| Autor: |
Pascale Tubert-Bitter, Jean-Jacques Chalé, Andrew Kramar, Thierry Moreau |
| Rok vydání: |
1994 |
| Předmět: |
|
| Zdroj: |
Computational Statistics & Data Analysis. 18:547-559 |
| ISSN: |
0167-9473 |
| DOI: |
10.1016/0167-9473(94)90084-1 |
| Popis: |
The performances of several linear rank tests for comparing two potentially censored samples is investigated for crossing hazards alternatives. A newly described test (Moreau et al., 1992) which is specifically designed for such alternatives is considered along with the ones most commonly used in survival analysis. Simulations experiments are carried out according to various survival patterns, since survival curves are more used in describing differences between two graphs than hazard functions. In the presence or absence of censoring, power is estimated in situations where survival distributions cross at the beginning, at the middle, at the end and beyond the maximum observed failure time. Simulation results suggest higher power of the Moreau et al. test for medium or late crossing survival curves. This trend is discussed through several examples in oncology. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
