Great - thanks. Yup, group 1 is young. So [1 -1] tests Y>O and [-1 1] tests O>Y... and I guess that would be a T-contrast since it's directional.

I suppose I could also set up an F-contrast like:
1 -1
-1 1

To test both Y>O and O>Y

On Wed, Aug 12, 2015 at 1:15 PM, MCLAREN, Donald <[log in to unmask]> wrote:

As with any analysis, you should start with the null hypothesis. In your case, the one you are probably most interested in is the difference between groups.

Ho: Y=O

convert to be equal to 0

Ho: Y-O=0 or O-Y=0

then use the coefficients as the column weights

Assuming column 1 represents Y and column 2 represents O, the contrasts would be:

[1 -1] or [-1 1]

Hope this helps.

Best Regards, 

Donald McLaren, PhD

On Wed, Aug 12, 2015 at 10:30 AM, Joelle Zimmermann <[log in to unmask]> wrote:

Hi Experts,

I'm setting up a contrast for a second-level model, where I have 2 groups (young and old), and set up a 2 sample t-test.

I'm wondering how I can set up the contrast for this comparison to make sense? (ie what numbers can I put in in the contrast manager?

Thanks,

Joelle