KEELE UNIVERSITY
CENTRE FOR MEDICAL STATISTICS
SEMINAR SERIES on BIOSTATISTICS 2002/2003
NO 5: Wednesday, February 5th 2003, at 2:30 pm
"Modelling covariance in longitudinal studies"
Professor Gilbert MacKenzie
Keele University
The idea that in longitudinal data analysis (LDA) the mean might be
modelled as a function of time and baseline covariates ( eg , the
treatment indicator in an RCT) is
well-established. The idea that the second parameter in the model, the
covariance, might be modelled is also well-establised, but that it might
be modelled in a regression. similarly to the mean is rather new. A
third idea, that selection of the optimal pair of parameters should be
"joint", renders much current practice redundant. We review the key
concepts behind joint mean-covariance modelling in LDA, when the
response is continuous, develop an augmented regression model for the
mean and covariance and present a new algorithm for finding the optimal
model in the joint mean-covariance model space.
All welcome!!
Venue: Room 2.22, Third Floor
MacKay Building, Keele University
http://www.keele.ac.uk
http://www.keele.ac.uk/depts/ma/news/#medstats
http://www.keele.ac.uk/university/campus/maps/
*****************************
Janet Drewery
Centre for Medical Statistics
MacKay Building
Keele University
Keele, Staffs ST5 5BG
Tel: (01782) 583269
Fax: (01782) 583269/584268
E-mail: [log in to unmask]
*****************************
|