While most of us tend to predict the outcome of presidential elections based on perceptions of the viability of candidates and what happens to be the hot issues of the day, it is useful to bear in mind that the real professionals (like political scientists) look at different things in their crystal balls.
A political scientist colleague of mine who tracks these things closely says that what they look at are markers of the state of the economy. The best predictor of presidential elections is the change in real disposable income and the GDP. Since there is usually a six-month lag in these numbers filtering down to voters, the figures that come in starting around May will give us a good indication of which way the election will go in November, largely irrespective of the candidates and the issues. Figures like unemployment do not matter so much because most people are employed and so it does not affect them directly.
This seems counterintuitive. Surely the GRAGGS (guns, race, abortion, gays, god, sex) issues that occupy so much time and space and arouse so much passion must influence the way people vote? But apparently they are not very good as predictors. I am guessing here but suspect that it is because most people’s views on these are fixed and thus the debates have little chance of changing the way people vote and merely serve to stoke the intensity of their feelings. And for those who can be swayed, the effects are random and largely cancel each other out, in that for every undecided voter who decides to vote for the Democrat because he or she is disturbed by the contraception uproar, there is likely another who it drives to the Republican camp.