Of the 251 children who peeked, 167 (66.5%) lied about peeking at the toy when later asked. All of
the non-peekers truthfully denied peeking. Logistic regression analyses with peekers who lied (coded
as 1) or confessed (coded as 0) as the predicted variable revealed that age, appeal, and the
Appeal Punishment interaction were significant predictors (see Table 2). Coefficients for the logistic
regression are displayed in Table 2. With every 1-month increase in age, children were 1.015 times
less likely to lie. Significantly more children lied in the No Appeal conditions (87.1% ps < .001) compared
with the External Appeal (46.4%) and Internal Appeal (65.9%) conditions. However, this was
qualified by a significant Appeal Punishment interaction. Further inspection of the Internal Appeal
conditions revealed that there was a significant difference between the Punishment and No Punishment
conditions (b = 2.14, SE = 0.86, p = .013, odds ratio = .118) (see Fig. 1). Specifically, whereas
the majority of children in the Punishment–Internal Appeal condition lied (86%), significantly fewer
children in the No Punishment–Internal Appeal condition lied (45%). Although there was a trend for
fewer children to lie in the Punishment–External Appeal condition (35%) than in the No Punishment–
External Appeal condition (61%), it was not significant (p = .144). There were no other significant
differences.