#36 - Gender Bias in Online Reviews - Starring Julia Barnett

Coffee Talk Episode 35 focuses on some illuminating research on gender bias. Margrét is joined by Julia Barnett, a PhD candidate at Northwestern University. Julia's research interests lie in algorithmic ethics and transparency, reducing the socio-technical harms of algorithmic systems, and deep generative applications in social contexts. Her recent study shows how patients perceive doctors differently based on the doctor’s gender. She examined over 400,000 online reviews of physicians.

She found a stark contrast in the words patients use to describe male and female doctors. Male doctors with a five-star rating are more likely to be called “brilliant,” “excellent,” or “superb.” Meanwhile, five-star female doctors are more likely to be called “nice.” So even when patients rate their female doctors highly, their language indicates a lower level of regard.

Julia also looked at doctors that provide lower-quality care. (These doctors had been sanctioned by their state medical boards for serious quality issues.) She found that lower-quality male doctors still garner positive reviews from patients. Meanwhile, patients use strong negative language to describe lower-quality female doctors.

So online reviews are biased, even for low-quality doctors. This finding has HR implications: many patients choose their doctors based on reviews, and reviews may affect performance evaluations. So it’s crucial for HR teams to consider how female and male professionals of the same quality are evaluated differently by those who receive their services.

Julia’s findings are also worth consideration by anyone writing personnel evaluations. Knowing that people tend to use different sets of words for men and women is an invitation for all of us to think about why we use the words we do.

Friday Coffee Talk from Planet Fair is a YouTube series co-hosted by PayAnalytics founder Margrét Bjarnadóttir and Henrike Von Platen, founder and CEO of the FPI Fair Pay Innovation Lab in Berlin.