JMIR Publications

ABSTRACT

Background:

Online doctor reviews are an important information source for prospective patients. In addition, they represent an untapped resource for studying gender bias in the doctor patient relationship. Understanding bias is important because it may impact the value of online reviews to patients. Documenting bias is also important for improving the doctor patient relationship. Yet studies of bias are currently limited to smaller data sets and specific conditions.

Objective:

In this research, we explore 154,305 reviews from across the U.S. for all doctor specialties. Our analysis includes a qualitative and quantitative examination of review content and rating as they relate to doctor and reviewer gender.

Methods:

154,305 reviews were downloaded from Google. For each review, the reviewer and doctor were assigned a gender based on name. Reviews were also coded by star rating (low or high) and content category (process, positive soft skills, negative soft skills). Our quantitative analysis made use of regression to model the relationships between the variables of interest (process, soft skills, physician gender, and reviewer gender).

Results:

Reviews for female physicians were overall significantly more critical (3.4 average rating for female and 4.0 average rating for male physicians). Reviews for female physicians were significantly more critical with respect to both process and soft skills, with negative comments on soft skills being twice as common for female (16.02%, 5,903 of 36,847) as for male physicians (8.80%, 6,529 of 74,189). Reviewer gender was not a strong predictor of review outcome or content, in general (p=0.046). The largest difference is in mentions of negative soft skills; 8.81% of female reviewers (5,978 of 67,857) and 7.79% of male reviewers (3,364 of 43,179). However there is a significant interaction effect between reviewer and physician gender. Female reviewers rate male physicians positively around 75.90% of the time (33,466 of 44,094), while male reviewers rate female physicians positively around 58.19% of the time (7,362 of 12,652). Our approach cannot identify causal relationships.

Conclusions:

Reviews for female physicians were more critical and focused more on soft skills than reviews for male physicians. Additionally, female reviewers are more likely to report on positive interpersonal skills than are male reviewers, and more likely to report experiences of disrespect and indifference in negative reviews than are male reviewers. This work complements existing smaller-scale studies of bias and suggest a need to include information about gender differences in review site summaries and search results. Clinical Trial: N/A