Time isn't collinear because it hasn't been measured. Time is not known for the control group, not zero. So you cannot use the time variable. You can examine the other variables, but since you have no data for the controls you can only examine the effect of time within the intervention groups.

Hi Ronán Michael Conroy ! Thank you for your answer. Not including the base level in my model would however practically eradicate the explanatory power of the model. The control group didn't experience the exposure, a more correct way would probably that Time is NA for Group = Control. This however does not solve my problem unfortunately.

Hello Hans,

It's not clear what specific model you're interesting in trying to build or evaluate. Could you possibly explain what that was intended to be?

If group membership is an IV (independent variable), then what Ronán Michael Conroy says is correct: you cannot include time as a variable in the same model, since you didn't (or weren't able to) collect that information for the control group. So, any model built either excludes the control group or excludes time.

Good luck with your work.

I'm trying to build a regression model. That argument makes sense, but no exposure would have meant that I couldn't have collected this data in the first place.

Did you try without base?

What was your software?

Do you have enough sample size (enough error degrees of freedom)?

Hello again, Hans,

I'm still a bit unclear as to where you're headed with the data.

Could you elaborate a wee bit more? What are the IVs you wish to incorporate in the model, and what is the DV? Is this a growth model? What specific research question(s) are you hoping to address via this analysis?

Good luck with your work.