Hi all,
I am struggling to understand when, using proc phreg to produce a frailty model, one would use the "unadjusted" p-value vs. the adjusted. See, e.g., https://documentation.sas.com/doc/en/statcdc/14.2/statug/statug_phreg_examples11.htm
Output 86.11.6. In this example, the adjusted p-values are cited in the text. However, I compared proc phreg frailty model results those produced from the survival package in R, and the results were closer to those from the "unadjusted" p-value. I understand that the adjusted p-value is just the Wald test with df adjusted using the Wald Adjustment, and I've studied Therneau and Grambsch (2000) and the formula for the Wald Adjustment, but from a practical point of view, I can't figure out why the Wald Adjustment is necessary, particularly since the R output is more similar to the unadjusted case.
Can someone explain the practical value to using the adjusted p-value? For those with experience fitting a frailty model, which result have you used when you do have a large frailty REML estimate?