Jeffrey D Picka (Associate Professor)
BASc, BSc, MSc (Toronto), PhD (Chicago)
If you look at water boiling through rapids or clouds in the sky, you see chaos. In what sense is this chaos difficult to predict (random) or difficult to describe (disordered)? My research seeks to take useful definitions of disorder, unpredictability, and uncertainty and to use them to establish criteria for making truth-claims about Nature from immensely complicated numerical models of natural phenomena. It also involves developing useful statistical methods for evaluating the match between simulations and data observed in Nature. These methods involve spatial statistical inference, but for spatial data which is neither stationary in space nor stationary in time. Applications are diverse, including models for climate change, powder flow, epidemics, and patterns formed in Belousov-Zhabotinsky reactions.
My other main research interests involve bringing new developments from science studies into the practice of statistical inference and into statistical education. Since the development of hypothesis testing in the 1930s, the view of science as revealed truth about Nature has been transformed into a view of science as the process by which communities of researchers seek to construct high-quality information about regularities in Nature. This newer view of science provides a far more accurate description of what science is than do many statistics texts. Ideas from the philosophy, sociology, history, and rhetoric of science can be used to help graduate students learn consulting, and to help all students understand the applied-philosophical aspects of probability and inference. In particular, they can be used to show how statistics differs from other forms of data science in that it uses doubt constructively to improve the quality of scientific information.