Self-selection bias in a field experiment: Recruiting subjects under different payment schemes

We examine a potential self-selection bias in different samples of experimental subjects depending on the payment scheme offered in the recruiting process. We ran four field experiments in which undergraduate students in a microeconomics course were invited to voluntarily set their own goal for the final exam. They were informed that they would be given a monetary reward (else nothing) if their actual grade were higher than or equal to their goal. Rewards were an increasing (quadratic) function of the goal. We aimed at studying whether subjects' willingness to participate in the experiment depends on their expected performance under different advertised reward criteria, like a rank-order tournament and piece-rate pay. Given that judgments about future performance are closely tied to previous performance, the midterm exam scores from the current academic course are compared between participants and nonparticipants in order to analyze sample-sorting effects. We find that when a rank-order tournament is offered alone or in combination with another reward mechanism, high-performing students are more likely than low-performing ones to participate in the experiment.