150 Chapter 5 preferred their airline to another airline [5]. The target population was frequent business travelers from New York to Chicago. The puzzling thing about this ad was that, in practice, only 8 percent of the people flying from New York to Chicago flew on this airline. If 84 percent prefer this airline, why were only 8 percent flying on it? The answer to this puzzle is that the sample had considerable selection bias in that it was not randomly drawn from the target population but, instead, was a survey of passengers on one of this airline's flights from New York to Chicago. It is not surprising that frequent business travelers who choose to fly this airline prefer it. It is surprising that 16 percent of the passengers prefer another airline. But, we can hardly expect this airline to advertise "Sixteen percent of the people flying on our planes wish they weren't." This particular kind of selection bias is known as self-selection bias because people choose to be in the sample. There is also self-selection bias when people choose to go to college, marry, have children, or immigrate to another country. In each case, we should be careful making comparisons to people who made different choices. For example, college graduates may have higher incomes, in part, because they are more ambitious. If so, it would be a mistake to attribute their higher income solely to college attendance. For a fair comparison, we would have to do something like the following: among all those who want to go to college, a random sample is allowed to go.