Identifying drivers of evaluation bias in online reviews of city destinations

Abstract

Online reviews are a significant part of decision-making processes of customers and organizations. Varying user rating behaviors and lack of consistency between ratings and expressed user opinion in reviews have been a subject of recent studies. This paper provides insights into forms of bias affecting users` evaluations in urban destination reviews. Our opinion mining approach identifies context-specific, intrinsic terms associated with rating levels. We assess the individual impact of negativity, positivity and user characteristics on ratings using regression and classification methods, to verify two main sources of biased rating behavior. First, opinion-related evaluation asymmetries, namely positivity bias, negativity bias and bias related to positivity of low arousal. Second, bias associated with users` past rating behavior. Building upon results, we distinguish five user clusters with distinct evaluation bias characteristics. Results enhance understanding of users` formation of rating choices, while our simplified segmentation approach can be useful for relevant user profiling practices.