Timeline for Pairwise Log-rank Tests on Cox Proportional Hazard Models: are they even necessary?
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 14, 2021 at 22:40 | comment | added | EdM | @lgreenst see the Wikipedia logrank test page, and Frank Harrell's course notes, Section 20.4. Logrank is just the score test on a simple Cox model. Harrell's course notes and book provide much insight not only into survival modeling but also into regression modeling in general. Taken together with the vignettes in the R survival package, you have some great resources from which to start learning. | |
| Feb 14, 2021 at 22:06 | comment | added | lgreenst | @FrankHarrell Thanks for the comment. Would you mind elaborating on what you mean by "logrank is a special case of Cox" and/or point me toward some reading? I'm new to CPH models, so I'm still trying to learn the ins and outs. | |
| Feb 14, 2021 at 12:25 | comment | added | Frank Harrell |
The likelihood ratio $\chi^2$ test from the Cox model is the gold standard for the global $H_{0}$. The OP should not refer to logrank tests; stick with the full Cox model and recognize the logrank is a special case of Cox and is not needed. Consider simultaneous confidence intervals for as many hazard ratios as desired (see the R rms contrast.rms and cph functions). Reduce use of statistical significance testing.
|
|
| Feb 13, 2021 at 20:40 | history | answered | EdM | CC BY-SA 4.0 |