Skip to content
 

Timing is everything!

A colleague emailed me with a question about the methods used by Groseclose and Milyo in their study of media bias.

Before getting to the question, I just wanted to comment that Groseclose has had really bad timing with this project.

First off, his article came out in 2005 when everybody was hating Bush. Even the Republicans who reelected him weren’t thrilled with the guy. Then his book came out in 2011. If his book had come out a year ago, that would’ve been perfect: the 2010 elections coming up, lots of anger at the Democrats and Obama, no peer-reviewed criticisms of his work, etc. Instead he waited until 2011, and then look what he got:

– Republicans feel they have a chance at winning in 2012 so they’re more interested in fighting and less interested in complaining.

– John Gasper shoots down Groseclose/Milyo in the Quarterly Journal of Political Science. That’s gotta hurt. (Until this point, Groseclose could respond to attacks by saying he was waiting until a criticism appeared in a peer-reviewed journal.)

– The biggest news in media bias is the story of Rupert Murdoch buying the London police force. Not the best time for a book claiming that Fox News is “centrist, and possibly even left-leaning.”

Here’s my colleague’s question:

Here’s the guts of their model:

Define xi as the average adjusted ADA score of the ith member of Congress. Given that the member cites a think tank, we assume that the utility that he or she receives from citing the jth think tank is

(1)    aj + bjxi + eij

The parameter bj indicates the ideology of the think tank. Note that if xi is large (i.e., the legislator is liberal), then the legislator receives more utility from citing the think tank if bj is large. The parameter aj represents a sort of “valence” factor (as political scientists use the term) for the think tank.

Since the “ideology” values range from 0 (the most conservative) to 100 (the most liberal), and since an individual’s ideology is multiplied by a thinktank’s ideology to determine the contribution of ideological affinityto the “utility” that he or she derives from citing it, it trivially follows that a completely conservative individual (politician or journalist) pays no attention whatever to think-tank ideology, but chooses purely on the basis of “valence”.

This seems to me [my colleague who sent me the email] to be gratuitously false to fact, and likely to distort other aspects of the model fit, since all or nearly all the choices of conservative congresscritters need to be accounted for using the valence terms.

Regarding the specific question, I haven’t looked at all the details but my guess is that Groseclose is using a standard ideal-point model and that the reparameterization (0 to 100 rather than, say, -1 to +1) won’t affect his substantive conclusions.  My guess is that if you shake out all the algebra you get back to the standard model.

I also have a more general comment about the presentation of the model. Groseclose is a well respected scholar but his strength is particularly in theory rather than empirical methods. Theoretical modelers tend to distrust models of data. What they want is a model of behavior. The result (from the perspective of an empirical modeler such as myself) is that they add unnecessary layers of implausible psychology on to empirical models that might very well stand up just fine on their own.

In particular, I don’t think it makes a lot of sense to say that a congressmember has a “utility” from citing an advocacy organization. I don’t know much about the Congressional Record but my guess is that sometimes a congressmember has a point he or she wants to make and finds a quote to support it, other times a lobbyist or advocacy organization gives the congressmember some material to insert directly into the Record, other times there is a particular study that is relevant to a debate. In any case, I don’t see this having anything to do with utility (except in the tautological sense that whatever you do, you must have utility for, or you wouldn’t do it).

Also, as I’ve written elsewhere, I don’t like Groseclose/Milyo’s use of the term “think tanks” for organizations that are not, in fact, think thanks. I’d prefer the term “research and advocacy organizations” or even simply “advocacy organizations” (as even the research organizations tend to have policy positions that they’re pushing).

4 Comments

  1. Brett Keller says:

    FYI, something is off with your RSS feed. I normally get two copies of every post, but got three copies of this one.

  2. John Mashey says:

    re: “think tanks”
    I’ve spent a lot of time studying these thing and from what I can see:
    1) Some actually do research.
    2) Some do advocacy and are straightforward about it
    3) Some are combined PR agency/lobbying entities who take money from wherever they get it, most often companies that privatize profits and socialize losses and risks. Then the thinktank can be an independent voice and can talk to legislators without being “lobbyist” and following those rules.
    Quite a few thinktanks have taken tobacco money, for example, as per the wonderful Tobacco Archives.

    So, what are the right names for these things?
    I don’t know if you’ve ever studied the rules around 501(c)3 entities, of which the public charity category usually claim to do research and education (as on public policy, or “sound science”)
    – donations are charitable deductions
    – any investment profits are tax-free.
    – there are limits on lobbying and they have to say what effort/$ they spend
    – Support has to be from the public, rather than just a few big donors, with somewhat messy rules.
    – It is worth a lot to have 501(c)3 status.

    Sadly, in practice, there can be a lot of murky money flows, since they need not name donors and sometimes donate money to unnamed entities.

    But I am pleased to learn that Fox news was centrist/left-leaning.

  3. his article came out in 2005 when everybody was hating Bush. Even the Republicans who reelected him weren’t thrilled with the guy

    I don’t think that’s true. Republicans venerated him from beginning to end, though it is true that he alienated non-Republicans.

    His approval didn’t dip below 80% among Republicans until 2006 or very late in 2005, and was always higher among self-described “conservative Republicans”. (I link to a bunch of places to substantiate that in my post here).

    Further problems with the study’s methodology were examined a while back here:

    For example, the ACLU and the NRA have very similar scores–49.8 and 45.9–both centrist, though slightly to the right. The authors take note of the ACLU’s position, and explain why–its opposition to campaign finance restrictions was cited repeatedly by Congressional opponents:

    “In fact, slightly more than one-eight of all ACLU citations in Congress were due to one person alone, Mitch McConnell (R.-KY), perhaps the chief critic of McCain-Feingold.” It does not seem to occur to the authors that this may be indicative of a much more pervasive problem with their basic, untested assumption that politicians cite sources for ideological reasons, rather than pragmatic ones. …

    There’s also the little matter of calling organizations like the NRA and the ACLU “think tanks.” They’re not. …

    Terminology is always tough for organizations described as “think tanks.” For Heritage and AEI, how about “Republican PR firms”.

  4. idiot says:

    According to Google’s Define, a think tank is “a body of experts providing advice and ideas on specific political or economic problems”. Granted, it’s the only entity I find that use that definition, but at least one dictionary agrees with Groseclose and Milyo (because I’m willing to believe that the NRA, NRA, ACLU, Heritage, AEI, and any other alphabet soup organization are staffed with people that call themselves experts).

    Considering how many people frequently use the word “think tank” to refer to outright political activity and how “research” can be poorly defined, I think it’s far better to just assume Google’s Define is right. Words can change meaning, after all.