Reading Homestyle with my Congress class right now. On pg. 17, a classic quote about congressional campaigns:
“Seventy-Five percent of all the money we spend in a campaign is wasted. But we don’t know which 75%.”
This quote will ring true to anyone who’s ever worked on a campaign. A friend of mine who has worked hundreds of campaigns actually thinks it drastically underestimates the wasted money.”Switch the 7 for a 9,” he says. “That rings more true.”
I recall one local campaign I worked on in New Haven, we had this real Carville-wannabee as campaign director. On election day, he was in such a frenzy that after we finished calling all the registered partisans to remind them to vote, he started yelling at us to call all of the people not registered to either party. I said to him, “we have no evidence that the independents are breaking in our favor; in fact, we have a lot of reasons to believe they are breaking against us. If we are correct that they are breaking for us, calling them might marginally increase the probability of an already-likely victory. If we are wrong, calling them might significantly increase the probability of snatching defeat from the jaws of victory.” He just stared at me for about 5 seconds. Then he barked quite loudly, “Trust me. Just call them.” And then he walked away.
It didn’t matter. We got creamed in that election, almost 2 to 1. But that only proved that the independents had broken against us, and that I probably got someone to vote who otherwise wouldn’t have…for the other guy!
Wasted money indeed. The crazy truth is that the vast majority of campaigns employ a kitchen-sink strategy of spending money any way that it could plausibly be effective, without any hard evidence that such methods have any effect, and certainly no idea of the magnitude of the effect. Just like that kooky campaign director in New Haven having us call indys, occasionally you can’t even be certain that the strategy is not having a negative effect. And there’s a horrible feedback loop: winning campaigns are seen to be models of good strategy, and losing campaigns are seen as models of bad strategy. Marginal effects? Good luck raising that issue.
I bring this up because Sasha Issenberg has written a new book, The Victory Lab, which details how Rick Perry brought on a host of political scientists to do randomized field experiments with his campaign for governor in ’06.
This is interesting in itself; double interesting to me because the political scientists who more or less started this movement (and presumably worked with Perry) were Yale political science faculty, Don Green and Alan Gerber, when I was in graduate school. Their methods aren’t complicated: they simply run controlled field experiments using random assignment, in order to understand what campaign techniques work, and more importantly, how cost-efficient different campaign strategies are. One of their first major pieces compared mailings, phone calls, and door-to-door personal contact, randomized in a get-out-the-vote (GOTV) drive in New Haven. After random assignment of citizens to the treatment groups (including a control group), they apply the treatments and then after the election check the effect on voter turnout, using public records. Instrumental variable regression can be run using the intent-to-treat, and a cost-per-vote can be precisely calculated. In the case of this experiment, door-to-door blew away phone calls and mailers, despite being on the surface far more expensive.
It’s a cool sub-field in behavioral American politics, and certainly one with significant real-world implications. And the extensions are endless; one of the coolest pieces I’ve read studied how social pressure affected GOTV — a treatment group was sent a mailer indicating that whether or not they voted would be revealed to their neighbors. (It significantly increased turnout). One drawback to the line of research is that it’s quite expensive; without start-up cash or a serious hook-up from a campaign, it’s not the kind of research you can just decide you want to do. But that’s one thing that makes the connection with Perry so interesting. If he’s the GOP nominee and allows continued experiments to be run during his national campaign, it could be a massive wellspring of research across a sample and a stakes that has never been subject to these methods before. Very exciting.
Still, there are limits. As another buddy of mine says, all of this research may someday make it possible for campaigns to only waste 2/3 of all the money they collect.
