Abstract

Consumers rate chain restaurants of the same brand in notably divergent ways, which has crucial consequences for customers, businesses, and online platforms. Through the lenses of multi-attribute utility and hierarchical information integration theories, this study aims to identify essential aspects that influence customers’ descriptions of their dining experiences in restaurants, and then leverage sentiment scores obtained from those descriptions to explain observed rating differentials across same-brand restaurants. This novel, aspect-specific sentiment analysis method can uncover critical aspects in user-provided reviews, then determine each aspect’s sentiment score. The tests of econometric models that specify the resultant impacts on customers’ ratings of restaurants, involving samples of actual reviews and ratings from Yelp, reveal that food, price, anecdotes, service, and the brand in general, as well as user activity and popularity on the platform, help explain differential ratings of same-brand chain restaurants. Food, service, price, and anecdotes are particularly important for lower-price restaurants (e.g., fast food); food, service, and brand aspects are more essential for midscale restaurants. Predictive models that incorporate the sentiment scores for these (discovered) influential aspects provide better estimates of customers’ ratings of same-brand restaurants than do models that rely only on features directly available on the platform.

DOI

10.17705/1jais.01006

Download

Share

COinS