Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Merrill, Samuel III; Grofman, Bernard; Brunell, Thomas L. 2008. Cycles in American National Electoral Politics, 1854-2006: Statistical Evidence and an Explanatory Model. American Political Science Review 102 (February): 1-17.
In 1924, Arthur Schlesinger famously predicted that "Coolidge-style conservatism would last till about 1932." Later, he added that the "prevailing liberal mood would run its course in about 1947." In 1949, he predicted once again that "the recession from liberalism was due to end in 1962," and that the "next conservative epoch will commence around 1978." In every case, he was startlingly correct; as predicted, the nation's ideological mood reversed about every 15 years.
Schlesinger's feat inspires the present study. The authors argue that there are cycles in American ideology. Roughly every 12 to 15 years since 1854, American voters have oscillated between preferring Democrats or Republicans in national office, with a complete cycle every 25 to 30 years.
The basis of this claim is nakedly empirical. In their first table, the authors perform elementary run tests to demonstrate that "both the House and the Senate have far fewer runs (i.e., fewer partisan switches) than" randomness would allow. That is, one party tends to control the House (or Senate) over several elections, then the other party gets a turn.
It's not enough to show that changes in partisan control are non-random, though, to explain Schlesinger's uncanny predictions. So the authors turn to a specialized technique, spectral analysis, to see whether there is a consistent cycle length causing these runs. They analyze House, Senate, and presidential elections separately, then all together. All four analyses yield a consistent result: A half-cycle (i.e. change in partisan dominance) of around 13 years, a full cycle (i.e. return to original party) of around 26.
This empirical finding influences our understanding of partisan realignment in two ways. First, and most importantly, it revives a literature severely criticized by Mayhew (2002). Realignments do occur. Nonetheless (and second), the authors carefully point out that their evidence suggests a sort of tidal ebb and flow, not a punctuated equilibrium (or "tipping point") model in which realignments occur suddenly following a "critical election."
They explain these findings with a formal model that has four main moving parts:
The first two considerations create a centrifugal-centripetal tension, simultaneously pulling the party toward the median and toward extremism. Over time, the second consideration overpowers the first.
The latter two points also conflict, reflecting a tension among voters between valuing incumbency and desiring change. When a party first acquires power, its newfound incumbency advantage helps it at first. But as time goes on, the advantages of incumbency are overwhelmed by changes in the electorate's mood.
Together, these two pairs of considerations create a cyclical pressure.
Research on similar subjects