When Do Consumers Value Positive vs. Negative Reviews? An Empirical Investigation of Confirmation Bias in Online Word of Mouth

In the online word-of-mouth literature, research has consistently shown that negative reviews have a greater impact on product sales than positive reviews. While this negativity effect is well documented at the product level, there is less consensus on whether negative or positive reviews are perceived to be more helpful by consumers. A limited number of studies document a higher perceived helpfulness for negative reviews under certain conditions, but accumulating empirical evidence suggest the opposite. To reconcile these contradictory findings, we propose that consumers can form initial beliefs about a product on the basis of the product’s summary rating statistics (such as the average and dispersion of the product’s ratings), and that these initial beliefs play a vital role in their subsequent evaluation of individual reviews. Using a unique panel data set collected from Apple’s App Store, we empirically demonstrate confirmation bias – that consumers have a tendency to perceive reviews that confirm (vs. dis-confirm) their initial beliefs as more helpful, and that this tendency is moderated by their confidence in their initial beliefs. Further, we show that confirmation bias can lead to greater perceived helpfulness for positive reviews (positivity effect) when the average product rating is high, and for negative reviews (negativity effect) when the average product rating is low. Thus, the mixed findings in the literature can be a consequence of confirmation bias. This paper is among the first to incorporate the important role of consumers’ initial beliefs and confidence in such beliefs (a fundamental dimension of metacognition) into their evaluation of online reviews, and our findings have significant implications for researchers, retailers and review websites.