Gaby,
I can't think of a better way to do it.
Including the zeros won't change the mean estimate of the covariate
effect but it will change the SD. But if your groups are similarly sized
this should be a small change.
If your covariate is of no interest, ie. a nuisance regressor, then
including it as you describe will achieve what you desire ie. subtracting
it from the data so this effect does'nt drive between group differences.
If your covariate is of interest (ie. you want to make an inference about it -
the small change in SD may be of concern) - then you could redo the analysis with
a new model containing just that groups data.
Best,
Will.
Gaby Pell wrote:
> Dear Dr Penny,
>
>
>
> I hope you don’t mind my forwarding the recently posted message since I
> know you must get many requests each day, but I forgot to add your name
> to the start of the message as a potential respondent.
>
>
>
> Please discard if already seen on list!
>
>
>
> *********************************************************************************
>
>
> Dear SPMers,
>
>
>
> I would like to carry out a t-test (or regression) analysis between 2
>
> groups with the addition of a covariate variable. However, the covariate
>
> is only relevant for one of the groups, and has no meaning for the other
>
> group. Other than setting the covariate value to 0s for this group which
>
> does not seem ideal, is there a better way to include the covariate in the
>
> design matrix.
>
>
>
> Thanks,
>
>
>
> Gaby
>
>
>
>
>
> ***************************************
> Gaby Pell, PhD
> Brain Research Institute
> Ground Floor, Neurosciences Building
> Austin & Repatriation Medical Centre
> Banksia Street
> West Heidelberg,
> Melbourne,
> Victoria, 3081
> Australia
> Tel: (+61 3) 9496-2868 (W)
> Tel: (+61 3) 9527-6441 (H)
> Fax: (+61 3) 9496-4071 (W)
> Email: [log in to unmask]
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|