Wednesday, May 8, 2013

PolitiFact and the 77-cent solution (Updated)

Is there gender discrimination in wages?

PolitiFact, a project of the Tampa Bay Times supposedly designed to help you find the truth in politics, has the answer.  In fact, PolitiFact does even better than giving us an answer.  It gives us two different answers to the same question.

Is it true that "women earn 77 cents for every dollar earned by a man"?

It's "Mostly True," says PolitiFact.  It's "Half True," says PolitiFact.

You'd think they might be able to settle on "Mostly Half True."

How is it that PolitiFact can reach two different conclusions about the same claim, know that it has reached two different conclusions regarding the same claim and yet fail to resolve the discrepancy?

This is supposed to be fact checking, not "Wheel of Fortune."

We've said for years that PolitiFact's rating system by its nature forces reporters and editors into making subjective judgment calls.  This case serves as yet another example supporting that claim.

Could some difference in the claims or the context of the claims justify a different rating?  PolitiFact mentions no such differences.  Yet PolitiFact has terrific motivation for explaining the different ratings.  In its recent fact check of Rep. Marcia Fudge's "77 cent" claim, PolitiFact Ohio cited other PolitiFact ratings of similar statements:
PolitiFact has made several examinations of the claim that women earn 76 to 77 percent as much as men, and found that they lacked context because they failed to account for factors like education, type of job, age of employee and experience level.
The hotlink associated with "several examinations" leads to a "Half True" rating from PolitiFact Georgia for a claim effectively identical to Fudge's.  Fudge received a "Mostly True" rating.

The writers and editors at PolitiFact apparently don't realize that linking to a closely parallel fact check with a different rating exposes a problem of inconsistency.

Inconsistency isn't bias!

By itself, inconsistency is not bias.  But patterns of inconsistency may provide evidence of bias.  We have that sort of pattern in PolitiFact's ratings of differences in pay by gender.

We can measure by tracking the frequency with which stories either favor one political party over another or cause harm to one party more often than to another.  My co-editor at PFB, Jeff D, points out that Republican presidential candidate Mitt Romney made a claim about differences in pay by gender during the 2012 election.  Romney noted that the equal pay candidate, President Obama, was paying male White House employees more than the female employees.  PolitiFact found that Romney was right.  And rated the claim "Half True":
In the broadest sense, the Romney campaign is on solid ground when it says that "women in Barack Obama's White House are earning less than men." But the closer you look at the data, the less striking this conclusion becomes.
...The statement is accurate but needs clarification or additional information, so we rate it Half True.
"The statement is accurate but needs clarification or additional information," so PolitiFact rates it "Half True."  There's just one problem.  That's the definition PolitiFact gives for "Mostly True":

MOSTLY TRUE – The statement is accurate but needs clarification or additional information.

Even aside from that PolitiFact blunder that somehow escaped the notice of layers of editors, we see a pattern of partisan inconsistency.

Romney's statement, as Jeff points out, avoids false precision.  Romney simply says men get paid more than the women at the White House.  It's very hard to argue that Romney's statement is in any way more misleading than any of the "77 cent" claims.  Indeed, it's hard to argue that Romney misled any more than did the National Women's Law Center with its claim that every state has a gender wage gap.  PolitiFact Georgia rated that claim "True."  PolitiFact simply doesn't provide reasoning that would distinguish one rating from another in this similar set of claims.

Whether the correct rating is "Mostly True" or "Half True," the Republicans draw the short straw with PolitiFact in comparison to Democrats.


Here's the list of similar gender gap stories, followed by two stories where claimants used the 77 cent figure claiming it's the difference where men and women do the same work.

Diana DeGette says women earn 77 cents for every dollar earned by a man
"Mostly True"

R.I. Treasurer Gina Raimondo repeats oft-quoted, but misleading, statistic in equal pay debate
"Half True"

Rep. Marcia Fudge cites wage gap between Ohio women and men
"Mostly True"

Gender wage gap claim needs more context
"Half True"

Tim Kaine says Virginia women earn 79 cents to every $1 made by men
"Mostly True"

[National Women's Law Center] Is there a gender wage gap in every state?

Mitt Romney says women White House employees earn less than men under Barack Obama
"Half True"

Same job, same work

U.S. Rep. David Cicilline says women earn only 77 percent of what men earn in the same job
"Mostly False"

Barack Obama ad says women are paid "77 cents on the dollar for doing the same work as men"
"Mostly False"

Update July 10, 2013

A reader alerted us to another PolitiFact rating that fits with this group.  Former U.S. president Jimmy Carter lowers the bar for "Mostly False" by making the same job, same work claim while naming the wrong percentage.  Carter said the wage gap for the same job and same work averaged 70 cents on the dollar.  "Mostly False," said PolitiFact Georgia.

We wonder how low one could go with the percentage and still rate higher than "False"?


  1. Very simply, because it depends on the claim the politician made. Some politicians claim women make 77 cents to the dollar _for the same work_ (which is largely untrue), while others claim it's 77 cents to the dollar in general. And if we average over all workers, divided only by gender, this is true. The differences comes from the types of jobs people have. Women and men in the same jobs have something like a 94 cent to the dollar ratio, from my understanding.

    1. Durinn, did you read our post before replying?

      At the end of the post we listed the PolitiFact stories grouped by whether they did or did not use the 77-cent statistic to represent different pay for the same work.

      From our perspective it looks like you completely missed the details in our post that support its point. If we somehow missed your point, feel free to add more comments.


Thanks to commenters who refuse to honor various requests from the blog administrators, all comments are now moderated. Pseudonymous commenters who do not choose distinctive pseudonyms will not be published, period. No "Anonymous." No "Unknown." Etc.