Testing Hypotheses for Effects on Survival by the Analysis of a Matched Retrospective Design
Abstract
In testing for the relation of risk factors to a particular cause of death, such as a rare disease, a longitudinal study requires the observation of many individuals for long periods of time before enough information is accrued to permit reliable statistical analysis. In the present paper, this difficulty is circumvented through the use of a matched retrospective design. In particular, tests of the hypothesis of no effect are obtained for the constant proportionality model and for a second model in which the risk factors are quantified. The asymptotic distribution of the test statistics are also derived.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1978
- Accession Number
- ADA063999
Entities
People
- Anastasios A. Tsiatis
- Bernard Harris
Organizations
- University of Wisconsin–Madison