Bruce has given the direct statistical answer that uses the fact
of matching.
When the matches were constructed from factors
like age and sex, etc., it is usually a better practice to
ignore the matching when doing the test, and control
for the matching variables as covariates or factors,
in an ANOVA with 2 groups.
If the controls were from different sources (which is
a more ambitious design), the setup would preserve
the three groups and give the test as a special contrast,
Case vs (Control1+control2)/2 .
--
Rich Ulrich
Date: Thu, 12 Jan 2012 15:19:52 +0000
From:
[hidden email]Subject: Compare Means
To:
[hidden email]Dear all,
I would compare two means between cases and controls taking
into account that I have matched 1 case
to two controls. How i can do it with pasw statistics 17
Thanks in advance
Jan
Locked
