| This dissertation addresses technology acceptance issues, specifically for website portals, and investigates the suitability of the Technology Acceptance Model (TAM) and assesses the causal structures implied by it using Bayesian Networks (BNs). The Bayesian methodology is applied to data collected from an existing website portal. The website portal data was used in a study by Hans van der Heijden (2002) and was analyzed, in the context of TAM, using multiple regression analyses.; Constructs used in the original TAM are further modeled as a simplified longitudinal Bayesian model. This longitudinal Bayesian model is tested with a set of computer simulated data to formalize the learning of causal associations in a temporal set of observations, and illustrates how such an approach can be used to model cross-sectional, non-experimental data that is common in the social sciences.; It is found that such an analysis of a causal model using Bayesian techniques will provide for possible improved prediction capabilities of user behavior and user satisfaction on the Internet. This research adds to the very limited existing research on the applicability of Bayesian networks to TAM and in broader terms to IS research. Furthermore, this research investigated whether an extension of TAM is more appropriate in the context of Web portals. |
