A study found that professors tend to give higher grades to students who take more than one course from the professor than to students who take only one course from the professor, suggesting that professors reward students who come back for more. Identify the most important statistical bias in this sample and describe a (hypothetical) controlled experiment that would get rid of this problem.