American Statistical Association
This presentation will provide methodology for power and sample size calculations for prospectively designing studies that aim to distinguish a more predictive marker from an inferior one, when the endpoint is subject to censoring. The method is applicable to any binary covariate, such as symptom grading (e.g. severe/non-severe), under the assumption that these covariates are measured on the same subjects (paired data). Simulations are performed under alternative schemes by varying the correlation between the covariates; degree of censoring; and parameters of the assumed survival distribution. Power calculations are obtained for a fixed difference in the predictive accuracy between the two covariates, as a function of the sample size. Software availability and limitations will also be discussed.
|Date:||Wednesday, February 11, 2009|
|Time:||4:00 P.M. - 5:00 P.M.|
Memorial Sloan-Kettering Cancer Center
Department of Epidemiology and Biostatistics
307 East 63rd Street
(between First and Second Avenues)
3rd Floor Conference Room
New York, New York
Note: To gain access to the building, please follow the directions by the telephone in the foyer.