Dxy value difference in log-logistic AFT survival model

1

Hi,

I’m getting two significantly different Dxy values from the model summary output and the validate() function for a log-logistic AFT model. In the model summary, the reported Dxy is 0.799, whereas the validate() function (with B = 300, or even B = 1000) gives Dxy values of index.orig = 0.3376 and index.corrected = 0.327 . Why is there such a large difference between these reported values?

Screenshot 2025-05-16 at 11.45.32 am
Screenshot 2025-05-16 at 11.45.32 am976×106 9.85 KB

Screenshot 2025-05-16 at 11.45.16 am
Screenshot 2025-05-16 at 11.45.16 am288×242 5.4 KB

Many thanks,
M

2

Please post this under the special topic for the rms package, which is RMS Discussions - modeling strategy - Datamethods Discussion Forum and I’ll remove it here. Post as a reply at the very end of that long topic. Ignore the chapters at the top of that page.

3

Thanks, I’ve posted this as a reply at the end of the long thread.

Kind regards
M