Analyzing continuance intention of recommendation algorithms

As recommendation algorithms have been increasingly applied to content personalization services, scholars are voicing concern about the negative impacts of these algorithms, for instance filter bubbles and ideological polarization. This research attempts to analyze the various factors influencing users' continuance intention of recommendation algorithms through structural equation modeling. Based on the Expectation-Confirmation Model, this study proposes an extended framework to empirically examine the impact of confirmation, perceived usefulness, perceived enjoyment, perceived ease of use, perceived risk, and subjective norm on satisfaction and continuance intention. Results indicate that confirmation positively impacts satisfaction, perceived usefulness, and perceived enjoyment. Furthermore, all constructs had a significant effect on satisfaction as well as continuance intention. A group comparison analysis of consumers primarily using news recommendation algorithms and multimedia recommendation algorithms uncovered differences between the two groups. Managerial implications on how to retain recommendation algorithm users are suggested based on the results.