There are still some places available for this PSI training course which is taking place next week. Please contact PSI Executive Office at Resources for Business, South Park Road, MACCLESFIELD, Cheshire, SK11 6SH, telephone: +44 (0)1625 511750, fax +44 (0)1625 267879 to register or obtain further information.
PSI TRAINING COURSE
LIKELIHOOD-BASED STATISTICAL METHODS:
FROM THEORY TO SAS
Date 8th, 9th and 10th November 1999
Time 10.30 to 17.00 on Day 1 (registration from 10.00)
09.00 to 17.00 on Day 2
09:00 to 16:00 on Day 3
Location The De Vere Bull Hotel, Gerard's Cross, UK
Course Fee PSI Members £775.00 (plus £135.62 VAT @ 17.5%)
Non Members £825.00 (plus £144.37 VAT @ 17.5%)
Linear modelling for non-normally distributed responses is now widely used in the analysis of clinical trials: within the pharmaceutical industry this is usually accomplished using SAS procedures such as LOGISTIC, GENMOD, PROBIT, LIFETEST and PHREG. Methodologically all of these procedures are applications of likelihood theory, and a thorough understanding of the definition of likelihood and of likelihood-based estimation and hypothesis testing provides practising statisticians with insight into their use and interpretation. In particular, it allows the relationships and differences between these methods to be better understood.
The course will be presented by:
Professor John Whitehead and Dr Nigel Stallard, Medical and Pharmaceutical Statistics Research Unit, The University of Reading
The course registration fee includes morning coffee, lunch, afternoon tea, dinner and overnight accommodation (with en-suite facilities) on 8th and 9th of November plus breakfast, morning coffee, lunch and afternoon tea on 10th November.
A small number of rooms will be reserved for attendees who wish to stay at the hotel the night before the course. However, the availability of these rooms is strictly limited.
COURSE OBJECTIVES
To make participants aware that most of the methods of statistical analysis used in practice derive from a small central core of likelihood methods.
To enable them to work through the derivations of simple examples and to appreciate how more complicated situations can be tackled.
To show how approximation underlies most methods used in practice and how different choices of approximation account for the variety of formulae available.
To identify the likelihood methods which underlie various items of SAS output, and to encourage participants to allow the choice of statistical method to drive their use of SAS rather than the other way round.
COURSE OUTLINE
Fundamental concepts
Anomalous output from SAS procedures
Definition and interpretation of likelihood
Maximum likelihood estimation
Likelihood ratio tests
Optimality of procedures
Sufficient statistics
Efficient score and Fisher's information
Likelihood ratio tests, Wald tests and score tests
Finding these tests in SAS output
Efficiency of estimation
Uniformly most powerful tests
Dealing with multiple parameters
Nuisance parameters
Profile likelihood
Conditional, marginal and partial likelihood
Classification of SAS procedures by their use of likelihood
Further topics
Sequential methods
Bayesian methods
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|