Karen Lidzba wrote:
> Hi!
>
> I have a small group of 7 patients with two control-groups, the sujects of
> which are closely matched to one patient each by age, sex and IQ-data. To
> find out differences in activation patterns in a given task I thought of
> using a paired t-test with patient-control pairs.
>
> I know that the paired t-test was originally invented for repeated measures,
> but would my approach be acceptable, too?
>
That one patient is age, sex and IQ matched to one control subject is
sufficient justification for using a paired t-test. Then your data will
come in pairs.
Of course, they won't be matched for other factors that are
beyond experimental control. But given that you've made an effort to control
these factors i think a paired t-test will be more sensitive than a two-sample
t-test.
Of course, the alternative, is to enter age, sex and IQ as regressors (at the 2nd level
for fMRI, 1st level for PET) and model these effects out. But you've effectively
already done this by selecting the data.
Best, Will.
> Any suggestions appreciated!
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7478
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|