A fortnight ago, my column made a stab at applying dispassionate analysis to come up with an operational definition for discrimination. Basically, discrimination is the act of choice, and choice is a necessary fact of life. Now let's turn to prejudice, keeping in mind that for sound thinking, one should avoid confusing one phenomenon with another.

Prejudice is a useful term that's often misused. Its Latin root is praejudicium, meaning "an opinion or judgment formed . . . without due examination." Thus, we might define prejudicial acts as decision-making on the basis of incomplete information.

In a world of costly information, people seek to economize on information costs. Imagine heading off to work, you open your front door, only to be greeted by a full-grown tiger. The uninteresting prediction is the average person would slam the door or otherwise seek safety.

Why they do so is more interesting. It's unlikely that person's decision is based on any detailed information held about that particular tiger. More likely his decision is based on tiger folklore or how he's seen other tigers behave. He prejudges, or stereotypes, that tiger.

If a person didn't pre-judge tigers, he would seek more information prior to his decision. He might attempt to pet the tiger, talk to him and seek safety only if the tiger responded in a menacing fashion. The average person wouldn't choose that path, surmising that the expected cost of getting more information about the tiger is greater than the expected benefit and concluding, "All I need to know is he's a tiger, and he's probably like the rest of them." By observing this person's behavior, there's no way one can say unambiguously whether the person likes or dislikes tigers.

In the late 1990s, the Washington, D.C., taxi commissioner warned cabbies against going into low income black neighborhoods and picking up "dangerous looking" passengers whom she described as young black males dressed a certain way. A few years ago, some St. Louis, Mo., pizza deliverers were complaining about delivering pizzas to black neighborhoods. Can one say anything unambiguous about cabbies' or pizza deliverers' likes or dislikes for blacks?

In the case of the taxi commissioner's warnings, the commissioner was black and so were most of the cabbies, and 75 to 85 percent of the complaining pizza deliverers were black. Are they racists? What about Rev. Jesse Jackson who once admitted that he is often relieved when the youths he hears walking along the street behind him turn out to be white, not black? Is he a racist?

As in the tiger example, the cabbies, pizza deliverers and Jackson are pre-judging. They are using a cheaply observed physical characteristic as an information proxy for a more costly to observe characteristic. The cheap-to-observe characteristic that a person is tall, female, Asian, black or white can indicate some probability of some other more costly to observe characteristic. In the minds of cabbies, pizza deliverers and Rev. Jackson, race was associated with a higher probability of being assaulted.

No one says that all young black males, not even a majority, pose a threat, but people are assigning probabilities. Such an assignment differs little from a physician, knowing that incidences of cardiovascular diseases are 30 percent higher among blacks than whites and prostate cancer is twice as high, giving his black patients more careful screening for these two diseases. Like the cabbies, pizza deliverers and Rev. Jackson, the physician is engaging in what some have called racial profiling -- using race as an indicator of something else.

For analytical purposes, it's important to correctly identify behavior. Asserting that a particular behavior reflects racial likes and dislikes, which it could, when in fact it does not, is to mislead and confound whatever problem or issue one is addressing.