Hello,

I want to make a design with two continuous variables like:

design=model.matrix(~ con_var1 + con_var2 , data=samples)

but my decide test gives me no results:

   (Intercept) con_var1 con_var2
-1          26      0      0
0           68    543    543
1          449      0      0

I dont understand why this is not working as it is with factorial variables.

If I do the separately it does work!

design=model.matrix(~ con_var1 , data=samples)

   (Intercept) con_var1
-1          37      4
0           52    536
1          454      3

design=model.matrix(~ con_var2 , data=samples)

   (Intercept) con_var1
-1          37      9
0           52    536
1          454      2

Thanks for any help!

Best regards,

Julian

It's not possible to say without more information, but I would guess that your two continuous variables are not orthogonal.

In more heuristic terms, if your continuous variables are not orthogonal, then they will both explain some of the same variation in your observed expression values. When you fit the individual models, each variable explains enough of the variation to achieve significance, but if you put both in the model, they have to 'share' the variation they explain, and neither gets enough to be significant.

Or it could be something else. Like I said, not enough information to say for sure.

Thank you, that indeed explains it! The continuous variables are measurements that correlate with each.

Best,

Julian