How Much Would You Make if You Were a Man?

President Barack Obama is now revisiting an issue that marked the very beginning of his presidency: Equal pay for men and women. On Jan. 29, seven years to the day after Obama signed into effect the Lilly Ledbetter Fair Pay Act, he acknowledged that the wage gap still persists to an alarming degree—and introduced several new measures to help address the problem.

One of these efforts is a proposal to increase the amount of data the federal government receives about employee pay. But a great deal of this information already exists in detailed data collected by the Census in their annual American Community Survey. According to a new analysis by TIME and Motto, women earn less than men at every age range, from 15% less between ages 22 and 25 to a staggering 38% less between ages 51 and 64.

We analyzed the wages earned by more than 15 million Americans surveyed by the Census Bureau between 2008 and 2012, who were classified into 460 different occupational categories. We found that 372 of those occupations contained enough information to determine that there was a statistically significant gender pay gap in at least one of the seven age ranges we examined.

In all, we found 1,490 combinations of age and occupation in which there existed a statistically significant difference in how much men and women make. Only 22 of those 1,490 involved women making more money than men. All 22 situations where women out-earned men occurred among the youngest cohorts in an occupation. (The gaps are statistically significant to at least a 90% level of confidence, the same threshold the Census uses).

There was not a single occupation where women age 30 and above took home significantly higher average salaries than their male colleagues, even if they start out making as much or more.

Read more: So You’re Making Less Than a Man—Now What?

Looking at the earning patterns for secondary-school teachers helps tell this story. From ages 22-25, women actually earn 10% more than their male colleagues. By the time those teachers turn 30, there is no statistical difference between them and their male peers. By 35, the men make more. It’s a pattern we see repeated with librarians, social workers, secretaries and preschool teachers. Notably, these are all female-dominated professions. In male-dominated professions like engineering and finance, women typically make less from the very start.

Some occupations stand out as particularly egregious in our research. For example, the tax-preparation industry, where we had a sample of over 5,000 people, has one of the worst records of pay disparity of any job in the country. From ages 26-30, female tax preparers, on average, earn around half as much as their male counterparts. It gets worse from there. Among those between 41-50, women in the tax-preparation business bring home only a third as much as their male colleagues.

One explanation for pay disparity is that many occupations offer higher- and lower-paying tracks. We could hypothesize, for instance, that more female lawyers choose public service and more male lawyers choose high-paying firms, perhaps explaining why male lawyers in their 40s earn 51% more than their female counterparts. But the sheer breadth of occupations in which men make more than women suggests something other than self-selection into tracks. For instance, male dishwashers in their 50s are paid 20% more than female dishwashers. Male janitors in their 40s are paid 63% more.

Joining a profession where women make up the majority of workers may not give women a pay advantage either—or even parity. There is research to show that in female-dominated professions, men get promoted more quickly than women do in their field, an effect that has been termed the “glass escalator.” Male and female registered nurses, our research shows, make roughly the same pay in the youngest cohorts, only to have a significant wage gap open up in their 30s and beyond.

The survey does confirm that there are more men than women in certain high-paying professions (financial analysts, dentists) and more women in certain low-paying professions (teaching assistants, telemarketers). But within all of those professions, men still make significantly more. Having more women in higher-paying professions, in other words, would not eliminate the overall gap.

Some point to the fact that women remain primary caregivers in many families as the cause for their wages lagging behind those of men. A woman might leave the work force to raise a child and when she returns enter at a lower wage and never catch up. While this may happen in some places, childcare alone cannot account for the vast inequality we measured across almost every profession and age group.

The calculations in our analysis exclude anyone who did not earn an income, but do include part-time workers. While men who work part-time still tend to earn more than women who work part-time, there are some occupations and age cohorts where female part-timers outnumber male part-timers. This could explain some of the gap in those groups, but it cannot explain it all. Researchers found that among full-time workers only there is still an intractable gap, even after controlling for things like occupational segregation and employment gaps due to maternity and childcare.

Alarmingly, the discrimination happens not just during promotion season but when a job candidate first shows up at a company. In a variety of professions, male-dominated or female-dominated, experiments have shown that employers are biased toward male job candidates. In a 2012 double-blind study, science faculty at research universities were asked to rate identical resumes for a lab tech position, one with a male name and one with a female name. When the name on the resume was John, faculty suggested a higher starting salary than when the name on the resume was Jennifer.

Amelia Showalter, a data analytics expert and political consultant, was director of digital analytics for Barack Obama’s 2012 presidential campaign. Chris Wilson is the director of data journalism at Time.com.

Source: IPUMS-USA, University of Minnesota, www.ipums.org.