|
Abstract:
|
Very little attention has been given to examine the performance of methods for analyzing continuous outcome from the stratified cluster randomized trials (CRTs). Recently, we have empirically examined the sensitivity of several methods for assessing the treatment effect of a school-based stratified CRT. These methods differ in terms of effect estimate, width of the confidence intervals and standard errors but the overall conclusions were similar. In this study, we are assessing the performance of methods for analyzing continuous outcome from stratified CRTs. Several methods will be considered to examine the treatment effect. The individual-level methods include: i) linear regression ii) mixed-effects method iii) GEE with exchangeable and AR(1) correlation structures. The cluster-level methods include: (iv) cluster-level linear regression and (v) meta-regression. Performance of these methods will be evaluated in terms of type I error, power, coverage and mean square error using the simulation study – which has already been designed. Also, these methods will be compared with or without adjusted for stratification.
|
