Latest / Women at Work: Career, Leadership & Modern Work / Why Women Leave Tech The Microaggression Tax
Transcript
- Lucas: So here's a number that stopped me this week: women in tech quit at a 45 percent higher rate than men. And for years, the standard explanation has been something like 'they leave for better work-life balance' or 'they just don't like the culture.' But there's a new study out of the Kapor Center that suggests the real driver is a lot more specific — and a lot more fixable. Luna: 45 percent higher — that's massive. What did the study actually track? Lucas: They followed 1,200 women engineers over three years. And what they found is that the best predictor of whether a woman would leave wasn't her salary, her title, or even whether she had kids. It was something they call the 'microaggression tax.' Essentially, the cumulative toll of small, everyday slights that pile up until the cost of staying outweighs the benefit. Luna: I love that framing — a 'tax.' It makes the invisible visible. So what kind of slights are we talking about? Lucas: The study identified four categories that accounted for about 80 percent of the incidents. First: interruption — being talked over in meetings. Second: credit theft — having your idea repeated by a male colleague and then praised. Third: competence skepticism — having your expertise questioned more frequently than peers. And fourth: isolation — being left out of informal networks like after-work drinks or the group chat where real decisions get made. Luna: That fourth one — isolation — is interesting because it's the least obvious. It's not something you can easily report to HR. Lucas: Exactly. And the data gets really specific here. Women who experienced five or more microaggressions per month were 3.2 times more likely to quit within the year. That's after controlling for every other variable the researchers could measure — including overall job satisfaction. It's not that they were unhappy with their work; it's that the tax became too high. Luna: 3.2 times — that's not a subtle effect. So what did the women who stayed have in common? Lucas: Good question. The study found that the strongest mitigating factor was having a manager who actively interrupted bias in real time. When a manager — male or female — would step in during a meeting and say something like 'Let's let her finish' or 'That was actually Sarah's idea, just to credit her,' the likelihood of quitting dropped by nearly 40 percent. It's a simple intervention, but it has to be consistent. Luna: So the burden of fixing the problem is on managers, not on the women experiencing the microaggressions. That's a shift from the usual 'resilience training' narrative. Lucas: Right. And that's where I want to push back a little on the 'lean in' framework that dominated the last decade. The Kapor study suggests that individual grit isn't the lever. In fact, women who reported 'high resilience' on personality tests were just as likely to quit as women who scored low — because the tax is structural, not psychological. You can't grit your way out of being interrupted three times in an hour. Luna: That's a powerful finding. So what's the fix beyond manager intervention? Did the study test any other interventions? Lucas: They did. The most promising one came from a pilot at a major cloud services firm — I won't name them because the data is still under peer review, but it's a company you'd recognize. They implemented what's called a 'bias interruption protocol' at the team level. Every meeting had a designated 'process observer' — a rotating role — whose only job was to flag microaggressions using a simple code. For example, if someone was interrupted, the observer would say 'Check' — just one word — and the speaker would get the floor back. Luna: A single word — that's so low-friction. Did it actually work? Lucas: Over 18 months, attrition among women on teams using the protocol dropped 26 percent compared to control teams. And here's the kicker: the protocol didn't slow down meetings. If anything, decisions came faster because fewer ideas were lost in the noise. The researchers estimate that if every tech company with over a thousand employees adopted something similar, the industry would retain an additional 18,000 women engineers per year. Luna: 18,000 — that's not a rounding error. But I wonder, is this only a tech problem? I've heard similar stories in finance and consulting. Lucas: The Kapor study focused on tech because the attrition gap is widest there, but the same pattern shows up in law, medicine, academia — any field where informal power dynamics are strong. A 2022 meta-analysis of 85 studies found that microaggressions predicted turnover across all white-collar sectors, with an average effect size almost identical to what the Kapor team found. So this is a general workplace phenomenon. Luna: Let me play devil's advocate for a second. Some people argue that microaggressions are subjective — that one person's 'interruption' is another person's 'enthusiastic collaboration.' How do you standardize something like that? Lucas: It's a fair concern. The researchers addressed it by using a technique called 'behavioral coding.' They trained observers to look for very specific, observable behaviors — for example, someone speaks for less than 20 seconds and is cut off, or an idea is paraphrased by a different person within 60 seconds and attributed to that person. They didn't rely on self-reporting. When they cross-referenced the behavioral data with exit interviews, the correlation was extremely high. So the behaviors are real, even if intent is ambiguous. Luna: That's rigorous. So for a woman listening who's wondering if she's experiencing the microaggression tax, what would you tell her to look for? Lucas: I'd say start with a simple log. For two weeks, every time you feel that small sting — the interruption, the credit theft, the questioning of your expertise — write it down. Don't judge it, just note it. At the end of two weeks, count how many you had. If it's more than five per month, that's the red zone the study identified. Then ask yourself: how much energy am I spending managing this versus doing my actual job? Luna: That reframes the problem as a cost-benefit calculation, not a character flaw. I like that. Lucas: Exactly. And if the tax is high, the next step is to look for leverage points. Is there one manager or peer who consistently has your back? Can you request a meeting structure that includes a process observer? The study shows that even small structural changes can reduce the tax by a third. You don't have to fix the whole culture overnight. Luna: What about the role of men in this? The study found that male allies stepping in was powerful. But do men actually know what to look for? Lucas: That's the second phase of the Kapor study, actually. They surveyed 600 male engineers and asked them to identify microaggressions from the same video clips the female coders had watched. On average, men spotted only 40 percent of the incidents that women identified. So there's a perception gap. But the good news is that a 30-minute training session — focused on specific behaviors, not general 'bias awareness' — closed that gap by 60 percent. So men can learn to see it, and once they see it, they can act. Luna: Specific behaviors, not general awareness — that seems to be the theme of the episode. What's the one takeaway you want listeners to leave with? Lucas: I think it's this: the microaggression tax is real, it's measurable, and it's not your fault. But it is your decision whether to stay and try to change the system or leave for one where the tax is lower. The research gives us a framework to make that decision with data, not just frustration. And for leaders listening: the fix isn't expensive. It's a one-word prompt and a rotating role. Luna: A one-word prompt. That's a surprisingly small lever for a problem that costs the industry tens of thousands of women every year. Lucas: Exactly. Sometimes the biggest problems have the smallest solutions staring us in the face. We just have to be willing to name them.