Inference with Bivariate Truncated Data

Christopher M. Quale, Division of Biostatistics, School of Public Health, University of California, Berkeley
Mark J. van der Laan, Division of Biostatistics, School of Public Health, University of California, Berkeley

Abstract

In this paper we build on previous work for estimation of the bivariate distribution of time variables when they are observable only on the condition that one of the times variables is greater than (left-truncated) or less than (right-truncated) some observed time variable. We propose various inferential methods, including variance and confidence interval estimation as well as a test for independence of events. We will demonstrate the implementation of these methods in simulation studies and a data analysis. In addition we present ad-hoc estimators for important special cases of truncation of bivariate right censored data.