Sampling Bias in Population Studies

Modiied versions of the life time distribution are often used in survival analysis. The modi-cations depend on how we choose individuals for the study and on the assumptions on the behavior of the population. A rigorous point process description of the Lexis diagram is used to make the sampling mechanisms and the preconditions transparent. The point process description gives a framework to handle all possible sampling patterns. The setup is generalized so it can handle more complicated life descriptions than just lifetimes, and the disability model is used as an example. Two setups can be used. Conditional on the birthtimes, the modiica-tions of the life time distribution are left truncation and either right censoring or truncation. Assuming that the birthtimes can be described by a Poisson process the modiications are length bias and the recurrence time distribution known from renewal theory.