chapter  5
Matching and Stratifi cation
Pages 20

In a simple unmatched CRT design, a set of clusters is randomly allocated between two or more treatment arms. If there is substantial between-cluster variability, it may be decided to fi rst group together clusters that are expected to be similar with respect to the outcome of interest, and to allocate the treatment within these groups. This is usually done in an attempt to prevent imbalance in important covariates between the treatment arms, and to improve the precision and power of the trial.