Turnout for What?: An altogether different concern with field experiments


To the degree that Political Science Twitter can blow up, it blew up this week with the controversy over a field experiment run by professors at Stanford and Dartmouth in Montana (For the latest in the debate, see John Patty; for overview, Derek Willis in NYT). I don’t have anything to add to the ethics in field experiments discussion, although I am sympathetic to the accused profs. (If you would like a totally ethical look at how information increases participation in races like these, see my MPSA paper from last year!)

My concern with this episode is that it illustrates a problem with the literature on turnout studies. They’ve gotten too good. One only needs to look around at the mail parties send to voters to see that voter shaming (like in Alaska), and possibly intimidation, is alive and well. These are techniques that were first tested by field experiments.

(Look, it even happened to me in 2012)


The combination of these techniques and the robust field office infrastructure of the Obama campaign means that voters in competitive races are being inundated with campaign contact in the last few days before election day. 

The problem is that increasing effectiveness of these techniques is approaching the ceiling. The concept from economics of diminishing marginal returns applies. Consider television ads, if you’ve seen 50 ads the 51st ad has little impact on you. With social pressure, the first postcard you get may make a difference. The tenth postcard that compares your voting record against your neighbors has little impact. If anything the Montana case shows that it may be putting voters off.

So I doubt that these techniques are going to move the needle much in the future. But they're not going away. American politics is a zero-sum game; so if one side is doing it, the other needs to be doing it too. Therefore, campaigns (and researchers) will surely be looking for the next new innovation that gives them an edge.

Ultimately, that’s why field experiments need to survive this current scandal. The lasting contribution of the turnout field experiments is not shaming techniques or that in-person contact is better than mail. The real innovation is it demonstrated the power of a systematic method of randomly assigning some voters to receive a treatment and comparing them to others. When I worked on campaigns we did everything we could think of to influence voters and just hoped it work. This political science research program has shown how to do a much better job of that. That's why it needs to survive this onslaught from critics.


Popular posts from this blog

10 Tips for Going on the Political Science Job Market

Image
It’s August, and academics are procrastinating so I’ve seen lots of advice on the job market. A lot of this advice seems good, but it usually comes from tenured professors who last went on the market before the advent of Twitter. So here are ten quick thoughts on my two recent trips to the academic job market: My qualifications for dispensing advice include spending many hours in the belly of the beast. Long term planning. One of the most sobering pieces of advice I got on the market came from Mike Horowitz at Penn, who emphasized to a group of us: “The market is a multi-year process.” This turned out to be the case for me, and the majority of people I have seen go through the market. At Penn we had 5 years of guaranteed funding, so planning for the market should begin in earnest in the second half of the fourth year. Once you lose that guaranteed funding, you’ll need to spend time on finding contingent employment, and anytime that is spent not on research you will not get
Read more

2022 Midterm Projection: Fundamentally Republican

Image
I think this is going to be an ugly election cycle for the Democrats. We can expect losses for the president's party in Congress in their first midterm election, and Joe Biden has not done anything to change my expectation. This is compounded by the fact that the economy seems to be the most pressing concern for voters, and inflation is a very difficult issue for policymakers to make go away or minimize. Since inflation is effectively making money less valuable, people that have a lot of money hate it , and even though unemployment is at a record low right, and many people in lower tax brackets have money in their pockets, it doesn't seem to go as far when you have to pay large bills for gas, heating your home or buying groceries.  So I believe we'll see the US House of Representatives swing to Republicans, probably with about 15-20 Democratic seat losses. Oddly, this is not actually that high of a number (Trump's party lost 41 in his only midterm, Obama lost 63 in his
Read more