We've already told you about the so-called "benevolent sexism" that conceals gender biases under seemingly-innocuous language. But in performance reviews, those same biases aren't so sublimated. That's what tech founder Kieran Snyder realized after studying hundreds of reviews written about men and women. Women receive far more negative feedback than men, she discovered, and are far more likely to be critiqued for their personalities than they are for their skills.

In an article for Fortune, Snyder says her interest in this topic was piqued when she talked to an engineering manager who was considering promotions for two members of his team: "'Jessica is really talented,' he said. 'But I wish she'd be less abrasive. She comes on too strong.' Her male counterpart? 'Steve is an easy case,' he went on. 'Smart and great to work with. He needs to learn to be a little more patient, but who doesn't?'"

Snyder says she focused on performance reviews because a) they represent written records of perceptions and b) they could provide statistical evidence. After asking men and women in tech to share their reviews (but not specifying the nature of her study), she received 248 reviews from 180 people across 28 large, mid-size, and small tech organizations. She noticed some definite trends across the entire sample—in reviews from all companies of all sizes, regardless of the gender of the person writing the review.

For starters, 59 percent of the reviews of male employees contained critical feedback, compared to 88 percent of the reviews of female employees.

Not only that, but men received different kinds of feedback than women did. Men were encouraged to develop their skills—e.g. "It would have been extremely helpful if you had gone deeper into the details to help move an area forward," and, "Take time to slow down and listen. You would achieve even more."

But women were told to "pipe down," as Snyder writes—e.g. "You can come across as abrasive sometimes … You need to pay attention to your tone," "Sometimes you need to step back to let others shine," and, "You would have had an easier time if you had been less judgmental about R—'s contributions from the beginning."

In fact, negative personality criticism showed up in 76 percent of the critical reviews received by women—compared to just 2 percent of the critical reviews received by men.

Finally, Snyder notes, these performance reviews mirror the pervasive stereotypes that women are "bossy," "abrasive," "strident" and "aggressive" when they lead and "emotional" and "irrational" when they object. Each of those words showed up at least twice in the women's reviews, some appearing much more often. The word "abrasive," for example, popped up 17 times in reference to 13 different women.

Out of all those words, however, only the word "aggressive" showed up in the men's reviews and only three times. And in two of those three instances, the reviewers were encouraging more aggression.

Not to make her sound "abrasive" or "bossy," but Snyder urges employers to reflect upon these trends and to check not just the quantitative substance of performance but the qualitative, as well. "At most mid-size or large tech companies, HR leaders supervise review scores to uncover and correct patterns of systematic bias," she writes. "This is a call to action to bring the same rigor to the review language itself."