Explaining ChatGPT Plus users’ continuance intention: comparing TAM, ECM, and extended ECM

Abstract

This study aims to compare the explanatory power of three influential theoretical frameworks in post-adoption research for understanding users’ continuance intention toward ChatGPT Plus. The three models are the Technology Acceptance Model (TAM), the Expectation Confirmation Model (ECM) and an extended ECM (EECM) that integrates core belief and evaluative constructs from both TAM and ECM. Data were collected from 333 active ChatGPT Plus subscribers and analyzed using structural equation modeling. Three separate models were specified and assessed using IBM SPSS and AMOS 31. Their performance was compared based on model fit indices and predictive validity. Each model’s explanatory power was evaluated through R² values for continuance intention. All three models showed acceptable goodness-of-fit. EECM showed the highest explained variance, while TAM showed the strongest parsimony based on information criteria. In terms of predictive validity, the EECM accounted for the largest proportion of variance in continuance intention at 69.50%, exceeding the explanatory levels of TAM (50.40%) and ECM (69.30%). These results show that while TAM remains a strong and parsimonious predictor of user continuance, the EECM provides the most comprehensive explanation of ChatGPT Plus post-adoption behavior, and ECM performs adequately but less effectively relative to the other two models. To the best of the author’s knowledge, this study is one of the earliest comparative examinations of TAM, ECM, and EECM in the context of GenAI subscription services. By empirically demonstrating the added value of combining pre-adoption beliefs with post-adoption evaluations, the findings of this study contribute to theoretical work on technology continuance and offer practical insights for improving user retention in AI-driven subscription services.