Hi Jenny,

Is zygosity only a nuisance in your model? If yes, and if you already included one EV per pair (regardless of zygosity), then zygosity can be removed from the model, and yet it will be taken into account by the pair-specific EVs.

So, just remove that final EV, and no worries as whether MZ or DZ will be taken into account.

When doing the permutations, remember to respect the exchangeability blocks. You may want to use a shuffling strategy similar to that used for HCP data: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/PALM/ExchangeabilityBlocks#EBs_for_data_of_the_Human_Connectome_Project

All the best,

Anderson


On 18 February 2017 at 19:00, Jenny Gilbert <[log in to unmask]> wrote:
Hello,

I am creating a GLM to model between weight dependent between twin differences across a sample of monozygotic and dizygotic twins. I am having trouble figuring out how to set up the matrix to include EVs for each twin pairs & zygosity without creating collinearity.

This is an abbreviated example of my matrix:
                Group   EV1                     EV2             EV3             EV4      EV5
                 1              BMI level       Pair1   Pair2   Pair3    Zygosity *(0=MZ, 1=DZ)
Input1       1          1                       1               0               0            0
Input2       1          -1                      1               0               0            0
Input3       1          1                       0               1               0            1
Input4       1          -1                      0               1               0            1
Input5       1          1                       0               0               1            0
Input6       1          -1                      0               0               1            0

My contrast of interest is:
                                    EV1       EV2       EV3             EV4      EV5
HighBMI > Low BMI       1            0            0              0           0

I think the collinearity warning is given since the pair EV and zygosity EV are the same for both twins in a pair.

Does anyone know a good way to set up theGLM  matrix to avoid this problem?

Thanks for your help!
Jenny