Sequential and Temporal Dynamics of Online Opinion

Abstract

We investigate the evolution of online ratings over time and sequence. We first establish that there exist two distinct dynamic processes, one as a function of the amount of time a book has been available for review and another as a function of the sequence of reviews themselves. We find that, once we control for calendar date, the residual average temporal pattern is increasing. This is counter to existing findings that suggest that without this calendar-date control, the pattern is decreasing. With respect to sequential dynamics, we find that ratings decrease: the nth rating is, on average, lower than the n-1th when controlling for time, reviewer effects, and book effects. We test and find some support for existing theories for this decline based on motivation. We then offer two additional explanations for this “order effect.” We find support for the idea that one's ability to assess the diagnosticity of previous reviews decreases: when previous reviewers are very different, more reviews may thus lead to more purchase errors and lower ratings.

Management science in the era of smart consumer products: challenges and research perspectives

“…Recent studies on star ratings using ordered logit or ordered probit have been published by Godes and Silva (2012) and Moe and Schweidel (2012). With dependent variables y l denoting the ordinal-scaled star rating observed with online review l, and x pro l (x con l ) denoting the binary vector that represents the corresponding occurrences of the smart product attributes of interest in the pros (cons) list of review l, the linear relationship y l = αx pro l + βx con l + ε∀l can be formulated.…”

Noise or Quality? Cross-Nested Hierarchical Effects of Culture on Online Ratings

“…Much of the literature specifically devoted to online feedback systems in a variety of virtual contexts has focused on explaining the impact that ratings have on promoting a variety of outcomes such as price premiums and trust in electronic commerce environments (Ba & Pavlou, 2002;Chevalier & Mayzlin, 2006;Clemons, Gao, & Hitt, 2006;Dellarocas, 2006;Pavlou & Dimoka, 2006;Son et al, 2006), consumer purchase errors (Godes & Silva, 2012), decision quality and time (Poston & Speier, 2005), and effort to contribute content in user-generated content websites (Chen et al, 2011). Unfortunately, this prior literature has reported mixed empirical results whereby sometimes the impact of ratings is positive, negative, or not significant (Ba & Pavlou, 2002;Ghose et al, 2006;Kauffman & Wood, 2005).…”

“…Much of the literature on online feedback systems in a variety of contexts has focused on explaining the impact that ratings in these systems have on a variety of social and economic outcomes (Ba & Pavlou, 2002;Chen et al, 2011;Chevalier & Mayzlin, 2006;Dellarocas, 2006;Godes & Silva, 2012;Pavlou & Dimoka, 2006;Poston & Speier, 2005;Son et al, 2006). Far less literature has investigated why something gets positively or negatively rated online.…”

“…Much of the literature on online feedback systems has focused on explaining the impact that ratings (as proxies for quality) have on a variety of social and economic outcomes (Ba & Pavlou, 2002;Chen, Xu, & Whinston, 2011;Chevalier & Mayzlin, 2006;Dellarocas, 2006;Godes & Silva, 2012;Pavlou & Dimoka, 2006;Poston & Speier, 2005;Son, Tu, & Benbasat, 2006). Unfortunately, this prior literature has reported mixed results whereby sometimes the impact of ratings is positive, negative, or not significant (Ba & Pavlou, 2002;Ghose, Smith, & Telang, 2006;Kauffman & Wood, 2005).…”

The Logic of Collective Rating

“…Closest to our study is the recent social-influence-in-rating literature [27][28][29][30] which we synthesize in the formulation of our model, relying on a rich model with four behavioral types. The model explains empirical regularities with a single model, particularly contrarian dynamics regarding initial-rating vs. end-of-sequence-rating dynamics.…”