Persistent Pay Gap: Survey Respondents Paid Women Less For The Same Job

People Recommended Lower Pay for a Job They Were Told Was Mostly Women

A new study set out to answer one timely career question: if respondents are told most workers in a job are women, will they recommend less monetary compensation for the position? The results suggest the gender pay gap is very much still here. Simply describing a job’s workforce as mostly female was enough to make respondents suggest lower pay for it.

Published in the journal Research in Social Stratification and Mobility, the study recruited a large sample built to mirror the U.S. population in age, gender, and roughly in education. Every respondent read the same three-paragraph description of a management consulting job. The only difference was one line near the end, stating that women made up either about a quarter, nearly half, or two-thirds of the people currently in that field. Then came the real question: what should a new college graduate be paid to start in that job?

That single detail, how many of the job’s current workers were described as women, is what makes this study different from decades of research before it. Past studies showed jobs held mostly by women pay less in the real world, but critics always had an out: maybe those jobs are just physically easier, or need less training, or differ in some other unmeasured way. This experiment closed that door. Every respondent read an identical job description. Only the stated gender makeup changed, and that alone shifted what respondents thought was fair pay.

Over Half the Survey Takers Got Dropped for Forgetting the Setup

Researchers surveyed people through Qualtrics in two rounds, in 2021 and 2022, aiming for a group that reflected the country by age, education, and gender. Since it was an opt-in online panel, results may not perfectly reflect every American.

Respondents were told the survey was a joint project between researchers and an anonymous company trying to figure out fair pay. They read about the job, learned its made-up gender breakdown, then set a salary between $40,000 and $100,000. To confirm they’d absorbed that detail, researchers later asked them to recall it. More than 12,000 people took the survey, but over half were dropped for misremembering the number. That left 6,705 in the main results. Checking the full, unfiltered group, the same downward pattern showed up, but weaker and no longer strong enough to rule out chance.

A Majority-Female Label Cost About $865 in Suggested Pay

Those core numbers tell the story plainly. Respondents told the job’s workforce was 25% women recommended $55,403 on average. Those told it was 67% women recommended $54,478, a real-world gap of about $925. Adjusting for who took the survey and when, that gap settled at about $865, and held up even after checking it against age, education, income, race, region, and marital status one at a time. Tossing every factor into a single calculation shrank the gap further, to around $714, and made it shakier, though it never fully disappeared.

Oddly, the effect didn’t build gradually. Going from 25% women to nearly half made no real difference in what respondents were willing to pay. The drop only showed up in the comparison against 67% women, one of just three workforce breakdowns the study tested, so it’s not clear this points to any specific cutoff. Also worth noting: female respondents themselves recommended lower salaries than male respondents did, across the board, no matter which version of the job they read.

Respondents Didn’t Rate Women Workers as Less Skilled, Yet Still Paid Them Less

Researchers didn’t expect what came next. They also asked respondents to rate the hypothetical workers in that job on things like skill, warmth, and how much they contribute to the company’s bottom line, betting that a “mostly women” workforce would drag those ratings down too. It barely moved anything. If anything, respondents rated profit contribution slightly higher when the workforce was described as more female, the opposite of what a straightforward bias story would predict.

So respondents weren’t rating women workers as less skilled or less valuable. They just recommended less pay for them anyway. What’s actually driving that gap remains an open question the researchers can’t answer from this data alone.

Whether This Holds Beyond One Job and One Survey Is Still Unknown

It’s worth being upfront about what this experiment can and can’t prove. It tested one job, three workforce breakdowns, and one moment, a starting salary decision, not a career. Nobody knows yet whether the same pattern shows up for a nurse, a coder, or a construction supervisor, or whether it holds outside a survey, in an actual HR office. What is reasonable to guess: if a bias like this shaves a few hundred dollars off a starting salary, and that gap compounds through every raise that follows, it could add up to something much bigger over a career. That’s a plausible extension of the findings, not something this study measured.

Even with those limits, the experiment offers a direct look at the moment a pay number gets chosen, and evidence that simply changing the stated share of women in a job can nudge that number down.

---

Disclaimer: This article summarizes findings from a peer-reviewed academic study. It is intended for general informational purposes and does not constitute financial, legal, or employment advice.

---