Statistical Analysis of Duration Data with Missing Time Origins

• Activity Number: 427 - Various Challenges and Strategies in Analysis of Real-Life Data
• Type: Invited
• Date/Time: Thursday, August 6, 2020 : 10:00 AM to 11:50 AM
• Sponsor: SSC (Statistical Society of Canada)
• Author(s): Joan X. Hu* and Yi Xiong and John Braun
• Companies: Simon Fraser University and Simon Fraser University and University of British Columbia
• Keywords: first-hitting-time model; interval censoring; joint modeling; longitudinal measures; mixed effects ; semiparametric regression

Abstract

Understanding the distribution of an event duration is essential in many studies. However, the exact time to the event is often unavailable and thus so is its duration. We link associated longitudinal measures and the event data via the first-hitting-time model (e.g. Lee and Whitmore, 2006). We estimate the duration distribution and conduct a semiparametric regression analysis of the data. The estimators are studied both asymptotically and numerically. By viewing the available data as interval-censored event times, we show that the proposed approach can be more efficient in the current setting compared to the Turnbull estimator, an well-established inference procedure with interval-censored data. A collection of wildfire records is used to motivate and illustrate the research. This is a joint work with Yi Xiong (Simon Fraser University) and John Braun (University of British Columbia).