Wednesday, June 29, 2016

Offsite Criticism of the Week

While PolitiFact continues its pattern of ducking public criticism, we will continue to model the opposite approach by dealing with a typical offsite criticism we run across (real example from real discussion board):


First, we'll note that generally citing PolitiFact Bias is the wrong way to defuse criticisms that Trump doesn't tell the truth. There are good reasons to dismiss the use of the PolitiFact statistics that some use to "prove" somebody lies more than somebody else. The liberal use of selection bias lays a poor foundation for reliable statistics. We've been pointing that out for years.

As for the criticism:
The first article in the first link says PolitiFact is wrong because it only used the data that was available to make its ruling. Oh the humanity!
No.

We believe the comment was directed at our evaluation of the Loretta Lynch fact check where Lynch said the LGBT community endures most of the statistically recorded hate crimes. It's true, as Dustin Siggins pointed out, that PolitiFact relied on incomplete statistics. In keeping with PolitiFact's common practice, that should have prevented Lynch from getting a rating any better than "Mostly True." Lynch received a "True" rating.

But our central criticism was quite different from what the writer said.

We noted that Lynch said nothing, in context, about using the per capita measure. If that measure isn't used, then PolitiFact's statistics say blacks experience most of the hate crimes recorded in the statistics. So what Lynch said was false. PolitiFact pretended otherwise to give her the "True" rating.

This is typical of the criticism we receive. We're dismissed based on bias (ad hominem fallacy) and on misrepresentations of our work (straw man fallacy).

We'd love to do "Offsite Criticism of the Week" for a criticism that is not fallacious. Comments are open here and on Facebook if anybody knows of one.

6 comments:

  1. I noted the issue to several people. I phrased it as the factcheckers had to add words to her statement to then come to a rating.

    Had they added different words (raw numbers vs. per capita) they could have come to a false rating.

    I also pointed out consistency given the gaps in the data issue because a year ago, they rated NAACP President Cornell William Brooks as Mostly True because the experts agreed but the data has inaccuracies.

    ReplyDelete
    Replies
    1. Forgot the link to last year's analysis:

      http://www.politifact.com/truth-o-meter/statements/2015/jun/25/cornell-william-brooks/how-many-hate-crimes-are-there-united-states-2/

      Delete
    2. **Had they added different words (raw numbers vs. per capita) they could have come to a false rating.**

      That's right, but it's even worse than that. The words Lynch chose are most naturally interpreted as raw numbers. For PolitiFact to ignore that dimension of the fact check shows grave incompetence if not bias.

      Delete
  2. Any statistician will tell you that per capita is the way to go when measuring rate of things among groups or else things get muddled in population. You guys look for bias so hard you lose the truth.

    ReplyDelete
    Replies
    1. Any statistician saying that, Mike Hunts, is likely also going to say that it is proper to identify the statistic as "per capita" to prevent confusion.

      You're not going to find a conservative or Republican getting the same type of break Lynch got. We see it all the time from PolitiFact.

      What truth do you think we lost on this one, by the way?

      Delete
  3. FBI Director James Comey Makes Statement About Hillary Clinton Email Scandal - FULL SPEECH

    http://commoncts.blogspot.com/2016/07/fbi-director-james-comey-makes.html

    ps. Would you please consider adding CC to your blogroll?

    ReplyDelete

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.