My previous columns have attempted to reduce confusion by suggesting operational definitions of discrimination and prejudice. Discrimination was defined as the act of choice, and prejudice was the act of decision-making on the basis of incomplete information. Good analytical thinking requires that we don't confuse one phenomenon with another.

The final behavioral phenomenon related to discussions of race is racial preference. We can think of preferences generally as likes and dislikes, and we all have them for many things. Some of us prefer Bordeaux wines to California wines, while others prefer the opposite. Some of us prefer jazz music while others prefer classical music. The list of differences in human preferences is endless.

There's no logically consistent argument that says to prefer one good, service or person is better, or more righteous, than another. Let's try it. Is my preference for California wines better, or more righteous, than your preference for Bordeaux? Is your preference to marry a white woman better, or more righteous, than my preference to marry a black woman? While we might like or dislike another's preferences, there are no analytical standards by which we can judge one set of preferences to be superior to another.

Preferences alone do not determine behavior. If we conducted a survey asking people which they prefer: filet mignon or chuck steak, Rolex watches or Timex, Rolls Royces or Dodge Neons, I'm guessing that filet mignon, Rolex and Rolls Royce would win hands down. Having found what people preferred the most, then watch what they actually do. You would find chuck steak outselling filet mignon, Timex watches outselling Rolex, and Dodge Neons outselling Rolls Royces any day of the week.

To fully understand behavior, we must go beyond preferences and take restrictions on choice into account, namely income and prices. That fact is very relevant to issues of race. Let's look at it. During South Africa's apartheid era, white labor unions that would never have a black as a member were the major supporters of minimum wages for blacks. Their stated intention was to protect white workers from competition with low-wage black workers.

Gert Beetge, secretary of the Building Workers' Union, said, "There's no job reservation left in the building industry, and in the circumstances, I support the rate for the job (minimum wages) as the second best way of protecting our white artisans."

In the United States, the Davis-Bacon Act of 1931 (still on the books), a super minimum wage law, was enacted to protect unionized white construction workers from competition with black workers. The support ran along the lines of Alabama Rep. Clayton Allgood's testimony: "That contractor has cheap colored labor that he transports, and he puts them in cabins, and it is labor of that sort that is in competition with white labor throughout the country" (Congressional Record, 1931, page 6513).

What minimum wage laws do is lower the cost of, and hence subsidize, racial preference indulgence. After all, if an employer must pay the same wage no matter whom he hires, the cost of discriminating in favor of the people he prefers is cheaper. This is a general principle. If filet mignon sold for $9 a pound and chuck steak $4, the cost of discriminating in favor of filet mignon is $5 a pound, the price difference. But if a law mandating a minimum price for chuck steak were on the books, say, $7 a pound, it would lower the cost of discrimination against chuck steak.

Minimum or maximum prices are one of the most effective ways to encourage people to indulge their preferences, be they racial or any other preference. In general, any kind of economic regulation that restricts peaceable, voluntary exchange has the capacity to lower the costs of preference indulgence. Decent people should be against such regulations.

Walter E. Williams is a professor of economics at George Mason University.