The present dataset was reported in a paper entitled “Effects of test difficulty messaging on academic cheating among middle school children” [1]. It reports the findings of an experimental study that used a naturalistic math test-taking paradigm to assess children's academic cheating behavior under different test difficulty messaging conditions. The participants were Grade 8 middle school children (N = 201). The primary dependent measures were whether each participant spontaneously decided to cheat (presence of cheating), and among participants who cheated, the specific number of test items on which they cheated (extent of cheating). We used logistic regression, ANOVA, and Pearson correlation to assess whether various predictor variables (e.g., conditions) predicted the presence of cheating or the extent of cheating. This dataset should be of interest to researchers who are interested in the development of moral behavior in children generally, and academic dishonesty in particular.
