It seems to come up a lot in the discussions of Cato. Cato has worked really hard to cultivate a favorable/neutral reputation with liberal intellectuals. "We're not the religious, homophobic, xenophobic right-wing crazies. We're the anti-war, anti-Fed, anti-pot-law right-wing crazies."
I would prefer to have a favorable/neutral reputation with liberal intellectuals myself. But I am wary of being manipulated because of that desire.
I am not a big believer in the affiliation heuristic. I think that people say that they use the affiliation heuristic a lot more than they actually do.
For example, when right-wingers say we need to examine President Obama's past affiliations with radicals, are they sincerely using those affiliations to evaluate his policy stance? If so, that would mean that they do not have enough information about President Obama's current policies to evaluate them, and instead they need to draw inferences based on his past affiliations. Really?
Similarly, when Ezra Klein says that he respects the Cato Institute, is he actually using the affiliation heuristic when he reads Cato policy papers? If so, that would mean that he is incapable of evaluating policy papers on their merits, but instead he must draw inferences based on scholarly affiliations. If you believe that, then I have a bridge across the Charles River I would like to sell you.
Regardless of the affiliation heuristic's actual merits (and I am pretty negative on it), I am offering a meta-heuristic. I am saying that when you see someone invoking the affiliation heuristic, your standard presumption should be that they are being hypocritical and manipulative. (Is that Robin Hanson whispering that this is an uninteresting statement, because, hypocritical and manipulative is what humans are all about?)
Readers are welcome to submit instances in which they have observed someone sincerely invoking the affiliation heuristic. Examples of the form "I used it sincerely when..." will be disregarded, because with p=1 you are not telling the truth. Examples of the form "Kling used the affiliation heuristic when..." will be particularly interesting, and I bet it turns out that they prove my point.