Modified Kendall's tau as a measure of discrimination in survival analysis. MICHAEL J. PENCINA AND RALPH B. D'AGOSTINO Boston University Abstract:  Measures of association based on ranks find applications in numerous fields [cf. 1]. We use a modification of Kendall's tau coefficient proposed by Noether [2] to assess the discrimination ability of a survival analysis model - a problem of considerable theoretical interest and important practical applications. In this context, discrimination is viewed as a measure of concordance between predicted and actual survival times. It is quantified by the overall C coefficient suggested by Harrell [3] as an extension of the ROC curve area to survival analysis.  We develop the overall C index as a parameter describing the performance of a given model applied to the population under consideration and use the statistic given in Harrell [3] as its sample estimate.  We use the developments presented in Noether [2] to a construct confidence interval for our measure and we investigate its properties via numerical simulations.   Bibliography: 1. Lehmann, E.L., Nonparametrics, San Francisco, Holden-Day Inc., 1975. 2. Noether, G.E., Elements of nonparametric statistics, New York, Wiley,    1967. 3. Harrell, F.E. et al., 'Tutorial in Biostatistics: Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors', Statistics in    Medicine, vol. 9, 487-503 (1990).