Confidence interval construction for a difference between two dependent intraclass correlation coefficients

Inferences for the difference between two dependent intraclass correlation coefficients (ICCs) may arise in Studies in which a sample of subjects are each assessed several times with it new device and a standard. The ICC estimates for the two devices may then be compared using a test of significance. However, a confidence interval for a difference between two ICCs is more informative since it combines point estimation and hypothesis testing into a single inference statement. We propose a procedure that uses confidence limits for it single ICC to recover variance estimates needed to set confidence limits for the difference. An advantage of this approach is that it provides a confidence interval that reflects the underlying sampling distribution. Simulation results show that this method performs very well in tot-ins of overall coverage percentage and tail errors. Two data sets are used to illustrate this Procedure. Copyright (C) 2009 Joint Wiley & Sons, Ltd.