JMIR Publications

ABSTRACT

Background:

Previously, most studies use 5-star and 1-star ratings to represent reviewers' positive and negative attitudes respectively. However, this premise is not always true because individuals’ attitudes have more than one dimension. In particular, as a credence product, to build durable doctor-patient relationships, patients may rate their doctors with high scores to avoid lowering doctors’ online ratings and to help build doctors’ online reputation. Some patients may only express some complaints in review texts, resulting in ambivalence such as conflicting feelings, beliefs, and reactions towards doctors. Thus, online rating platforms for medical services may face more ambivalence than platforms for search or experience goods.

Objective:

Based on the tripartite model of attitudes and uncertainty reduction theory, this study aims to consider both the numerical rating and sentiment of each online review to explore whether there is ambivalence and how the ambivalent attitudes influence the helpfulness of online reviews.

Methods:

This study collected 114378 reviews for 3906 doctors on a large doctor review website. Then, based on existing literature, we operationalized the numerical rating as the cognitive dimension of the attitude and sentiment in review texts as the affective dimension of the attitude. Several econometric models, including the ordinary least squares model, the logistic regression model, and the Tobit model, were employed to test our research model.

Results:

First, this study confirms the existence of ambivalence in each online review. Then, by measuring the ambivalence through the inconsistency between the numerical rating and sentiment for each review, this study found that the ambivalence in each online review has a different impact on the helpfulness of online reviews. Specifically, for reviews with positive emotional valence, the higher the degree of inconsistency between the numerical rating and sentiment, the more helpfulness (β_1^Positive=0.046,p<.001). For reviews with negative and neutral emotion valence reviews, the impact is opposite, and the higher the degree of inconsistency between the numerical rating and sentiment, the less helpfulness (β_1^Positive=0.046,p<.001; β_1^Neutral=-0.030,p=.233). Considering the traits of the data, the results are also verified using the logistic regression model (θ_1^Positive=0.056,p=.005; θ_1^Negative=-0.080,p<.001; θ_1^Neutral=-0.060,p=-.034) and Tobit model.

Conclusions:

This study confirms the existence of ambivalence between the two dimensions in single reviews and finds for reviews with positive emotional valence, the ambivalent attitudes in each online review will lead to more helpfulness, but the ambivalence leads to less helpfulness for negative and neutral emotion valence reviews. The results contribute to the online review literature and inspire a better design for online review website rating mechanisms to enhance the helpfulness of reviews.