Here is the abstract and link, where you’ll find a link to 16 pages of pdf. After you’ve studied them, please come back and let us know what you found out. Or, you can just say, “The hell with it. If I have to dig it out myself, it must be lie.”
American presidential elections run in cycles that have turning points after about two to three terms of a party’s control of the White House. This is not the pattern associated with realignment eras that are presumed to last 30 years or so. The cyclical dynamic is estimated with a second-order autoregressive model. More than three years before the day of the next presidential election is not too early for the cyclical model to offer its forecast. With parameter estimates and the requisite values for the predictors at hand, the cyclical forecast is able to make an unconditional forecast for the 2016 presidential election: 51.4% of the two-party popular vote for the Republican candidate. The 2016 contest shapes up as “Time for a Change†election. After two terms, consistent with the logic of the cyclical model, change looms larger than continuity.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2303042
Thanks! I will read it. Many moons ago I had an experiment published in JASA - like I said, MANY moons ago.