Skip to content

The Great Society, Reagan’s revolution, and generations of presidential voting


Continuing our walk through the unpublished papers list:

This one’s with Yair and Jonathan:

We build a model of American presidential voting in which the cumulative impression left by political events determines the preferences of voters. This impression varies by voter, depending on their age at the time the events took place. We find the Gallup presidential approval rating time series reflects the major events that influence voter preferences, with the most influential occurring during a voter’s teenage and early adult years. The fitted model is predictive. It explains more than ninety percent of the variation in voting trends over the last half-century. The fitted model is also interpretable. It divides presidential voters into five main generations: New Deal Democrats, Eisenhower Republicans, 1960s Liberals, Reagan Conservatives, and Millennials. We present each generation in context of the political events that shaped its preferences, beginning in 1940 and ending with the 2016 election.

It’s currently in umpteenth round of revision and almost done. An earlier version was featured in the newspaper several years ago. Since then we’ve cleaned up the model a bit and also updated to include more recent data.

In addition to anything this work has taught us in political science and sociology, the project is notable as one of the early applications of Stan, not just a use of Stan to fit a model so that we could check that Stan works, but an applied problem where we built, fit, and developed a model of sufficient complexity that we never would have constructed, had Stan not existed.


  1. Ben Hanowell says:

    Disentangling age, period, and cohort effects would be a huge boon in my field (people analytics), where we try and understand patterns of worker attrition and engagement. We have lots of unproductive fights about tenure, hiring cohort, and tenure effects (at least once we are able to educate folks that these are three different things!).

    Anyway, where’s the Stan code for this masterpiece? The narrative version of the model is so rad. I love the clarity of thinking about generational effects in terms of weighted sums of period effects. So much more meaningful than tossing year and quarter of hire into a model and then either period or tenure, and shrugging.

  2. Benjamin Hanowell says:

    “We have lots of unproductive fights about tenure, hiring cohort, and tenure effects”

    Obviously I mean period, hiring cohort, and tenure effects.

  3. Martha (Smith) says:

    Looking at cohort/birth year rather than age-at-time-of-survey makes a whole lot of sense to me. After all, we so often talk about differences between the sociological “generations”.

  4. jim says:

    Haven’t read the paper but at first glance the data lined up by birth year (right above) seems to dispute the idea that there’s not much swing vote. A *lot* of people bailed on Republicans in 2008, my goodness! There are significant variations in other parts of the chart as well. I think this data says that the “swing” is composed of a set of independent groups. Mostly, they average each other out. But it’s possible in any election that the right candidate could line up the 7s and hit the jack pot.

Leave a Reply