Deception detection and the role of self-selection

We consider a lie-catching experiment with 9240 judgements. A set of videotapes shows subjects participating in a tax compliance experiment. The subjects chose whether or not to misreport. Subjects knew that underreporters were chosen for an audit with some probability. An audit led to detection and to a punishment fee. This compliance framework induced only persons with high deceptive abilities to underreport and, so, caused self-selection. Among the students who judged these videos, we find that the deception detection rate was significantly below 50 percent and even lower if the self-selection pressure in the tax compliance experiment was higher. This suggests that, when subjects can choose whether to state the truth or to lie, there is a self-selection effect by which individuals with higher deceptive ability are more likely to lie.