Sarah,
In SPM, a t-test is typically one-tailed. If you are interested
in a two-tailed t-test, you can define the same contrast as an F-test. This is
because a square of t-test statistic (df=N) is an F-test statistic (df=1,N). This
means both highly negative and highly positive t-scores will manifest as large
F-scores.
-Satoru
From: SPM (Statistical
Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Pulaski,
Sarah (NIH/NINDS) [V]
Sent: Monday, November 16, 2009 1:17 PM
To: [log in to unmask]
Subject: [SPM] One Sample test - one-tailed or two tailed
Hello,
I
was wondering if anyone could help me figure out what is likely, a very simple
question. I have been using SPM5 to analyze lesion data in one sample of
patients, correlating their brain atrophy with behavioral data. I have set my
contrast code as 0 1 0 0… (and so forth). Is this analysis a 1-tailed
t-test or a 2-tailed t-test. Is it even possible to do a 2-tailed t-test in
SPM?
Thanks
so much,
Sarah
Sarah
Pulaski, M.A.
Cognitive
Neuroscience Section
National Institute of Neurological Disorders and Stroke
Building 10; Room 7D48
MSC 1440
E-Mail: [log in to unmask]
Work Phone: 301-496-8163
Work FAX: 301-480-2909