Politico's Ben Smith wrote a long article about America's fact-checking industry (PolitiFact, FactCheck, etc.), and he does a good job of describing the tug-of-war between these sites and political spin-meisters, as well as addressing some of the inherent weaknesses in the criteria they use to find the line between truth and falsehood. But there's one very important question missing from the article: Does fact-checking work?
By "work," I mean a couple of things. The first is, does it change politicians' behavior? Is a candidate who gets called out for a lie in a fact check going to stop saying it? I posed that question to Bill Adair, who runs PolitiFact, when I interviewed him for a story about this topic that never actually found its way into print (long story). Adair's response was that changing politicians' behavior isn't his job; he and his organization put their best assessment of the facts on the record, and then whatever happens next is basically out of their hands.
One could design a study to determine whether lies are less likely to be repeated once the fact checkers have judged them harshly, but no one that I know of has done it. The consensus from people I've talked to about this seems to be that it depends on who the liar is. The narrower their constituency, the more likely they are to continue on unashamed even after being called out for lying. Michele Bachmann doesn't really care if PolitiFact says one of her claims is bogus. Mitt Romney, on the other hand, is more concerned about his reputation and therefore more likely to stop saying something once it has been called a lie.
But it's important to remember that keeping politicians honest—not just providing a record—was really the reason fact-checking got started. As Brooks Jackson of FactCheck.org relates in the piece, this whole enterprise was born out of the 1988 campaign, in which reporters felt that they had been manipulated by George H.W. Bush's campaign into not only passing on a lot of falsehoods, but also letting the campaign be dominated by troubling issues like the Willie Horton case.
(A brief digression for a fact-check of my own. Smith's article contains this sentence: "'After [George H.W. Bush’s 1988 ad featuring] Willie Horton, there was a lot of hand-wringing by Democrats,' said Jackson..." Don't know if this is Ben's fault or Brooks', but the Bush campaign never actually aired an ad featuring Willie Horton; the Horton ad was aired by a shadowy outside group. The Bush campaign did, however, air ads about crime that were almost certainly meant to evoke Horton, who was often featured on the news and whom Bush mentioned repeatedly on the stump.)
The initial response to 1988 was a whole lot of "adwatches," in which television ads are evaluated by news organizations for their veracity. These are far less prominent now than they were from about 1992 to 2000, and for a while there was a lively debate among communication scholars about whether they actually persuade people to disbelieve the misleading claims in the ads they evaluate. Which brings us to the second part of the "Does fact-checking work" question: Do they persuade the people who read them? After all, fact-checkers depend on nonpartisan credibility, the idea that these neutral arbiters can settle the competing claims and tell you who's really telling the truth, even if it may be that your team is the one lying. But a growing body of research shows that people are remarkably resistant to accepting that the things they believe are actually false, even when the people telling them so appear to be neutral. We are extremely resourceful at finding ways to ignore, argue against, discount, and forget anything that challenges our beliefs. Information, furthermore, is inextricably entwined with the source it comes from. When you hear a politician or media figure you trust tell you something, then a fact-checker tells you it's untrue, your impulse is going to be to side with the person you trust, particularly in an environment where facts are constantly in dispute. In some cases, a correction can actually make some partisans hold even tighter to their false beliefs (see here). For example, you can tell conservatives that the Affordable Care Act contains no "death panels," but many of them will just refuse to believe it.
This goes to the heart of whether fact-checking is worthwhile. After all, if everyone who reads your fact-check disbelieves it unless it says their team is right and the other team is wrong, there's not much point to it. But nobody knows exactly what to do about that problem, so it isn't surprising that the fact-checkers themselves prefer not to think about it.