Making and Assessing Epidemiological Predictions

In the previous chapter we saw, among other things, that a good prediction claim is one that is justified by a good prediction activity. In this chapter, we apply the notion of stability to prediction in the epidemiological context to yield a simple philosophical theory of what makes a good prediction activity in that context. We compare this view to some other possible ideas about what good epidemiological prediction might be: extrapolation; inference from laws of nature; understanding of underlying mechanisms. None of these is either necessary or sufficient for good prediction. A good prediction activity is one that explains why the prediction claim is true rather than alternatives identified by best current knowledge as ways the prediction might go wrong, in a suitably restricted sense of “might”.