On a new approach to modeling responses to “rolling panels” such as Annenberg’s. Here’s the link in case you missed it:
I’ve been at the Library of Congress for two days now, capturing a selection of archived campaign websites that are only available on-site (because of legal reasons) for a dataset I’m putting together as part of my dissertation. And in that time I’ve looked at literally hundreds of pages of content, and it’s a generally-maddening experience—if a 30-second ad full of BS gets you mad, try reading that stuff in 8-hour shifts. But I think what gets me most is the desecration of logical argument practiced by candidates who try to persuade voters of the correctness of their positions.
“In West Virginia, we have the highest per capita gun ownership in the country with also one of the lowest – if not the lowest – crime rates. This is proof that disarming American citizens actually makes them less safe and secure, and I will fight to preserve our Second Amendment rights and freedoms.” (John Raese, Republican Senate Candidate for WV, 2006)
This sort of thing makes me almost yell at my laptop screen, and in the LOC’s Main Reading Room (which is absolutely beautiful, btw—see the previous post) I imagine such things are greatly frowned upon. But for the love of science, is there anyway we can get No Child Left Behind (or its successor) to require an understanding of the difference between correlation and causation in order for a student to graduate 8th grade? Because the saddest thing isn’t that so many politicians make these kinds of false arguments, it’s that so few voters can tell that they’re wrong.
One thing that’s been a consistent theme in talk about the presidential election next year is that the main motivator for Republican voters seems to be dislike of Obama, rather than enthusiasm about any of the GOP candidates. And today, Newt Gingrich predicted big GOP pick-ups in the House and Senate again next year. Which got me wondering: we know that anti-incumbent moods can swing a midterm election, but does the same hold for congressional races in presidential election years?
My intuition is to say no. Since midterms have much lower turnout on average, a strong anti-incumbent mood among out-party voters can provide a strong motivation in otherwise uninspiring races (i.e., when few voters have strong feelings about the individual candidates). This could thus sharply skew the relative turnout between each party’s base in a midterm. In presidential years, though, the presidential contest provides a positive motivation for in-party voters to turn out; if out-party voters have only a negative motivation (voting against the incumbent, but not necessarily for their nominee), the relative turnout difference should be much smaller.
(Note that I’m assuming overall preferences are held constant; in other words, that voters aren’t changing their minds about whom to vote for, they only choose whether or not to show up/donate/mobilize others and so forth.)
But this is just a conjecture, and I’m sure somebody’s already studied it. If so, can anyone recommend some reading on this topic? I’ve got some original survey data on the strength of each candidate’s coattails in 2008, and am thinking of combining that data with a broader study of the effects of presidential candidates on congressional races. As such, any suggestions about where to start re: previous work would be put to good use.
(Since I don’t keep up a comments section on this site, you can email me at email@example.com. Thanks!)
Quick note before departing Phoenix: a few people have asked for copies of the paper I presented yesterday (which won AAPOR’s 2011 student paper award), and in case you’re one of them, you can download it here:
It’s a bit long, so I’m currently working on a new version which focuses specifically on informing effects in greater detail (the other results may be used in a subsequent paper). Also, I’ll need a new name for it–the most recent APSR (or was it AJPS?) had an article with the same main title–so suggestions as well as other feedback are welcome.