Trigonometric Maximum Likelihood Estimation and Application to the Analysis of Incomplete Survival Information
Journal of the American Statistical Association · 1979 · Volume 74 · Issue 365
DOI: 10.1080/01621459.1979.10481629
Abstract
A procedure is derived for computing maximum likelihood estimates from the sample trigonometric moments. Several examples are given which demonstrate the convenience of this hybrid approach. The Zippin-Armitage method for dealing with incomplete survival information is generalized. A simple expression is obtained for the survival curve given nonparametric estimators of (1) the probability density of survival times for patients who have died, and (2) the probability density of known survival times for patients still alive. It is shown that this new estimator approaches the Kaplan-Meier product limit as the sample size approaches infinity.
Enable JavaScript for citation tools and additional paper interactions.