Our experiments support the widely documented poor ability of humans to detect lies holds for both self-selected and instructed liars

Human Lie-Detection Performance: Does Random Assignment versus Self-Selection of Liars and Truth-Tellers Matter? Karl Ask, Sofia Calderon, Erik Mac Giolla. Journal of Applied Research in Memory and Cognition, December 25 2019. https://doi.org/10.1016/j.jarmac.2019.10.002

Deception research has been criticized for its common practice of randomly allocating senders to truth-telling and lying conditions. In this study, we directly compared receivers’ lie-detection accuracy when judging randomly assigned versus self-selected truth-tellers and liars. In a trust-game setting, senders were instructed to lie or tell the truth (random assignment; n = 16) or were allowed to choose to lie or tell the truth of their own accord (self-selection; n = 16). In a sample of receivers (N = 200), we tested two alternative hypotheses, predicting opposite effects of random assignment (vs. self-selection) on receivers’ lie-detection accuracy. Accuracy rates did not differ significantly as a function of veracity assignment, failing to support the claim that random assignment of liars and truth-tellers alters the detectability of deception. Equivalence tests indicated that, while a small effect of random assignment cannot be ruled out, moderate (or larger) effect sizes are unlikely.

Keywords: DeceptionLie detectionRandom assignmentSelf-selectionDetection strategy

General Audience Summary: In everyday communication, people typically decide whether to lie or to tell the truth of their own accord. In most studies on lie detection, however, researchers instruct individuals to lie or tell the truth on a random basis. This approach has received critique from experts in the field, because it does not reflect what happens in real life. Since self-selected and instructed liars and truth-tellers differ in several ways (e.g., motivation, proficiency of lying), the two modes of veracity assignment may give rise to different cues to truth and deception. In the current study, we tested whether random assignment, as compared with self-selection, improves or impairs people's ability to detect deception. Liars and truth-tellers (senders) tried to convince participants (receivers) to trust them with their money, promising cooperation and financial gain in return. Half of the senders had been randomly assigned to lie or tell the truth, whereas the other half had chosen to lie or tell the truth of their own accord. We tested two competing hypotheses: First, on the assumption that it prevents good liars from choosing to lie (and poor liars from choosing not to lie), random assignment would improve receivers’ ability to detect lies. Second, on the assumption that there are detectable differences between senders who are likely to lie when given the opportunity and those unlikely to lie, random assignment would make such differences uninformative and impair receivers’ ability to detect lies. Our results did not support any of the hypotheses (lie-detection accuracy was near chance level in all experimental conditions), thus failing to support the claim that random assignment of liars and truth-tellers alters the detectability of deception. Instead, they indicate that the widely documented poor ability of humans to detect lies holds for both self-selected and instructed liars.