Use this URL to cite or link to this record in EThOS:
Title: Theory and practice of mixed models applied to medical research
Author: Brown, Helen K.
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2006
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis examines in depth the properties of mixed models and considers their application in a variety of designs used in medical research. Mixed models are a broad class of models which allow variation in the data to be modelled at several levels and take into account correlations occurring between observations. They offer several potential advantages over the more conventional fixed effects approaches: more efficient estimates, effective handling of missing data and more appropriate inference. The different types of mixed model are placed into a unified format and the properties of various fitting methods, including likelihood-based methods, least squares methods and the Bayesian approach, are considered in detail. The practical implications of using mixed models are examined and the submitted material would appear to be the first to consider these in such depth. The particular features of applying mixed models to a range of designs are considered including repeated measures, crossover, multi-centre, meta analysis, cluster randomised, hierarchical, bioequivalence and several more ad hoc designs. Novel approaches are introduced for sample size estimation and for analysing crossover designs with multiple periods, bioequivalence studies and case-control studies. Comparisons of mixed models with fixed effects models, which have often previously been the conventional approach, are given particular attention. Models suitable for both normal and non-normal data are considered and examples involving original analyses are used to illustrate the properties described. The published material comprises two editions of a textbook and ten journal publications.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available