Beyond stars: role of discrete emotions on online consumer review helpfulness

Traditional Word-of-Mouth (WOM) literature shows that emotions embedded in advertising appeals or referrals/reviews influence consumer buying journey. However, there is a paucity of research exploring the distribution and impact of emotional content in “online” consumer reviews (OCRs). Hence, in this study, an attempt is made to examine the emotional content in OCRs and to study the influence of discrete positive and negative emotions on the perceived OCR helpfulness, by classifying them based on their valence and arousal. Further, the impact of the inconsistency between the star rating of a review and the qualitative review-related factors (i.e. emotions and valence) is analyzed. This study uses Natural Language Processing (NLP) and text-mining techniques to retrieve valence and emotions from 100,000 reviews from Yelp.com and employs model testing methods to verify the hypotheses. The results reveal that: (1) both qualitative/latent (emotions and valence) and quantitative/manifest review message factors (star ratings, word count) are important in determining the OCR helpfulness; (2) the difference in arousal (high/low) and valence (positive/negative) has a differential impact on OCR helpfulness; (3) within high-arousal emotions, negative-valence emotions influence OCR helpfulness more than positive-valence emotions; and (4) consumers’ perceptions of OCR helpfulness depends on the consistency between the qualitative and quantitative content factors.