Timeline for I think coxph post estimation with big data - confused about the results
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 24, 2022 at 12:48 | comment | added | tchoup | Thank you so much! | |
| May 24, 2022 at 12:48 | vote | accept | tchoup | ||
| May 23, 2022 at 20:55 | comment | added | EdM |
@tchoup the standard coxph() function has an optional argument robust = TRUE that will use a jackknife-type approach to estimate the coefficient standard errors. That doesn't depend on the assumptions of asymptotic normality, proportional hazards, etc that underlie the usually reported Wald-type standard errors. See Section 7.2 of Therneau and Grambsch. Even things like incorrect functional forms of covariates and omitted covariates affect those "robust" standard errors less.
|
|
| May 23, 2022 at 20:40 | comment | added | tchoup | I was hoping you could explain an answer from one of your links a little more for me: The first link has three solutions, and suggests using robust error estimate. I don't really understand what that is in regards to a cox ph. | |
| May 14, 2022 at 19:21 | history | answered | EdM | CC BY-SA 4.0 |