Weekend Reads: In Employee Reviews, Words Matter

by Kevin Schofield

This weekend's read is a report by tech company Textio on biases in job performance feedback. Specifically, the report looks at biases in the language used to provide feedback — not a surprising topic given that Textio sells a product that purports to help employees improve their on-the-job writing.

In an earlier research survey, the company looked at the quality of feedback employees receive, including the amount of feedback, the mix of complimentary and critical comments, how much of the feedback is actionable, and how often the feedback is about personality traits rather than substantive work habits. It found that women receive harsher feedback, more of it is focused on their personality, and there is less constructive or actionable feedback. Performance feedback is an equity issue: Studies have shown that people who receive less feedback — and less actionable feedback — as a result have fewer opportunities for professional growth and promotion.

In the first part of this new study, the company broadened the questions to look at potential biases by age and by race/ethnicity in addition to age. And it found plenty. Of 500 people surveyed, 100% of women reported receiving personality-related feedback, but only 75% of men did. Black and Latino employees also reported receiving personality feedback, while only 71% of Asian employees did. The specific language used in describing employees' personalities also showed clear biases.

Since personality feedback is by definition less actionable, the amount that performance reviews focus on these aspects can make a huge impact on the overall quality of feedback. They also reinforce cultural biases in the workplace; the report has an interesting side discussion of "stereotype threat," in which many employees constantly spend more time questioning their own workplace actions for fear that they are providing ammunition to those looking for confirmation of well-trodden but unfounded stereotypes.

In the second half of the study, the company obtained access to the written performance reviews of 25,000 people: 13,000 from one large organization, 2,800 from one midsize company, and 10,000 from a collection of over 250 small organizations. Analyzing the actual written texts gave Textio the opportunity to understand the extent to which they align with employees' perceptions of the feedback they are given — and for the large and midsize companies, whether their performance reviews reflect their state's company cultures.

It found there were substantial biases in the amount of feedback given by race and ethnicity, as well as by age. Asian employees received 105% of the company average, while Latino employees received 88% and Black employees only 81%. In a similar vein, employees under age 30 received only 80% of the company's average amount of feedback. Again, less feedback means less opportunities to improve and advance.

Looking at personality feedback alone, there were two clear biases: Men receive less than women, and Asians also receive less. But personality feedback is intersectional: For example, Black women receive 2.7 times the amount of personality-related feedback that Asian men do.

There were also dramatic biases in actionable feedback: Women, Latino, and Black employees receive substantially more feedback that's not actionable, as do employees over age 40. The intersectionality here is also huge: According to the report, "For every piece of feedback that's not actionable received by white men under 40, women over 40 receive 4.4; Black women (across age groups) receive 8.8."

The study also looks at exaggerated feedback, as well as feedback that focuses on fixed or innate characteristics rather than aspects people can change.

There are things we can critique about the study: First, it comes from a company that sells a product to address these kinds of issues. On one hand, it's good it is doing (and sharing) customer research; but we need to ask hard questions about whether that might lead it to exaggerate the problems in order to sell more product. That doesn't necessarily mean Textio's research results are wrong, but we should be looking to other independent studies to confirm them. Second, about 60% of its written performance feedback came from only two companies; that should lead us to question whether the results broadly generalize to all other large and midsize companies. Third, the segmentations the report uses are oversimplified: Gender isn't binary, nor is race/ethnicity in many cases.

At the same time, there are clear biases that show up, ones that have significant impact on individual employees' careers and livelihoods. Performance feedback directly affects employees' opportunities to improve, succeed in the workplace, and advance their careers. But studies have also shown that racial/ethnic and gender biases in compensation persist in the workplace, and those differences in compensation align with the reported findings on biases in negative feedback. In other words: Words matter — a lot.

Language Bias in Performance Feedback

Featured image by fizkes/Shutterstock.com.