Jen,
(1) A quick test for correlations
(homepages.gold.ac.uk/aphome/correlnotes.doc):
Fisher’s Z transformation of the correlation coefficient r is used.
This is not the Z transformation we use when standardising. It is
often denoted by r’
r’ = 0.5*loge{Abs([1 + r]/[1 - r])}
where Abs(x) is the absolute value of x (i.e., ignore sign). The
standard error of r’ is denoted sr’ and has a value of 1/sqrt(N - 3),
where sqrt(x) means the square root of x, and N is the number of
subjects involved in the correlation. We can calculate a z-statistic
for the difference between two r’ values:
Z = (r1’ - r2’) / sqrt([1/(N1 - 3)] +1/[N2 - 3])
Howell gives a worked example. We also use Fisher’s Z transformation
to test whether a sample correlation is equal to any specific value,
i.e. could the sample value we have obtained have come from a
population with a correlation of a particular value. Fisher’s Z can
also be employed to put confidence limits around a sample estimate of
a correlation.
(2) In SPM, you an get the t-value for the behavioral covariate and
convert it to an r using the following equation (imcalc would work):
r=sqrt(t*t/(t*t+df))
(3) Are you wanting to compare the correlations or the slopes of the
relationships? If the later is the case, then you want a two-sample
t-test (or flexible factorial model).
Two-sample T-test with 2 covariates (1 covariate for each group, DO
NOT MEAN CENTER).
Flexible Factorial Model (group factor and 1 covariate per group).
CONTRAST(s): 1 over one covariate, -1 over the other covariate
depending on the direction of the test you want to perform. The order
doesn't matter with an F-test.
Multiple-regression: constant term, covariate for all individuals,
group term (one for each experimental group(s)), covariate term (one
covariate term for each experimental group(s)).
CONTRAST(s): 1 or -1 over covariate for the group you want to test for
a difference. Sign determines direction on a T-test. Sign is
irrelavent for an F-test.
Hope this helps.
On Tue, May 19, 2009 at 12:19 PM, Jen Labus <[log in to unmask]> wrote:
> Surprisingly I am struggling at finding out how to determine group
> differences in a correlation. This seems like a fairly basic analysis but I
> have run across differing opinions on how to set this up. I am hoping
> someone my provide a very specific answer so that all SPM users can use the
> appropriate analysis.
>
>
> I have 2 groups and behavioral measure. I want to determine the group
> differences in correlations between the brain and the behavioral measure at
> the second level in SPM. What is the proper model to set up to get this
> answer in SPM?
>
--
Best Regards, Donald McLaren
=====================
D.G. McLaren
University of Wisconsin - Madison
Neuroscience Training Program
Office: (608) 265-9672
Lab: (608) 256-1901 ext 12914
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain
PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED
and which is intended only for the use of the individual or entity
named above. If the reader of the e-mail is not the intended recipient
or the employee or agent responsible for delivering it to the intended
recipient, you are hereby notified that you are in possession of
confidential and privileged information. Any unauthorized use,
disclosure, copying or the taking of any action in reliance on the
contents of this information is strictly prohibited and may be
unlawful. If you have received this e-mail unintentionally, please
immediately notify the sender via telephone at (608) 265-9672 or
email.
|