Dear NBS experts,

Recently, I did some correlation analysis in NBS by using the YMRS scores as co-variate and found some significant correlation between resting state function connectivity and YMRS scores. Now I want to further identify that whether bipolar subtypes moderated the relationship. So I want to conduct a two-way interactions between subtype and YMRS score to see if there is some interaction effect (the bipolar sub types included mania state, euthymic state and depressive state).For example, we have 9 bipolar patients (3 mania, 3 euthymia, and 3 depression). For each subject, I have their YMRS scores. Can you kindly help me with the design matrix?  And what's more, how to build the contrast that manifest the interaction effect?

I would be so appreciated if any one could help.

Looking forward for your reply. Thanks so much,

Hua

Hi Hua,

First, 9 subjects is most likely inadequate to evaluate an interaction effect. A larger sample size is required.

The design matrix might look like this, where M, E and D denote your three subtypes and the YMRS scores for the 9 subjects are denoted with the letters [a b c d e f g h i]:

M1  1  1  0  a  a  0
M2  1  1  0  b  b  0
M3  1  1  0  c  c  0
E1  1  0  1  d  0   d
E2  1  0  1  e  0  e
E3  1  0  1  f   0  f
D1  1  0  0  g  0  0
D2  1  0  0  h  0  0
D3  1  0  0  i   0  0

To test for an interaction between YMRS score and subtype, the contrast would be [ 0 0 0 0 1 1] and the test would be "F-test".

Andrew

Originally posted by Hua Yu:

Hi Andrew,

That makes great sense to me! Thank you so much!

Hua