Seeing the Future: A Crystatistical Ball.

Abstract

We are used to identifying risk factors, or predisposing factors, for outcomes of interest, whether those outcomes are negative or positive, using the tools of statistical modeling. We are far less accustomed to making predictions of outcomes for specific individuals or testing the accuracy of those predictions. In this commentary, we consider this more difficult task and discuss, in this context, why statistics has such difficulty talking about individuals, even as it easily informs us about populations.