Large sample theory for Schemper's measures of explained variation in the Cox regression model

摘要: Two new measures aimed at quantifying the predictive precision of a proportional hazards regression model have recently been introduced by Schemper, who argued that could provide useful generalization to notion proportion variation explained for linear models. Other approaches problem potential advantage over Schemper measures, having clearly defined population quantity which their suggested converge. The purpose this paper is work out characteristics measures. Large sample comparisons then become possible. On basis study we conclude some weaknesses not previously observed, e.g. fact counterparts depend on unknown censoring mechanism even mechanisms are independent time, common working assumption such contexts. For balanced exponential without censoring, studied derive exact asymptotic properties, enabling us two converge same and they bounded absolutely value 0.5. enables see how it would be possible an estimator converging does mechanism.