A colleague asked me whether a project he was considering doing was scientifically and ethically viable. The study was designed to replicate (and extend) a published study that had been conducted using Amazon’s Mechanical Turk. In the original study, participants received small amounts of money as an incentive for participating. My colleague did not have money to pay participants but he could offer them the chance of winning an iPod as an incentive. He recognized, however, that offering a chance to win an iPod was not the same as offering a certainty of acquiring money, and he worried that readers would insist that he had not fully replicated the earlier study. To address this concern, he suggested telling participants that they would receive money for participating “as long as they solved a problem” that they would receive after performing the other components of the experiment. The “problem” would be insolvable, so he would not be obligated to pay them money (which he didn’t have), but in the spirit of compensating participants, he planned to enter them into a lottery for the iPod after they were told that they had not solved the problem.

He reasoned that the study was ethical for two reasons. First, technically he did not lie to them when he told them that he would give them the money if they completed the problem (although he would be lying by omission become most if not all participants would assume that the problem was solvable). Second, because he intended to award someone an iPod, he did intend to compensate participants.