|
Activity Number:
|
296
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Stat. Sciences*
|
| Abstract - #300321 |
|
Title:
|
Statistical Treatment Rules for Heterogeneous Populations
|
|
Author(s):
|
Charles Manski*+
|
|
Affiliation(s):
|
Northwestern University
|
|
Address:
|
2003 Sheridan Road, Evanston, Illinois, 60208, USA
|
|
Keywords:
|
statistical decision functions ; missing data ; risk ; identification ; ambiguity ; treatment response
|
|
Abstract:
|
An important objective of empirical studies of treatment response is to provide decision makers with information useful in choosing treatments. Identification problems combine with the statistical problem of inference from finite samples to limit the informativeness of the available studies. I have earlier shown how identification problems generate ambiguity about the identity of optimal treatment choices. Here I use Wald's concept of the risk of a statistical decision function to study treatment choice using finite-sample data. Consider a planner who must choose among alternative statistical treatment rules, these being functions that map observed covariates of population members and sample data on treatment response into treatment choices. I first view the planner's problem in generality and draw broad conclusions for the conduct of research on treatment response. I then address a specific question of analytical interest and practical importance, this being the use of covariate information in treatment choice when the data are from a classical randomized experiment.
|
