Statistical decision theory for treatment choice and prediction

Statistical decision theory, developed by Wald (1950), provides a coherent frequentist framework for the use of finite-sample data to make decisions. Wald posed the task as choice of a statistical decision function, which maps potentially available data into a choice among the feasible actions. Statistical decision theory has great generality in principle and considerable potential usefulness in practice. This masterclass discusses recent applications to treatment choice and point prediction with sample data. The lectures cover both procedures for use of existing sample data and approaches to choosing sample designs.

Programme:
Session slides contain links to papers and books discussed in the masterclass.

Day One: Statistical Decision Theory with Given Data

10.15 – 10.45: Registration and coffee

10.45 – 12.15: (Manski) Introduction to statistical decision theory and simple illustrations
Download session 1 & 2 slides

12.15 – 13.45: Lunch

13.45 – 15.15: (Manski) Minimax-regret theory theory for treatment choice
Download session 1 & 2 slides

15.15 – 15.30: Coffee

15.30 – 17.00: (Tetenov) Treatment choice with many covariate values
Download session 3 slides

Day Two: Statistical Decision Theory for Data Collection

08.30 – 09.00: Coffee

09.00 – 10.30: (Tetenov) Choosing sample size in randomized experiments
Download session 4 slides

10.30 – 10.45: Coffee

10.45 – 12.15: (Manski) Choosing sample size in survey research
Download session 5 slides

12.15 – 13.15: Lunch

13.15 – 14.45: (Tetenov) Statistical decision theory with economic incentives
Download session 6 slides

14:45: Close