What Drives Extremity Bias in Online Reviews? Theory and Experimental Evidence
50 Pages Posted: 6 Sep 2019 Last revised: 11 Sep 2019
Date Written: September 10, 2019
In a range of studies across platforms, online ratings have been shown to be characterized by distributions with disproportionately-heavy tails. We focus on understanding the underlying process that yields such “j-shaped” or “extreme” distributions. We develop a simple analytical model to capture the most-common explanations: differences in utility or differences in base rates associated with posting extreme versus moderate reviews. We compare the predictions of these explanations with those of an alternative theory based on differential rates of attrition from the potential reviewer pool across people with moderate versus extreme experiences. The attrition rate, by assumption, is higher for moderate reviews. The three models yield starkly different predictions with respect to the impact on the relative prevalence of extreme versus moderate reviews of a review solicitation email: while existing theories predict a relative increase in extreme reviews, our attrition-based model predicts a decrease. Our results from a large-scale field experiment with an online travel platform clearly support the predictions from the attrition-based explanation, but are inconsistent with those from the utility-based and base-rate explanations alone.
Keywords: online reviews, online word of mouth, field experiment
Suggested Citation: Suggested Citation