Information inconsistencies in multi-dimensional rating systems

To help consumers evaluate products that often differ along multiple dimensions, many online review platforms have started to implement multi-dimensional rating systems, wherein a user provides product ratings on multiple product dimensions, in addition to an overall product rating. In contrast with prior work that has primarily focused on the effects of overall product evaluation, we explore how dimensional ratings affect product sales from an information-inconsistency perspective. Drawing on the diagnosticity theory, we examine two types of information inconsistency in multi-dimensional rating systems: inconsistency across ratings on different dimensions of a product and inconsistency across reviewers for a product dimension. Based on panel data analysis of a unique proprietary data set combining multi-dimensional rating information on 456 car brand models with their corresponding monthly sales, we find that (1) while the highest rating among different dimensions does not have a significant impact, the lowest rating among different dimensions increases product sales; (2) the rating variance among different dimensions has an inverted U-shaped relationship with product sales; and (3) the negative impact of the variance across reviewers for a dimension is contingent on whether the dimension is a vertical or a horizontal attribute. Our findings offer important theoretical and managerial implications for a better understanding of multidimensional rating systems.