Resampling Dependent Concordance Correlation Coefficients

Abstract

The concordance correlation coefficient (CCC) is a popular index for measuring the reproducibility of continuous variables. We examine two resampling approaches, permutation testing and the bootstrap, for conducting hypothesis tests on dependent CCCs obtained from the same sample. Resampling methods are flexible, require minimal marginal and joint distributional assumptions, and do not rely on large sample theory. However, the permutation test requires a restrictive assumption (exchangeability) which limits its applicability in this situation. Simulation results indicate that inference based on the bootstrap is valid, although type-I error rates are inflated for small sample sizes (≈30). For illustration we analyze data from a carotid stenosis screening study.

Key Words:

• Agreement
• Bootstrap
• Concordance correlation coefficient
• Correlated data
• Permutation test
• Resampling

ACKNOWLEDGMENT

Sara Crawford's research was supported in part by an appointment to the Research Participation Program at the Centers for Disease Control and Prevention, National Center for Infectious Diseases, Division of Parasitic Diseases administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and CDC.

Notes

^a p-value for the permutation test.

^bConfidence interval for the BC_A bootstrap method.

^a p-value for the permutation test.

^bConfidence interval for the BC_A bootstrap method.