QOTD: Sampling the Future

By Paul Kedrosky · Friday, July 3, 2009 ·

The following innocuous sampling theory comment from Andy Gelman set me thinking in a bunch of dimensions today. The question had to with how to handle statistical analysis when your sample population is the entire population, and Andy's answer is important and instructive:

So, one way of framing the problem is to think of your "entire population" as a sample from a larger population, potentially including future cases.

Precisely right, and a point that many naive hypothesis-generators might keep in mind, whether in financial markets or elsewhere.

More here.