chapter  2
Variability between Clusters
Pages 14

In a conventional clinical trial, in which individual subjects are randomly allocated to the different treatment arms, we can generally assume that observations on all individuals are statistically independent. In a CRT, in which groups or clusters of individuals are randomly allocated to treatment arms, the assumption of independence is usually invalid because observations on individuals in the same cluster are correlated. Much of this book is devoted to describing statistical methods of design and analysis that take such correlations into account.