proptest: tests of the proportional hazards assumption

Description

proptest is an R package for testing the proportional hazards assumption in the Cox model for right censored survival data. Two types of tests for identifying nonproportional covariates are implemented:

• data-driven Neyman type smooth tests (the function smoothproptest),
• tests based on the score process (the function scoreproptest).

Download and installation

You can install the package standardly directly from CRAN (using install.packages or R CMD INSTALL or a menu item in your GUI).

To download the package please visit its page on CRAN.

References

Data-driven smooth tests of the proportional hazards assumption [PDF, 212 kB]

Lifetime Data Anal., 13, 1–16, 2007
The original publication is available at www.springerlink.com
doi:10.1007/s10985-006-9027-8

Identifying nonproportional covariates in the Cox model [PDF, 97 kB]

Comm. Statist. Theory Methods, 37, 617–625, 2008
doi:10.1080/03610920701669744