An Explanatory-Sequential Analysis Of Students' Attitudes Towards Chatgpt, Epistemic Beliefs, And Academic Honesty For Segmenting AI Use

Abstract

This study employs an explanatory-sequential mixed-methods design to explore the relationship

between college students’ attitudes toward ChatGPT, epistemological beliefs, and tendencies toward internet-triggered academic dishonesty. Grounded in the Theory of Planned Behavior, the study integrates quantitative and qualitative methods to provide a understanding of how students perceive and use ChatGPT in academic contexts. Quantitative data were gathered from 279 students through web-based survey followed by a k-means cluster analysis to segment students into profiles. Qualitative data from focus group discussions enriched these findings, generating themes explaining subjective norms and behavioral control. The cluster analysis revealed two distinct profiles: Cluster 1, characterized by mature epistemological beliefs, below average attitude towards ChatGPT, and minimal academic dishonesty, and Cluster 2, defined by pragmatic attitudes of ChatGPT, and higher dishonesty tendencies, and naïve epistemological belief. Mediation analysis found no significant indirect effect of epistemological beliefs between attitudes toward ChatGPT and academic dishonesty, although epistemological beliefs independently discouraged dishonest behaviors. Integration of quantitative and qualitative findings through cross-case quantitization linked clusters with themes such as academic pressures and strategies, ethical concerns, and technological accessibility which highlighted varied motivations and contextual influences. Behavioral controls also differed between clusters. Cluster 1 exhibited balanced control through themes like knowledge and inspiration and collaborative roles, while Cluster 2 predominantly relied on utilitarian usage. The study's findings underscore the tension between students’ ethical considerations and their pragmatic reliance on AI tools influenced by academic demands and social norms. These findings hold implications for policy and instructional strategies aimed at balancing AI’s academic potential with its ethical challenges.