摘要: In some problems in survival analysis there may be more than one plausible measure of time for each individual. For example mileage a better indication the age car months. This paper considers possibility combining two (or more) scales measured on individual into single scale. A collapsibility condition is proposed regarding combined scale as fully informative survival. The resulting model regarded generalization usual accelerated life that allows time-dependent covariates. Parametric methods choice scale, testing validity assumption and parametric inference about failure distribution along new are discussed. Two examples used to illustrate methods, namely Hyde's (1980) Channing House data large cohort mortality study asbestos workers Quebec.