The impact of construal level on review consistency and helpfulness in online evaluations

Abstract

This study investigates the antecedents and consequences of internal evaluative consistency between verbal and numerical evaluations in online reviews. As an antecedent, we argue that the review's construal level affects its consistency, with abstract reviews being more internally consistent than concrete ones. As for consequences, we argue that internally consistent reviews are perceived as more helpful and useful. Empirically, we examine reviews from two major online review websites, Amazon and Yelp. To assess the internal evaluative consistency of reviews, we build a deep learning framework that analyzes review texts and predicts the “correct” rating and compares this predicted rating to the actual rating. We find confirmation for our predictions. Finally, we consider the implications of our findings for both theory and practice in the context of online reviews.