Dear Kosha,
> I have a 2x2 (time x condition) within subject design.
> The time factor has two levels: pre and post
> The condition factor has two levels: Drug and Neutral
> In addition i also have two additional covariates
> which i want to adjust for.
> These are : age and
> Number of days since last use.
>
> To test for interaction effects,
> I set up a standard 2x2 repeated measures design.
> In the above design i would like to adjust for the additional
> covariates.
>
> Even though i am removing subject specific effects, is this
> adjustment possible in GLM? i did set up something similar
> while setting up paired T test as shown below based on some of the
> earlier posts i read.
>
> To adjust for age and Duration since last use, i demeaned the
> covariates, multipled them with the first EV and then entered them
> in the model as shown below (attached Design_paired). is this setup
> right for looking at Pre vs Post effect adjusting for age and days
> since last use?
>
>
>
> EV1 EV2-EV5 EV6:Pre EV7:Post EV8:Age
> -1 1.. 45.75 0 -3.25
> -1 0.. -38.25 0 2.75
> -1 0.. 25.75 0 0.75
> -1 0.. -33.25 0 -0.25
> 1 1.. 0 44.5 3.25
> 1 0.. 0 -39.5 -2.75
> 1 0.. 0 29.5 -0.75
> 1 0.. 0 -34.5 0.25
>
>
you cannot, and need not, include any regressor that has a single
value per subject when regressing out subject specific effects. The
subject specific regressor already accounts for age (as well as
anything else specific to this subject).
I am not really sure what type of variable "duration since last use"
is. Is that also something that is subject specific, or is it
something that is differs between the two conditions/occasions the
subjects were scanned under?
Good luck Jesper
|