Bayesian Statistics the Fun Way

19FROM HYPOTHESIS TESTING TO PARAMETER ESTIMATION

So far, we’ve used posterior odds to compare only two hypotheses. That’s fine for simple problems; even if we have three or four hypotheses, we can test them all by conducting multiple hypothesis tests, as we did in the previous chapter. But sometimes we want to search a really large space of possible hypotheses to explain our data. For example, you might want to guess how many jelly beans are in a jar, the height of a faraway building, or the exact number of minutes it will take for a flight to arrive. In all these cases, there are many, many possible hypotheses—too many to conduct hypothesis ...