Large Sample Theory for Sequential Analysis of the Proportional Hazards Model.
Abstract
An appropriate large sample theory for sequential analysis of the Cox proportional hazards model is developed. For clinical trials with simultaneous entry of patients, the efficient score process of the partial likelihood is easily seen to be a martingale. It follows that, in a time scale based on the observed Fisher information, the score process and the properly normalized maximum partial likelihood estimator behave asymptotically like Brownian motion. When entry is staggered, the efficient score process is no longer a martingale in general. However, if patients in a staggered-entry clinical trial are assumed to be independent and identically distributed, independently of entry time, then the score process is well approximated by a martingale. The asymptotic results involving weak convergence to Brownian motion hold as before. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1982
- Accession Number
- ADA120646
Entities
People
- Thomas Sellke
Organizations
- Stanford University