I like think tanks. Some of my best friends work for think tanks. I think they do a lot of good work. But the political policy ones do their best work when they are trying to decide policy within a movement; that's when you start seeing real innovative work. They are also very good at providing critiques of academic work in their areas of interest.

When they turn to fighting outsiders over, say, the minimum wage, the quality of their work sharply degrades. They have limited ability to change their policy position, because the donors will revolt; if they can't get an answer the donors will like, they don't ask the questions. They also only hire scholars who agree with them. That already biases their work, but then you have to contend with the groupthink problem: when everyone at the office agrees with you that your opponents are idiots, and you socialize mostly with other people in the movement, your thinking gets a tad lazy.

So if the only support for your positions comes from movement think tanks (plus maybe a few marginal academics), your position is probably extremely weak. Indeed, if someone from the other side were pulling the same trick, you would be the first to notice this. Independent studies commissioned by think tanks are especially suspect. You can't check their calculations, and survey design is easily manipulable to get the answer you want.

That's why I rarely grab, say, a Heritage or CEI study on the minimum wage and offer that as evidence for my claims. As it happens, on this issue I broadly agree with them. But even if I were willing to vouch for their numbers, it's pointless, because no one who disagrees with me would accept them. So I go to the BLS, the Census Bureau, the CBO, the JEC, the GAO, or an academic study instead. In cases where I can check some of their numbers, I'll use it as a secondary source. But it's never my primary source for a policy position.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.