Yesterday was the Idaho Republican primary. As far as primaries go it wasn’t very meaningful, since the presumptive nominee has already almost certainly wrapped things up. The Democratic nomination is not quite so certain, but we’re getting to the point where there’s realistically not much doubt remaining as to the outcome.
How can we assign terms like “almost certainly” and “not much doubt” to something that hasn’t happened yet, and will only happen once? It’s not like we can repeat the nomination process many times and use the results to estimate the probabilities. This type of ambiguity has resulted in the quixotic battle between the Bayesians and the Frequentists over the interpretation of probability. I don’t know which side is Don Quixote and which is the windmill, but I don’t worry about it too much. If it isn’t testable experimentally there’s not much point from a physics perspective.
However, even for events that only happen once, we can meaningfully define probabilities if there’s a general theory describing that class of events. Want to shoot an electron through a magnetic field and measure its spin? If you know what state the electron started in, you can find the probabilities of each of the two possible spin states being observed at the end. Even if you only do this with one electron and your lab catches fire afterwards, you can still talk about what the chance was for each outcome even if you only were able to do one measurement. The theory tells you what those chances were.
Unfortunately there is no general theory of politics. People are just too complicated. But we’ve seen previous nomination processes and how they turned out. Despite the uniqueness of this year’s situation, we’re not without precedent that allows us to at least formulate a good guess as to each candidate’s chances. We could be wrong about the probabilities, and even then a probability is nothing like a guarantee of a result. But it’s a start for those interested in predicting the political future. Those hypotheses can be further refined by looking at the results of elections and comparing with the “theoretical” predictions.
We in the real sciences (kidding!) sometimes sneer at those disciplines like political science which don’t readily yield to mathematical description or easily repeatable experiments. But it’s very appropriate to remember that the methods of science are useful all the time - not just in the lab, and even if strict rigor is precluded by the circumstances. Even trying different ways of cooking your dinner is science in a sense. You make a hypothesis about what tastes good, and you test your supposition by doing the experiment of making dinner. Politics is just a few million people trying to test their hypotheses about government using the same experimental apparatus simultaneously. Will demographic group X vote for my candidate? Will candidate Y help the economy? Every election is an experiment, and politicians, pundits, and citizens will learn more about the statistical mechanics of the electorate by paying attention to the results. Now mathematics and rigor are simultaneously the crown jewels and Swiss army knives of the hard sciences. But for the softer sciences… well even they can get results sometimes, and I don’t begrudge them just because they have a long way to catch up! To the extent that their ideas are built up from the observed facts, they’re on the right track.