Chris Blattman links to this paper by Gwyneth McClendon, a Ph.D. student in political science at Princeton. She writes:
bq. In this paper, I discuss a field experiment in which 1294 South African local politicians were each contacted by a fictitious constituent, whose name signaled either co-ethnicity with the politician or out-group membership. The constituent raised a concern either about a public goods issue that is prioritized equally across South African ethnic groups or about one that is more ethnically divisive. I found that politicians of all ethnic groups were more likely to respond to co-ethnic constituents than to non-ethnic constituents (even when the non-ethnic constituents were co-partisans) and were much more likely to respond to a unifying issue than to a divisive issue. The paper concludes by discussing whether the findings provide an argument for descriptive representation.
A similar finding has emerged in American state legislatures. From a paper by Daniel Butler and David Broockman that got some notice a little while back:
bq. We use a field experiment to investigate whether race affects how responsive state legislators are to requests for help with registering to vote. In an email sent to each legislator, we randomized whether a putatively black or white alias was used and whether the email signaled the sender’s partisan preference. Overall, we find that putatively black requests receive fewer replies. We explore two potential explanations for this discrimination: strategic partisan behavior and the legislators’ own race. We find that the putatively black alias continues to be differentially treated even when the emails signal partisanship, indicating that strategic considerations cannot completely explain the observed differential treatment. Further analysis reveals that white legislators of both parties exhibit similar levels of discrimination against the black alias. Minority legislators do the opposite, responding more frequently to the black alias.
Blattman wonders about the ethics of this:
bq. …targeting politicians experimentally is getting increasingly common, and there are clearly important things to learn, but what’s the basis for human subjects approval (and ethical research) in these kinds of experiments? There can’t be consent or it would negate the experiment. These are public figures, so do we get to forego consent?…I don’t see much issue with this specific paper [JMS: McClendon’s], but still, my gut tingles a bit with fictitious constituents.
For one critical view, see Karl Kurtz at The Thicket — the blog of the National Council of State Legislators:
bq. I think of it more as treating legislators as lab rats: Performing an experiment by telling them a lie and then measuring how they respond. Ring a bell and see if they salivate.
bq. It’s a little bit like the FBI creating a crime and then trying to ensnare public officials in it. Only the political scientists’ deception doesn’t have an FBI sting’s justification of uncovering corruption. The only saving grace is that this study reports only group behavior, not individual responses, so there’s no “gotcha.” There is, however, a profound disrespect for the work that legislators do, the enormous demands many of them work under, and the idea of public service.
I’m more sanguine than Kurtz. All experiments involve deception. Obviously, you can’t tell subjects in advance what the treatment is and expect the treatment to work. I think the deception here was quite mild. I also think legislators are fair game as subjects. Answering a single letter or email doesn’t really detract from their busy lives. Nor does it disrespect their work. In fact, both of these experiments are profoundly respectful of the task of representing constituents. The question is simply whether legislators do so fairly. I think that’s a reasonable question to investigate in this fashion.