Not only was the number that Sanders quoted significantly lower, but his campaign provided Politifact with the study that they referenced while Trump's campaign did not respond.

The campaign providing them with a study has no effect on how true the statement is, which, as a fact checking organization, was their job to discover. This is surely evidence of, if not bias, gross incompetence.

That's not the point. The point is 2 sides of the aisle reported similar metrics and politifact 'ruled' at the 2 different extremes.

1. 51% and 59% aren't "similar" on such a large scale.

2. The vagueness of Trump's statement allows for far more misinterpretation, which it is his responsibility to avoid in order to be factual. Sanders mentions a specific age range of 17 to 20, which the study that his campaign later referenced was looking at.

3. As you may know, these "rulings" were clearly not at the two different extremes. Source: http://www.politifact.com/about/

If you read the two pages, it is pretty bad. The fact checkers point out the same BLS numbers for each, but then go on to expand into areas Sanders was actually less clear on. They gave him a huge benefit of the doubt to using the think tank numbers, but then they don't apply those same numbers to Trump.

> They gave him a huge benefit of the doubt to using the think tank numbers

It's not "benefit of the doubt" when you contact someone and ask for the basis of their claim and they provide it, and it matches the claim numerically quite specifically but the terminology used is still misleading and so you dock for that.

> but then they don't apply those same numbers to Trump.

Trump's camp didn't say that they were referring to those numbers, and their claim doesn't match those numbers. The fact checkers actually go out of their way to find another labor statistic that approximately matches Trump's numbers (but which is not the unemployment rate), which seems to be the entire basis for rating the claim only Mostly False false rather than simply False, which is an example of benefit of the doubt.

> It's not "benefit of the doubt" when you contact someone and ask for the basis of their claim and they provide it,

Of course not, but now that they knew about that data, it should have also applied equally to Trump's claim. They are trying to see who can debate it better or provide better numbers. They should use the same set of facts to judge everybody.

The fact checkers are incredibly biased. There is a Repub one out there does does the same. If you think they two aren't even slightly biased, then you probably drank too much of the Cool-Aid already.

> but now that they knew about that data, it should have also applied equally to Trump's claim.

But Trump's claimed number doesn't match that data either (nor has Trump claimed that was the intended referent.) It can be interpreted to approximately match a different potential measure of workforce participation (that is even less like an unemployment rate as usually understood) that the fact checkers went out of their way to find, which is why it is rated merely "Mostly False", rather than flatly False.

> They should use the same set of facts to judge everybody.

But the two claims aren't the same, and do not have the same relation to the facts.

They should have used that institutes data for both. I'm saying he should have been adjudicated as correct, but was abut the same as Sanders, may slightly more hyperbolic, but not enough to dock him that much.

At the very least they should have made a reference to the study that Sander's team brought up and said why it didn't apply.

The biggest issue is that this isn't an isolated incident. That fact checker routinely errors on the Dem side. If this was one of many more balanced cases this would be different.

They are not only similar but indiscernible. There is no single definition of "unemployment", therefore there is no single unemployment percentage.

"Similar" is a weasel word; they were notably different numbers, and one had an identified basis which allowed showed a clear connection to the truth despite inaccurate terminology which resulted in it getting not rated True.

And the rulings were not at the two extremes, they were two points apart on the six-level scale, with one one position removed from the positive extreme and the other two positions removed from the negative extreme.

They're both really just summaries of Tara Sinclair's view, aren't they? Not that it makes them more objective for showing someone else's bias, but the interpretation of the metric selection is essentially a summary of her statements in both.

I looked into this when it made the rounds and I don't agree that it is evidence of intentional bias on the part of Politifact.

I do believe it is a good case study that exposes the weak points of their approach toward condensing a quote down to a true/false rating. I think they would do much better to decompose a quote into separate claims and fact check those separately.

The Sanders quote that was fact checked can be read as making three separate claims, and Politifact discusses each before lumping them together for the true/false stamp:

  1. The "real unemployment rate for young people" is not the same as the typically accepted rate
  2. unemployment for blacks >> unemployment for Hispanics > unemployment for whites
  3. The "real unemployment rate for young people" is 51%, 36%, and 33% for those groups, respectively

Sanders doesn't explicitly spell out (1), but he at least uses the qualifier "real" instead of simply saying "the unemployment rate", and his campaign later clarified what he meant. (2) is supported by the study his campaign supplied. The numbers in (3) match the study his campaign supplied.

In contrast, the Trump quote is: "If you look at what’s going on in this country, African-American youth is an example: 59 percent unemployment rate; 59 percent"

and Politifact is only fact checking the "59 percent unemployment rate for black youths". Trump's campaign did not respond, but the Politifact research found that "it appears likely it comes from a computation of all 16- to 24-year-old blacks who aren’t working and may not even want a job, including high school and college students." Since the fact-check is focused on the statistic itself, this tips it into the false category.

Of course, a potential criticism is that Politifact should be fact-checking the higher-level claim that Trump is making. But from this quote, it's very difficult to tell exactly what this claim is. Trump tends to use words to paint emotional pictures that are intended to resonate with a broad audience, rather than making a chain of argument to support specific claims. How do you fact-check a claim about "what's going on in this country"? Any interpretation of such vague phrasing is going to rely as much on the personal experience of the audience as it will on the actual quote.

So I do agree that this exposes a weakness of Politifact. But in this case the Sanders and Trump claims are not equivalent, and Politifact has a clear chain of reasoning behind their ratings.

Another, higher-level potential criticism is that Politifact, as an organization with limited resources, has to make editorial decisions and clearly can't fact-check every public statement made by a candidate. I don't know how they decide which quotes to fact-check, so I don't rely on them to provide a complete account of what is happening in the world. But I'm happy they're able to dig in deeply in some cases where clarification is needed.

The lesson to take away is to read articles from fact-checking organizations for what they're good at. Read and understand their cited evidence and reasoning, not just the true/false stamp.