Dear Tom,
Thank you, it helped a lot! I have some follow-up questions:
>> 2. Demeaning the values in randomise: There are two options:
>> - to add an additional EV (with "1" only) and add -D in the command line
>> - to enter already demeaned values into the design.
>> Is it correct?
>
> Not quite. If you model the mean in the design matrix (e.g. w/ a EV of 1's)
> you *do* *not* demean your data. If your design matrix is centered (all EV's
> are mean zero), then it is crucial that you use the -D option to center the
> data.
What I did is: I demeaned the values in EV1 and also used D option fir
randomise (somehow I did it correct:).
Then, I am not sure if I should model my mean or not? I did it both with and
without the EV with ³1² only, results look roughly similar, what does it
change and what does it mean?
In group comparison, I dont assume the same variances. In correlation within
one group I may. Is it connected to this mean modeling?
>
>
>> 4. Is the design below correct for the correlation of TBSS data with 1 EV of
>> some demeaned parameter x? Should I run it with -D or without?
>
> If the EV is demeaned (and there's no other EV's) you must demean the data,
> with the -D option.
>
So is this single EV all I need (yes, my values are
demeaned and I used D option) or do I need to model my mean ? And then I
have 2 EVs, does it change anything with demeaning?
>
>> 5. Is this a correct design for correlation of TBSS data with parameter x,
>> with parameter y as a covariate ("controlling for y")? Do I enter other
>> parameters in randomise as before (eg. t-value threshold, etc.?)
>> /NumWaves 2
>> /NumPoints 17
>> /PPheights 6.651000e+03 8.000000e+00
>>
>> /Matrix
>> -1.492000e+03 0.000000e+00
>> 5.159000e+03 1.000000e+00
>> 8.620000e+02 -4.000000e+00
>> -1.361000e+03 -4.000000e+00
>> -9.570000e+02 0.000000e+00
>> -1.162000e+03 -1.000000e+00
>> -1.444000e+03 0.000000e+00
>> 6.610000e+02 -1.000000e+00
>> -4.190000e+02 0.000000e+00
>> -9.280000e+02 0.000000e+00
>> -1.287000e+03 2.000000e+00
>> 3.091000e+03 -2.000000e+00
>> 2.097000e+03 -1.000000e+00
>> -1.013000e+03 4.000000e+00
>> -1.038000e+03 -1.000000e+00
>> -6.870000e+02 2.000000e+00
>>
>> /ContrastName1 x
>> /ContrastName2 y
>> /ContrastName3 x(y)
>> /NumWaves 2
>> /NumContrasts 3
>> /PPheights 7.017095e+03 8.046171e+00 8.045551e+00
>> /RequiredEffect 5.870 5.827 5.827
>>
>> /Matrix
>> 1.000000e+00 0.000000e+00
>> 0.000000e+00 1.000000e+00
>> 1.000000e+00 1.000000e+00
>>
>> -7.600000e+01 -3.000000e+00
>
> Yes, this is all OK, but, again, assuming that both EV's are mean zero, you
> must deman your data.
>
(I did with D)
>
> Also, the third contrast "x(y)" is probably not meaningful... it's testing if
> the average of the two regression coefficients are zero.
>
Hmm, I in fact thought this is the main contrast to look at:(. So is it then
the first contrast that gives me info I am looking for (how TBSS values
correlate with x, controlling for y)?
Thank you again very much!
Best,
Aga
|