A study of college graduates found that economics majors had higher average starting salaries than art history majors, suggesting that the study of economics increases your chances of getting a good job. Identify the most important statistical bias in this sample and describe a (hypothetical) controlled experiment that would get rid of this problem.