Date:         Wed, 23 Feb 2000 11:48:05 -0600
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: Log-rank test

SPSS offers two other similar kinds of tests. All of these can be used as
omnibus tests, and as pairwise comparisons tests. None offer the ability to
compare one group vs. a combination of others. I've never seen any
multiplicity corrections other than general probability based ones, such as
Bonferroni or Sidak, which you'd have to implement "by hand" by using a
smaller critical alpha level.

David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.

> -----Original Message-----
> From: CREV - Statistics [mailto:ricerche.crev@MAIL.REGIONE.VENETO.IT]
> Sent: Friday, February 18, 2000 3:56 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: [SPSSX-L] Log-rank test
>
>
> Hi!
>
> I'm using the survival analysis log-rank test to compare mortality in
> different groups of my study population.
> I have four groups, and the test shows a significant
> difference between
> them (p=0.0004).
> My question is: how can I compare subgroups of this population (for
> example, group 1 vs. group 3, or group 1 vs groups 2+3)?
> Shall I use again
> the log-rank test restricted to these subgroups, or does it
> exist something
> like post-hoc tests for analysis of variance?
>
> Thanks in advance for your help and suggestions,
>
> Manola Bettio
> ricerche.crev@mail.regione.veneto.it
>
>
>
> Servizio di Posta elettronica della Regione del Veneto
> Per le regole e le configurazioni consultate:
> http://www.regione.veneto.it/informatica/mail.asp
>