Consumer Reviews and Actual Purchase Behavior in E-Commerce: Linear and Nonlinear Effects

Abstract

The rapid expansion of e-commerce has intensified the significance of online reviews as key informational cues that help mitigate uncertainty in digital transactions. Although prior research has widely explored the role of reviews, relatively few studies have addressed how their quantitative attributes function across different product contexts. This study investigates the impact of positive and negative reviews on consumer purchase behavior, emphasizing a comparison between absolute review counts and proportional measures, and assessing whether these effects differ for search versus experience goods. Drawing on purchase data from a leading South Korean e-commerce platform, both linear and threshold regression models were applied to capture potential nonlinearities. The findings reveal that negative reviews exert a considerably stronger influence on purchase behavior than positive reviews, consistent with Prospect Theory and the concept of negativity bias. Furthermore, absolute review volume showed significant effects, whereas proportional indicators were not statistically meaningful, suggesting that consumers react more strongly to the number of reviews rather than their relative distribution. Product category also moderated these relationships, with review counts greater impact for search goods than for experience goods. Threshold regression highlighted nonlinear patterns, such as diminishing returns from positive reviews and heightened deterrent effects of negative reviews when their volume is low. Overall, this study extends electronic word-of-mouth research by integrating volume, asymmetry, and product heterogeneity into a unified framework, while offering practical guidance for improving review management and consumer trust in digital marketplaces.