We use a sequential prisoner's dilemma game to measure the other-regarding behavior in samples from three related populations in the upper Midwest of the United States: 100 college students, 94 non-student adults from the community surrounding the college and 1,069 adult trainee truckers in a residential training program. Both of the first two groups were recruited according to procedures commonly used in experimental economics (i.e., via e-mail and bulletin-board advertisements) and therefore subjects self-selected into the experiment. Because the structure of their training program reduced the opportunity cost of participating dramatically, 91% of the solicited trainees participated in the third group, so there was little scope for self-selection in this sample. We find no differences in the elicited other-regarding preferences between the self-selected adults and the adult trainees, suggesting that selection into this type of experiment is unlikely to bias inferences with respect to non-student adult subjects. We also test (and reject) the more specific hypothesis that approval-seeking subjects are the ones most likely to select into experiments. At the same time, we find a large difference between the self-selected students and the self-selected adults from the surrounding community: the students appear considerably less pro-social. Regression results controlling for demographic factors confirm these basic findings.