GOFCOX - A COMPUTER-PROGRAM FOR THE GOODNESS-OF-FIT ANALYSIS OF THE COX PROPORTIONAL HAZARDS MODEL

被引:1
|
作者
LIN, DY
机构
关键词
CENSORING; COX REGRESSION; FAILURE TIME DATA; FORTRAN; LACK OF FIT; MODEL MISSPECIFICATION; SURVIVAL ANALYSIS;
D O I
10.1016/0169-2607(92)90080-Q
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
GOFCOX is a user-friendly FORTRAN program for assessing the adequacy of the Cox proportional hazards model. The underlying methodology is based on the comparison of the maximum partial likelihood estimator and a weighted parameter estimator. The latter is the root to an estimation equation that assigns varying weights to the individual contributions to the partial likelihood score function. The weighted and unweighted parameter estimators have the same expectation under the Cox model, but tend to differ when the model is inappropriate. The GOFCOX program computes a rich class of weighted parameter estimators and corresponding goodness-of-fit test statistics. The program runs on both mainframe computers and microcomputers. The running time is minimal even for large data sets. A simple example is provided to illustrate the features of the program.
引用
收藏
页码:101 / 105
页数:5
相关论文
共 50 条