
2
Sample Size Determination for Clustered
Outcomes
2.1 Introduction
Clustered data frequently arise in many fields of applications. We frequently
make observations from multiple sites of each subject (called a cluster). For
example, observations from the same subject are correlated although those
from different subjects are independent. In periodontal studies that observe
each tooth, each patient usually contributes data from more than one tooth
to the studies. In this case, a patient corresponds to a cluster, and a tooth
corresponds to a site.
The degree of similarity or correlation is typically measured by intraclus-
ter correlation coefficient (ρ). If