Political Scientists' Voting: Predictions and Methods

A week and a half ago, I posted a brief, frustrated reflection on my failure to find any good research on the rates at which political scientists vote in public elections. After several more search attempts, I've given up. As far as I can see, no one has explored this issue since a few studies in the 1960s and 1970s -- studies so problematic as to be utterly useless.

I've informally asked a number of people to guess what Josh Rust and I will find when we analyze the data. Everyone except the political scientists said that they suspect we'll find that political scientists vote more often than other professors. The political scientists, however, were cagey. A couple mentioned a minority view in political science that voting is for suckers: Your vote never makes a difference, so voting is a waste of time. Yet one of these same professors said that he himself has voted in every single election, down to the tiniest little runoff, since the turn of the century.

So here's my prediction: Political scientists will have a more broadly spread distribution than other professors -- there will be more at the extreme of voting in almost every election but there will also be many who vote rarely or never. On average, though, I predict, the political scientists will vote more. Compared to the average non-political-science professor, the average political scientist will be more informed about elections, more invested in and interested in the outcomes, and more likely to have publicly embraced the view that one should vote.

Well, we'll see! Here's what Josh and I plan to do:

From university websites, we'll gather names of philosophers, political scientists, and a sample of professors in other fields. We'll then look for these names on voting records that have been provided to us by several states, calculating a rate of votes per year for each individual since that individual's first recorded vote in the state.

There are two main weaknesses in this method, and I especially welcome readers' reflections or suggestions about these. First, since we don't have street addresses for the professors in question, we will not be able to disambiguate between voters with identical names. If there are four John Millers who live within commuting distance of So-and-So College, we won't know which one is the professor, so we will have to discard the data. And second, if no voting record matches the professor's name, we will not know whether that professor is registered under a different name, registered in a different locale, a non-citizen, a felon, or simply a non-voter. So we'll have to exclude those professors too.

Because of these difficulties, we won't be able to reach conclusions about absolute rates of voting participation among political scientists, just comparative rates -- more or less than professors in other departments. But will these difficulties undermine our ability even to draw that conclusion? Although I don't see any reason to think there will be large differences in the rates at which professors in different departments are registered under different names or in different locales, there is reason to suspect that different departments may have different rates of common names and of non-citizens. But hopefully we can keep those confounds under control: We'll have an exact count of the common-name professors in the different departments, so we can attempt analyses that account for that; and hopefully we can estimate the rates of non-citizenship in departments by accessing c.v.'s or biographies of our non-voting professors where possible and by looking at general data on the citizenship of professors.

What do you think?