Hi Carlos -
You should not model the baseline.
You should only model the active conditions.
These will then be compared against the implicit baseline in the GLM
analysis.
So you need one EV for OSPAN and one EV for Arith.
The contrasts are then
1 0 for OSPAN > Base
0 1 for Arith > Base
1 -1 for OSPAN > Arith
-1 1 for Arith > Base.
Cheers
T
On 28 Jan 2008, at 20:41, Carlos Faraco wrote:
> I am setting EVs and just wanted to make sure I am doing this
> correctly.
>
> The study consists of an OSPAN (working memory task) portion), an
> arithmetic
> portion, and a baseline. It goes as follows.
>
> -----------I------------I ---------I
> ---------------
> OSPAN 30s Response 15s Arith 30s Repeat
>
> ----------I ---------I
> Base 30s Base 30s
>
> So, the task w/ the response and the arithmetic are done 3 times
> total, and
> baseline is done 6 times. I am not interested in including the
> response
> condition in the analysis (this is just to verify they are paying
> attention,
> at least for now).
>
> I have setup the EVs as follows:
>
> For the OSPAN: Skip = 0s, Off = 105s, On = 30s, Phase = 105s, Stop
> = -1.
>
> For Baseline, I set up two EVs because the intervals vary, so the
> Baseline
> after the OSPAN is: Skip = 0, Off= 105, On = 30, Phase = 60, Stop =
> -1.
>
> Baseline after Arithemtic is: Skip = 0, Off = 105, On = 30, Phase =
> 0, Stop
> = -1.
>
> Arithemtic = Skip = 0, Off = 105, On = 30, Phase = 30, Stop = -1.
>
> Is this correct? Also, when I set up my contrast for OSPAN vs.
> Baseline, I
> choose OSPAN = 1, Baseline 1 = -.5, and Baseline 2 = -.5. This
> should make
> it as if there was just one baseline, correct? Or are the statistics
> affected differently somehow?
>
> Lastly, I am setting the highpass filter at 135; this is the total
> time for
> one cycle. Is this too high though?
>
> Thanks.
>
|