Many thanks to Timothy Groseclose for confirming that I understand the basic methodology of the paper underlying his book. Here’s my follow-up question, if I may: Am I right that the paper assumes that politicians cite think tanks for the same basic reasons that journalists cite think tanks? Put another way, am I right that the paper assumes that you can compare journalist-citations and politician-citations because they measure the same thing?
If I’m right, that assumption seems problematic to me. Here’s why. Journalists see their goal as informing their audiences. They cite think tanks to inform their audiences about what is happening. They want to get the reasonable range of views on a topic, so they cite think tanks that reflect what they see as the reasonable range of views.
Politicians cite think tanks for a different reason. Politicians make speeches and issue press releases to persuade rather than to inform. As a rule, politicians will cite think tanks if and only if they can find a think tank that said something that supports the politician’s view. To maximize the persuasiveness of the argument, the politician will cite the think tank that is the most respectable source that said something echoing their view. And of course think tanks occasionally exist to fill that role: If a powerful person or group needs someone respectable-sounding to say X, they will fund a person who sits in an office as the Executive Director of The Center for American Goodness to reliably say X to whoever will listen.
If I’m right about these differences, then comparing journalist-citations and politician-citations seems a bit like comparing apples and oranges. Of course, I would imagine the two correlate somewhat. It’s easy to see why. If a journalist has very liberal views, they will probably see the range of reasonable opinion as tilting to the left; and if a politician is very liberal, they will probably only find liberal think tanks to support their views. So a correlation makes sense. But as I understand it, the paper is not assuming that the two correlate. Rather, the paper assumes that the two measure exactly the same thing, and then seeks to show that that the media is biased because journalist-citations don’t exactly match the citation practices of a centrist politician. That seems like a weak assumption given the very different reasons journalists and politicians cite think tanks.
Of course, none of this means that the media isn’t biased. But unless I’m missing something — which is certainly possible! — the different reasons journalists and politicians cite think tanks leaves me unconvinced that comparing the citation practices generates “scientific proof” of media bias.