Nonparametric Estimates of Survival Function from Interval Censored Observations

In the paper estimates of a survival function of a time to event random variable are discussed for data containing only interval censored values. In practice these data are usually transformed to datasets with exact and right censored data and well implemented methods in statistical software are used for estimation. In the paper four possible transformations are used and their properties are compared with results of proper methods for interval censored data. Kaplan-Meier and maximum likelihood estimates of survival function are computed. Only small samples are studied by simulation study for lognormal distribution.