With a pioneering application of probability models in political science, Stokes and Iversen established "the existence of forces restoring party competition." Whatever the margin of victory in a given election, the partisan vote subsequently tends to return to the point of equal division. The authors introduce an expanded test of electoral equilibrium that allows for effects of major realignments and regional differences, using congressional elections since 1828. They find that the vote division gravitates to the mean but that the mean vote, in most periods of American history and in several regions, departs significantly from the point of equal division and in some instances is prone to a pronounced drift. Hence, during much of their lifetime, many Americans do not experience, in congressional elections, party competition that gives the opposition much of a chance to win.
Keywords: elections; voting behavior
The two real political parties in America are the Winners and the Losers. The people don't acknowledge this. They claim membership in two imaginary parties, the Republicans and the Democrats, instead.
-Kurt Vonnegut Jr., 1974
Democratic politics requires both fear and hope- the fear of losing among parties in office and the hope of winning among parties out of office.1 The prospect of losing office theoretically induces incumbent office holders to attend to the public will. Failure to do so is seen as opening the door to those out of office at the next election, when the public has an opportunity to administer vengeance. Yet is change of party control through elections common enough in American politics to make incumbents fearful enough of losing and the opposition hopeful enough of winning? Do the forces that fuel party competition trump the forces inhibiting it?
With a pioneering application of probability models in political science, Stokes and Iversen (1962/1966) established "the existence of forces restoring party competition." Without identifying or measuring any of these forces, the test showed that over time, the partisan split of the major-party vote in American elections gravitates toward the point of equal division. In other words, the winning party sooner or later suffers a setback, while the losing side stages a comeback. It would simply be too unlikely for this to happen, they concluded, without the gravitational pull of competitive forces.
How compatible is this notion with the view that electoral politics runs in long cycles, where political domination by one party persists over a considerable span of time followed by domination on the part of another party? Moreover, such partisan alignments may have a strong regional flavor, giving one party a stronghold in one region of the country and another party a stronghold in another. The "solid South" under Democratic rule is the most obvious case.
Our test for electoral equilibrium controls for the effects of major partisan realignments and for regional differences. We also offer separate assessments of (1) the return of the vote to the mean, (2) the closeness of that mean to the point of equal division, and (3) the extent of a drift of the mean vote. We do so with a universe of elections (1828-2004) nearly twice as large as used by Stokes and Iversen (1868-1960). To gain maximum leverage, we chose House elections, the longest running and most frequently obtained measure of the national popular vote. Both midterm and presidential elections were included, going back as far as popular votes for major-party candidates across the country can be obtained (1828). We conclude with some speculation about the forces that may help restore party competition and those that inhibit it.
The Flow of the Vote
By comparison with other countries, one of the most remarkable features of American politics is that only two and essentially the same two parties have controlled elections almost as long as popular vote tallies are available. …