The average score in the Die Under Cup task was 3.63 (SD = 1.67) and the distribution of responses deviated from the theoretical symmetric (honest) distribution, χ2 = 23.73, p < .001, indicating that in general, some cheating took place. As expected, the average die roll score was positively correlated with the lying frequency questionnaire, (r = .39, p < .01). The more people self-reported lying, the more they earned in the Die Under Cup task.
We further classified subjects as Honest and Dishonest according to the likelihood that their reported scores were honest. Taking an alpha of 10%, 15 subjects (29% of the sample) showed a score which was higher than the score predicted by chance. A one way ANOVA was used to compare the scores of the two groups (honest [n = 36] and dishonest [n = 15] subjects). For the lying frequency questionnaire and the YPI total score, results revealed a significant difference between males and females, with males scoring higher on both measurements. Hence, gender was entered as a covariate.
A significant difference was found for the lying frequency score, F(1, 49) = 9.45, p < .005, η2 = 0.14, with dishonest subjects scoring higher (M = 3.93, SD = 3.49) than honest subjects (M = 1.58, SD = 1.93). A significant difference was also found for the confession question, with dishonest subjects scoring higher (M = 7.00, SD = 11.92) than honest subjects (M = 0.69, SD = 1.75), F(1, 49) = 9.83, p < .005, η2 = 0.17. A nonsignificant difference in the expected direction was found for the FT score, with dishonest subjects scoring lower (M = 5.61, SD = 2.30) than honest subjects (M = 6.74, SD = 2.10), F(1, 49) = 2.85, p = .09, η2 = 0.06. No difference was found for the Words score, F(1, 49) = 2.68, p = .11, YPI total score, F(1, 49) = 1.07, p = .31, YPI LIE score, F(1, 49) = 0.54, p = .46 or DIT score, F(1, 49) = 0.64, p = .42.