnicity. So far as we are aware, few whites from Dubuque have been systematically victimized by racial prejudice. But numerous African-Americans, Native Americans, and Latinos have. No, not all African-Americans, not all Latinos, not all American Indians, are suffering from the effects of past and present discrimination. Not all will contribute more to diversity than the average white applicant. Of course not. But common sense, and the overwhelming preponderance of evidence, should make it obvious to college admissions officers where to concentrate their energies.

In both cases - terrorism profiling and affirmative action - race or ethicity is used as a proxy for other characteristics in order to help overcome the problem of imperfect information. If we knew who is a terrorist and who isn't, there would be no argument for security profiling. If we knew each college applicant's degree of victimization by racism or degree of contribution to diversity, the case for racially based affirmative action would be greatly weakened. Since we don't know these things and it would be difficult or impossible to find out, race or ethnicity are used as a crude proxy for them.

Some of the disagreement between liberals and conservatives may be due to a difference of opinion on the relative efficacy of the two policies. For example, liberals may think that racially based affirmative action is effective in achieving its goals, while terrorism profiling is not; conservatives may think the opposite. However, this does not account for the large number of conservatives who oppose affirmative action because they think it is intrinsically wrong, regardless of its effectiveness. And ditto for the large number of liberals who oppose ethnic profiling for national security purposes irrespective of how effective it might be. There are several possible ways to distinguish between security profiling and affirmative action. What is striking to me, however, is that most liberals and conservatives seem to completely ignore the potential contradiction between their thinking on these two issues.

UPDATE: I am embarrassed to say that I was unaware of my colleague Nelson Lund's paper on this subject until it was pointed out by Jonathan Adler in his post responding to mine. Nelson's article does precisely what I urge other conservatives to do in my post: it considers the implications of the conservative critique of affirmative action for the conservative defense of ethnic profiling for national security purposes. However, Nelson wrote the article in part precisely because most other conservatives have simply ignored the tension between their positions on these two issues. It's my impression that most continue to do so, despite Nelson's well-taken admonitions. The same, of course, is true of most liberals.