FAQ
Common questions about the gender earnings gap analysis.
The percentage difference in hourly earnings between male and female workers. In 2023 ACS data, women earn 16.8% less per hour than men among prime-age wage/salary workers (raw gap). After controlling for observable characteristics, the adjusted gap is 13.2%.
The gap after statistically controlling for age, race, education, state, occupation, industry, hours worked, commute, work-from-home status, marital status, and children. It represents the residual difference not explained by these observable factors.
ACS (16.8% raw, 13.2% adjusted), CPS (15.8% raw, 17.0% adjusted), and SIPP (15.1% raw, 10.9% adjusted) use different survey designs, sampling frames, and variable definitions. The raw gaps are similar; adjusted gaps vary more because each dataset has different available controls.
Adding occupation and industry controls reduces the gap by about 8 percentage points in ACS data. This is the single largest observable channel.
A supplemental decomposition that splits the total gap into an "explained" portion (observable characteristics) and an "unexplained" portion (residual). In 2023 ACS data, 87.8% of the Oaxaca gap is unexplained, but this repo treats sequential OLS as the headline series because it is more stable over time.
The total gap stayed stable, but the explained component collapsed. That looks more like a decomposition-composition issue than a sudden structural break in the adjusted gap, which is why Oaxaca is kept as a supplemental diagnostic rather than the public headline metric.
ATUS data shows women spend 68 fewer minutes per day on paid work but 43 more minutes on housework and childcare. The net gap (paid + unpaid) is about 25 minutes/day, with women doing less total paid work but more total work including unpaid labor.
Substantially. The raw gap is 19.3% among White non-Hispanic workers, 20.4% among Asian workers, but only 5.6% among Black workers. This reflects differences in both male and female earnings distributions across racial groups.
No. The gap generally widens with education, peaking at the Master's degree level (27.4%). Only at the doctorate level does it narrow slightly (17.7%).
Yes, MIT License. All data sources are free, public U.S. federal datasets (ACS PUMS, CPS ASEC, SIPP, ATUS, SCE).