To give you a little personal background, Becky and I basically grew up together in the profession, so in a sense we are academic siblings. We met in the early 80s when both of us were assistant professors. Actually, I think she was simultaneously a graduate student and an assistant professor when we met – in New Orleans, of course. Or as I called it at the time, New OrLEEEENS, which clearly grated on Becky’s ears, and of course resulted in a no-nonsense lecture on pronunciation. This was the first of many meetings at the same circle of conferences we both attended, notably ESA and Public Choice, which were pretty small at the time so you got to know almost everyone who attended. I don’t keep good records, but it’s a good bet that our paths crossed at conferences at least once a year since that first meeting in NO. The most recent one was at Columbia University less than a year ago. It’s hard to believe that it was also our last. Our paths and interests overlapped throughout our careers, not just by working on similar research problems, but also having similar interests as normal people. We both had a travel bug – although in the last five years or so I was badly outmatched in that dimension – and had similar culinary tastes, for example sharing an appetite for oysters, which we consumed together on many occasions, ranging from Acme in New Orleans to Aquagrill in Soho. Of course I benefited from many stimulating discussions with her over the years, but especially so at Princeton in 2004-5 when she was a Fellow at the Center for the Study of Democratic Politics in Princeton.
Let me focus for a moment on her work on voter turnout and strategic models of political participation. This includes at least two chapters of her dissertation, which launched her career. Voter turnout was naturally a central subject of our earliest conversations, as both of us were working on strategic models of voter turnout (she on her own and I, not on my own, but together a brilliant senior coauthor Howard Rosenthal). Her pair of papers, “Groups in Rational Turnout Models” (AJPS 1991) and “A Group Majority Model of Public Good Provision” develop a solution to the paradox of not voting, once dubbed by Mo Fiorina as “the paradox that ate rational choice theory” (mistakenly, I might add). This does not mean that the free rider problem is fully solved and all members of the group will turn out in her voting equilibrium, rather is related to Harsanyi’s notion of group utilitarianism. Groups turn out so as to maximize their total benefits of voting, net of costs, given the turnout rates of the other groups. It’s a Nash equilibrium between groups, rather than individuals. An equilibrium exists with positive turnout. She then goes on to extend this “partial equilibrium” analysis to the “general equilibrium” case where candidate locations are endogenous. These are really great papers.
Becky and I finally collaborated about 15 years ago on several joint papers related to strategic abstention and the swing voter’s curse as a result of our fortuitous convergence in Princeton. Not long before that she had studied with Ken Williams the question of how sequential and simultaneous voting procedures can make a difference when voters have information asymmetries. Coincidentally, Marco Battaglini, also at Princeton at the time, had just published a pure theory paper on the same subject. That paper identified conditions under which simultaneous or sequential voting led to better outcomes when voters were imperfectly informed and abstention was allowed. The three of us designed an experiment to test those predictions of the model. Of course, it was a pretty complicated game and not all of the details of the predictions were supported, but the qualitative predictions and comparative statics held up surprisingly well. This motivated us to embark on the next project, which was a test of the swing voter’s curse model of Feddersen and Pesendorfer. That is, poorly informed voters should strategically abstain because of the risk that their uninformed vote could swing the election to an inferior candidate. Or in case there is expected to be a strong partisan tide in one direction, the voter may strategically vote against the partisan tide, even with no information. This experiment worked like a charm. When we showed the initial results to Wolfgang Pesendorfer that strongly supported his theory, he was amazed, saying that the results looked “fishy”. Well, the results held up with replication, even with much larger committees.
Becky wrote quite a few other papers on turnout and of course many other subjects, but I think it’s time to wrap up. But before doing so, let me add that these “other papers” illustrate her amazing methodological breadth. Yes, she made seminal contributions to formal theory and experimental political science. But she also made significant contributions to political science employing other empirical methodologies as well, including large scale surveys and field data. In fact, her most recent paper on voter turnout looked at the effect of exit polls on turnout based on a natural experiment using data from voting in French Overseas Territories. To say that her methodological breadth was remarkable is an understatement.
We have lost a major player in our discipline, but at the same time we can be grateful for the legacy she has left us.
– Thomas Palfrey, Cal Tech