Specific comparisons in binomial GLM
- From: Stats Wolf <stats.wolf@xxxxxxxxx>
- Date: Mon, 7 Apr 2008 12:57:08 -0700 (PDT)
Hi,
When performing GLM for a binomial variable, I receive a result that a
factor does or does not influence the variable. If it does, I can also
easily compare the groups with the reference group. However, I can't
do that for other groups. As far as I know, I shouldn't simply change
the reference group and perform several analyses in a row. Or maybe I
can? Is there any way to compare two (or three or more) specific
groups I am interested in? For your information, I am against the
classical multiple comparisons in ANOVA and perform the specific
comparisons based on the standard error of the difference estimated
with the pooled variance--no protected LSD and Tukey's HSD and such
stuff; is there anything of this sort in GLM?
Many thanks in advance,
Wolf
.
- Follow-Ups:
- Re: Specific comparisons in binomial GLM
- From: Stats Wolf
- Re: Specific comparisons in binomial GLM
- Prev by Date: Third CEU Summerschool on Advanced Statistics and Data Mining (June 30th-July 11th, 2008)
- Next by Date: benefits of longitudinal study
- Previous by thread: Third CEU Summerschool on Advanced Statistics and Data Mining (June 30th-July 11th, 2008)
- Next by thread: Re: Specific comparisons in binomial GLM
- Index(es):
Relevant Pages
|
