B946 - Methods for modelling repeated measures in a lifecourse framework - 21/01/2010

B number: 
B946
Principal applicant name: 
Prof Kate Tilling (University of Bristol, UK)
Co-applicants: 
Dr Jon Heron (University of Bristol, UK), Prof Fiona Steele (University of Bristol, UK), Prof John Rasbash (Deceased) (University of Bristol, UK), Prof Jonathan Sterne (University of Bristol, UK), Mark Gilthorpe (Not used 0, Not used 0)
Title of project: 
Methods for modelling repeated measures in a lifecourse framework
Proposal summary: 

Objective 1: To develop methods for detecting and modelling autocorrelation.

Objective 2: To examine the impact of variance assumptions on multilevel models

Objective 3: To develop methods for developing individual trajectories and spline models

Objective 4: Taking extra variance from estimating splines into account. Including comparison with SEMs/WINBUGS.

Objective 5: To examine and develop methods for using class assignment probabilities

Objective 6: To develop methods for the initial conditions of cross-lagged models

Objective 7: To develop methods for handling missing data in cross-lagged models and in GMM models

Objective 8: To develop more objective tests for assumptions of multilevel models, and identify tests for when departures from these assumptions may cause problems. Assumptions: normality of residuals (see objective 3), autocorrelation (see objective 1), identification of mixtures (see objective 3 and latent class section), variance function (see objective 2).

Objective 9: To examine the plausibility and usefulness of methods developed above.

Date proposal received: 
Thursday, 21 January, 2010
Date proposal approved: 
Thursday, 21 January, 2010
Keywords: 
Primary keyword: