In survival analysis, several tests are available to compare two independent survival curves, e.g. Logrank test with different weights, Gehan's test, Cox-F test, Peto's test, etc.
I found in my daily work as a statistician that the Cox-F test has more power than the log-rank tests.
I think the Cox-F test is not the test within the Cox-regression model. It is a separate test.
Now, I want to make Monte Carlo simulations to compare the power of the Cox-F test with the family of log-rank tests.
In order to do that, I need the underlying formulas for the Cox-F test. I ask you to give me a hint where I can find such a formula, e.g. a pdf file, weblink or other reference.
I think this is important with a far reaching impact, e.g. in medical statistics most of the time logrank test is used, but I strongly suspect that the Cox-F test has much higher power.
Thank you.
Guess the paper "The Relationship Between Two Statistical Tests for the Comparison of Two Exponential Distributions" by Le (Biometrical Journal, 33) contains the answer. The very last formula is the one proposed by Cox in his 1953 paper "Some simple approximate tests for Poisson variates" (Biometrika, 40, 354-360).