A Performance Study of Cox's Model

• Main Author: 江英程
• Other Authors: 高正雄
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/20558639685536845528

碩士 === 國立中正大學 === 數理統計研究所 === 88 === Proportional hazards model has become one of the most widely-used semi-parametric models in survival analysis. In this work, we try to investigate the performance of Cox's model in the case when its covariates are continuous and time-dependent variables. It is found that estimates of the explanatory coefficients in the model by the so-called partial likelihood method are not as good as traditionally assumed. Serious bias of the estimators could occur when the time-dependent covariates are random and especially with small variance. Also the bias of the estimators increases as the proportion of censoring increases. Such phenomenon causes alarm on the proper use of Cox's model for attacking real problems.