Royal Statistical Society
Research Section





Meetings 2002/2003 (all are RSS ordinary meetings)

2002
5.00 p.m.,
16 October
SUSAN A MURPHY (University of Michigan)

Optimal Dynamic Treatment Regimes


A dynamic treatment regime is a list of decision rules, one per time interval, for how the level of treatment will be tailored through time to an individual's changing status. The goal of this paper is to use experimental or observational data to estimate decision regimes that result in a maximal mean response. To explicate our objective and state assumptions, we use the potential outcomes model. The proposed method makes smooth, parametric assumptions only on quantities directly relevant to the goal of estimating the optimal rules. We illustrate the proposed methodology via a small simulation.
From left to right Elja Arjas (Proposer), Susan Murphy (Speaker), Chris Jennison (Seconder)
From left to right Chris Jennison (Seconder), Susan Murphy (Speaker), Elja Arjas (Proposer)
From left to right David Cox (Discussant), Susan Murphy (Speaker), Bernard Silverman


The full published version with discussion will be published in The Journal of the Royal Statistical Society, Series B.


2002
5.00 p.m.,
11 December 2002
AUGUSTINE KONG (deCode Genetics, Iceland),
PETER MCCULLAGH, (University of Chicago)
XIAO-LI MENG (Harvard University),
DAN NICOLAE AND ZHIQIANG TAN (University of Chicago),


A theory of statistical models for Monte Carlo integration.


The task of estimating an integral by Monte Carlo methods is formulated as a statistical model using simulated observations as data. The difficulty in this exercise is that we ordinarily have at our disposal all of the information required to compute integrals exactly by calculus or numerical integration, but we choose to ignore some of the information for the sake of simplicity or computational feasibility. Our proposal is to use a semi-parametric statistical model that makes explicit what information is ignored and what information is retained. The parameter space in this model is a set of measures on the sample space, which is ordinarily an infinite dimensional object. Nonetheless, from simulated data the baseline measure can be estimated by maximum likelihood, and the required integrals computed by a simple formula previously derived by Vardi (1985) and by Lindsay (1995) in a closely related model for biased sampling. The same formula was also suggested by Geyer (1994) and by Meng and Wong (1996) using entirely different arguments. By contrast with Geyer's retrospective likelihood, a correct estimate of simulation error is available directly from the Fisher information. The principal advantage of the semi-parametric model is that variance reduction techniques are associated with sub-models in which the maximum-likelihood estimator in the sub-model may have substantially smaller variance than the traditional estimator. The method is applicable to Markov chain and more general Monte Carlo sampling schemes with multiple samplers.
From left to right Mike Evans (Proposer), Peter McCullagh, Xiao-Li Meng, Dan Nicolae, Augustine Kong, Christian Robert (Seconder)


Electronic version of the paper.

Venue
Royal Statistical Society Lecture Hall,
12 Errol Street, London EC1Y 8LX.
Tea was served from 4.30pm.

For further information
The full published version with discussion will be published in The Journal of the Royal Statistical Society, Series B.


2003
5.00 p.m.,
21 May 2003
OLE E BARNDORFF-NIELSEN (MaPhySto, University of Aarhus, Denmark),
RICHARD D GILL, (Mathematical Institute, University of Utrecht and EURANDOM, Eindhoven, Netherlands) and
PETER E JUPP, (School of Mathematics and Statistics, University of St Andrews, UK)


On Quantum Statistical Inference


Interest in problems of statistical inference connected to measurements of quantum systems has recently increased substantially, in step with dramatic new developments in experimental techniques for studying small quantum systems. Furthermore, theoretical developments in the theory of quantum measurements have brought the basic mathematical framework for the probability calculations much closer to that of classical probability theory. The present paper reviews this field and proposes and interrelates a number of new concepts for an extension of classical statistical inference to the quantum context.

Picture: left to right Luigi Accardi (Proposer), Richard Gill, Dorje Brody (Seconder).


Electronic version of the paper

Venue
Royal Statistical Society Lecture Hall,
12 Errol Street, London EC1Y 8LX.
Tea was served from 4.30pm!

For further information
The full published version with discussion will be published in The Journal of the Royal Statistical Society, Series B.


  • Back to the Royal Statistical Society Research Section home page