[R] multiple log-rank test for adjusted P value

David Winsemius dwinsemius at comcast.net
Tue Dec 27 20:14:07 CET 2011


On Dec 27, 2011, at 12:05 PM, Cheung Zihong wrote:

> Thank you  for your reply.
>
> I didn't know what adjust methods can apply to log-rank tests.Or the  
> kind of  tests are just not relevant? SAS seems to use sidak
> method to do  pairwise log-rank tests. which one is sidak method in  
> p.adjust methods?
> p.adjust.methods
> # c("holm", "hochberg", "hommel", "bonferroni", "BH", "BY",
> #   "fdr", "none")
Multiple comparisons adjustments have their own "rules of engagement"  
and they do not generally come from the test-statistic used but rather  
from the design and goals of the study. Since "SAS" is not a sentient  
being, much less a statistical consultant, appealing to SAS's  
"choices" for statistical consultation would seem unwise. Instead,  
since you are feeling unprepared for this task, you should describe to  
your consulting statistician how your data was c=gathered and what  
your goals were.
>
> 2011/12/27 David Winsemius <dwinsemius at comcast.net>
>
> On Dec 27, 2011, at 9:52 AM, Elliott wrote:
>
> Hello R users:
> I wonder if there are functions for multiple log-rank tests with  
> adjusted
> P-values. I found survdiff( ) only give pooled tests but
> without pairwise ones.
>
> Klein and Moeschberger suggest Bonferroni as a starting point. Why  
> not use p.adjust which would be a bit less conservative?
>
> David Winsemius, MD
> West Hartford, CT

David Winsemius, MD
West Hartford, CT



More information about the R-help mailing list