Date of Degree


Document Type


Degree Name





Maria Hartwig

Committee Members

Charles Stone

Mark Fondacaro

Michael Leippe

Charles Bond, Jr.

Timothy Luke

Subject Categories

Law and Psychology | Other Psychology | Social Psychology


Cheating, Dissonance, Moral Disengagement, Self-Concept Maintenance, Ethical Decision-Making, Legal Decision-Making


For the past fifteen years, the Russano et al. (2005) cheating paradigm has dominated research in the forensic psychological literature. While this paradigm successfully activates theoretical mechanisms for ethical decision-making, applying the methods for online data collection is cumbersome and retains a confound inherent in the design. Alternative cheating paradigms from both the psychology and economics literatures were evaluated for their suitability for an online cheating paradigm. The impossible anagram task was selected as most likely to elicit the same internal and external cost-benefit analyses online as the Russano et al. (2005) cheating paradigm does in-person: self-concept maintenance, ethical dissonance, and moral disengagement. Three studies developed the online impossible anagram task and tested the task as a guilt manipulation in a larger experimental design. First, Study 1 developed the stimuli for the anagram task by evaluating the difficulty of 149 anagrams in the target population. Second, Study 2 examined cheating behavior on different versions the task with the goal of randomly assigning participants to guilt condition. I also hypothesized cheaters would experience changes in self-concept, ethical dissonance, and moral disengagement. Finally, Study 3 provided participants an opportunity to confess to cheating on the anagram task to examine the influence of morality inductions on truthful disclosure.

Study 1 successfully identified several “impossible” anagrams participants in the target population were unable to correctly unscramble. In Study 2, these “impossible” anagrams were used to classify cheating during the anagram task. Unfortunately, the rates of cheating among the different versions of the task did not meet a priori thresholds for random assignment to guilt condition. However, in one version of the task, participants cheated approximately 50% of the time, allowing comparisons between those who self-selected their guilt status. Results from Studies 2 and 3 partially supported the hypotheses regarding changes in ethical dissonance and moral disengagement as a result of cheating on the anagram task but not self-concept. The morality inductions included in Study 3 moderated some participants’ reports of ethical dissonance and moral disengagement, partially supporting the hypotheses. Exploratory analyses suggest there may be fundamental personality differences between participants who cheated to the fullest extent and those who abstained from cheating. Implications of the results are discussed and recommendations made for further refinement of the impossible anagram task for use as a guilt manipulation in online data collection.