Abstract
A cluster randomisation trial (CRT) is an experiment in which a researcher randomises intact clusters (e.g. families, schools, communities) rather than individuals in experimental trials. Reasons for adopting cluster randomisation trials are diverse. In this thesis, issues such as the choice of design, sample size calculations and approaches to the analyses, that arise in conducting such trials are discussed, with the emphasis on trials with binary outcomes.
Three approaches that are widely employed in CRT’s, namely the aggregated approach, the disaggregated approach and the adjusted chi-square approach, were employed on our data sets and it was found that in some cases the methods gave differing results. In particular, the disaggregated analysis is able to adjust for individual level covariates. Therefore, gaining power, and sometimes detect statistically significant differences where the other methods have not been able to.
|
indicates that a file or directory is accessible from the campus network only.