Thursday, February 4, 2016

As The "Wheel O' Meter" Turns

In the dog-bites-man department, PolitiFact continues to do fact-checking the wrong way this week.

It started with a fact check of Republican presidential candidate Ted Cruz. Cruz said blacks gained $5,000 in median income under Reagan. PolitiFact's crack fact-checking team, veteran writer Louis Jacobson and founding editor Bill Adair's heir, Angie Drobnic Holan performed their fact-checking magic and found Cruz's claim "Mostly True."

But a day passed, and PolitiFact unpublished the article and replaced it with a rewritten version giving Cruz a "Mostly False" rating. Why? Somebody complained:
CORRECTION, Feb. 4, 2016: After we published this fact-check, a reader wrote us to say that in our original article, we had used the wrong Census Bureau data table to analyze Cruz’s statement and rate it Mostly True. However, using the proper data table, Cruz is actually incorrect, so we have changed the rating to Mostly False. 
This case gives us a smorgasbord of items to criticize.
  • PolitiFact's A team committed a substantial flub
  • The archived "original version" hides Holan's role in editing the flawed version
  • The archived "original version" wipes (with a cloth?) the hotlinks leading to the wrong table 
  • Neither the new nor the old version bothered to look at how Cruz fared using nominal dollars
Having noted the main problems with PolitiFact's treatment of this issue, we'll address two questions. First, what specific mistake did PolitiFact make before publishing its less-than-totally-transparent correction? Second, was Cruz correct using nominal dollars?

The Wrong Chart

We found PolitiFact's key figures from its original article on the same chart it used in its revised article. The chart gave figures for the United States as a whole followed by corresponding figures for different regions of the country. It appears PolitiFact used figures for the West region to test Cruz's claim.

Nominal Dollars

Using nominal dollars, Cruz's claim was correct. That's typically enough to eke a "Half True" from the PolitiFact gang's "Truth-O-Meter."

We reproduced the relevant part of the chart, putting the figures PolitiFact used in its new fact check in black boxes. The adjacent figures for nominal dollars ("Current dollars") we placed within red boxes. The difference over Reagan's two terms in office was $5,643.

We're not saying there's nothing wrong with giving increases in median income using nominal dollars. It's potentially every bit as misleading as when Democrats use the raw gender wage gap to emphasize the need for legislation punishing pay differences based solely on gender discrimination. What we are saying is Cruz made a true statement given a normal interpretation of his words and PolitiFact failed to acknowledge it either in the text of the story or in its (new) rating of Cruz.

PolitiFact tends to rate that bit of liberal misdirection no lower than "Half True," by the way.

What will it take for PolitiFact to regard its "Truth-O-Meter" rating system as a mistake?


Does PolitiFact's faulty first rating of Cruz show PolitiFact has a pro-conservative bias?

We don't think that's the best explanation. The more parsimonious explanation is that PolitiFact simply made a mistake. So this example shows PolitiFact's fallibility. This was the type of mistake bloggers make.

This example does point toward another potential evidence of bias at PolitiFact. Which political party receives the most benefit when PolitiFact runs a correction?


  1. I am wondering had that been rated on a liberal and a conservative had pointed out the mistake, would they have corrected it to show that the liberal politician was wrong?

    I know I have pointed out their mistakes before and have never gotten a response.

    1. As we noted at the end, we can (and will) take a look at which poltical party gains the most from PolitiFact's corrections.

      That study is worthwhile in part because we, like you, have pointed out mistakes that PolitiFact has never bothered to correct. Unfortuanately that will also count as a weakness of the study: we don't know how many criticisms PF has ignored from both ends of the political spectrum.


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.