•  
  •  
 

Abstract

Many online review systems adopt a voluntary voting mechanism to identify helpful reviews to support consumer purchase decisions. While several studies have looked at what makes an online review helpful (review helpfulness), little is known on what makes an online review receive votes (review voting). Drawing on information processing theories and the related literature, we investigated the effects of a select set of review characteristics, including review length and readability, review valence, review extremity, and reviewer credibility on two outcomes—review voting and review helpfulness. We examined and analyzed a large set of review data from Amazon with the sample selection model. Our results indicate that there are systematic differences between voted and non-voted reviews, suggesting that helpful reviews with certain characteristics are more likely to be observed and identified in an online review system than reviews without the characteristics. Furthermore, when review characteristics had opposite effects on the two outcomes (i.e. review voting and review helpfulness), ignoring the selection effects due to review voting would result in the effects on review helpfulness being over-estimated, which increases the risk of committing a type I error. Even when the effects on the two outcomes are in the same direction, ignoring the selection effects due to review voting would increase the risk of committing type II error that cannot be mitigated with a larger sample. We discuss the implications of the findings on research and practice.

Share

COinS