Hi - basically you can't have both of these things in the same model:
- You have a covariate (eg age) that has a different group mean in the
different groups, and you want to add this as a single confound
regressor in an attempt to model out the non-balanced nature of the
data. (Note that there is no guarantee that this will successfully
model out the confounding factor.)
- You want to test how the correlation between the data and the
covariate (eg age) is different in the different groups - so you have
a separate covariate for each group, with the covariate demeaned for
each before being padded by zeros for the other groups.
These two approaches aren't directly compatible with each other in the
same modelling - you need to choose one or the other.
Cheers, Steve.
On 18 Apr 2009, at 07:32, Mahinda Yogarajah wrote:
> Dear Experts,
>
> I am a little confused by previous posts as how to include
> confounding
> variables in a 3 group TBSS analysis. If you have 3 groups and are
> interested in correlations between FA/MD and a behavioural score,
> and how
> these correlations compare between groups, but there are confounding
> factors
> measured in all subjects (which vary within groups and between groups,
> though the between group variability is greater) such as age and IQ,
> how
> should they be included ?
>
> Should one:
> a) Include each confounder as one ev, demeaned across all 3 groups
> (would
> this account for between group variability in confounders only?)
> b) Split each confounder into 3 evs and demean across each Ev before
> padding
> with zeros (would this account for within group variability in
> confounders
> only?)
> c) Demean each confounder across all three groups and then split each
> confounder into 3 evs (would this account for within group AND
> between group
> variability in confounders ?)
>
> Thanks.
>
> Mahinda
>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|