No - not quite. I think that you are probably interested in the interaction between sleep and memory. In this case, your design matrix, would include columns for the main effects of sleep and memory as well as a third column for their interaction. The interaction column is simply the elementwise multiplication of the sleep and memory column. Your contrast would then be either a 1 or -1 placed on the interaction column and 0's elsewhere. You may also need to include a column of 1's.

Using a contrast of -1 and 1 will test whether the slopes of the memory and sleep are different. Probably not what you want...

Andrew

Originally posted by Nicholas Riccardi:

Hello,

I am using NBS to look at sleep disturbance and memory (both are continuous measures). If I specify the contrast as 1 -1 for sleep disturbaces and memory, respectively, is this identifying a subnetwork that is simultaneously associated with higher sleep disturbance and lower working memory? Is this reasonable/interpretable?

Thank you,

Nicholas Riccardi