Equal Pay Day: a wage gap fact check

Hey there, hope your week is going well. I’m delivering this week’s fact check a bit early. Yesterday, I got an email from Daniel in Lancaster, California, that said: “How would someone go about checking wage gaps? How can you check for gaps across genders and across racial/ethnic groups?”

Great question, Daniel, and a timely one for Equal Pay Day which is on 4 April this year. That date is symbolic – it shows roughly how many days of 2017 women need to work to earn as much as men did in 2016. Tuesdays are symbolic too, they represent “how far into the next work week women must work to earn what men earned the previous week”, according to the National Committee on Pay Equity.

But there are people who doubt that it’s as simple as that. Like the author of this Time article who cites the wage gap among “feminist myths that will not die”. Or contributors to mens’ rights forums. Let’s check the math.

Annotated Census Bureau screenshot

Photograph: Census Bureau

Step 1: Find a reputable source. The Census Bureau is a good place to start. I search for “census bureau wage gap” and click on the first link in the results. I’m not interested in “news” though, so I click on the “publications” part of their site and choose the most recent report.

The table of contents says that figure 2 contains the “Female-to-Male Earnings Ratio”. That sounds about right. It says that in 2015, women earned 80% of what men did. More specifically, the table below it shows that men on average earned $51,212 that year while women earned $40,742. If I want to check that “80 cents on the dollar” claim that is so frequently cited, I need to divide 100 by $51,212 and then multiply by $40,742 to arrive at the percentage. The answer is 79.5555729126. I don’t blame activists for rounding – “79.6 cents on the dollar” doesn’t have quite the same ring to it now does it?

Step 2: Read the small print. That chart shows earnings for “full-time, year-round workers who are 15 years and older”. And, in case you were wondering, the Census Bureau explains “a full-time, year-round worker is a person who worked at least 35 hours per week (full time) and at least 50 weeks during the previous calendar year (year round)”.

Hm. Well Daniel, I think we’re starting to understand some of the “why” of the wage gap. See, when I search for “hours worked by gender”, I quickly find that men work more hours than women in a typical work week according to the Bureau of Labor Statistics (BLS). That’s something people are quick to point out but slow to understand.

On average, women spend less time in paid employment because there are only 24 hours in a day and women are more likely than men to spend their time in unpaid work. Things like raising children, taking care of elderly family members and doing housework. What’s more, men are more likely to hold senior positions and work in occupations that pay better (5% of architectural and engineering managers are women, 5% of childcare workers are men). All of these factors are important in understanding why the pay gap exists.

Step 3: Check the source. One of the simplest ways to do this is to see if other reliable sources come up with a similar number. It turns out they don’t. The BLS claims that women earn 82 cents for every $1 a man makes. But again, the small print is crucial here: the BLS are looking at data for 2016, not 2015, and they’re using weekly earnings rather than annual ones. The difference between their figures and the Census Bureau’s isn’t too large to warrant concerns that either is false.

Step 4: Find out if the statistics accurately reflect all groups. It kind of sounded like that was the real thing you wanted to check, Daniel. And it’s one of the arguments made by those who dub the wage gap a “myth”. Well, that BLS page I just mentioned already provides some clues.

See, that 80 cents on the dollar statistic varies a whole lot by profession. Last year, female physicians and surgeons earned a paltry 63 cents for every dollar their male counterparts earned, while female cashiers earned 85 cents for every dollar male cashiers made. (If you’re interested in looking up specific occupations, this interactive from the Wall Street Journal is pretty great).

But there’s a lot more to it than that. The gender wage gap also varies depending on where you work, how old you are, your educational status and your race or ethnicity.

The problem is, we can only look at each of those things one at a time. Finding out the wage gap for a black woman with a master’s degree working in engineering in San Francisco is impossible to calculate reliably because survey analysts will only have a few individuals that fit that description to make their estimate. You need a lot of people if you want to avoid making conclusions based on outliers.

So let’s look at some of these factors one at a time. First up, geography. I used the Census Bureau’s Fact Finder site to get these numbers (warning: the site is a real pain in the butt – you have to search by selecting “sex” and “Income/Earnings – Individuals” and then pick the table you need). I found that the wage gap varies a lot across the country. Women in Utah earn 55 cents for every $1 that men in the state earn – in DC that figure is 86 cents, behind Puerto Rico, the only state or territory in the country where women actually out-earn men by two cents on the dollar.

Wage gaps by state

Wage gaps by state

And the Fact Finder site has data on race too. This is where you can really use statistics to tell whichever story you choose. See, you could compare within a race (Hispanic women earned 91 cents for every $1 earned by Hispanic men in 2015) or you could compare between race and gender (Hispanic women earned 54 cents for every $1 a white man earned that year).

So yes, Daniel. That 80 cents on the dollar figure is misleadingly simplistic. For women of color, women in the midwest and those working in certain occupations, the gap is much larger than that.

Would you like to see something fact-checked? Send me your questions! mona.chalabi@theguardian.com / @MonaChalabi

This entry was posted in Data Visualization. Bookmark the permalink.