Dear Cross Validators,
I have a data set of the following structure (don't mind the numbers):
| subject_ID | Position_of_the_measure | Environment_condition | Variable_of_interest |
|---|---|---|---|
| 1 | A | 10.3 | 40 |
| 1 | B | 10.3 | 36 |
| 2 | A | 15.2 | 62 |
| 2 | B | 15.2 | 48 |
| 3 | A | 21 | 27 |
| 3 | B | 21 | 29 |
I want to investigate the effects of the Position_of_the_measure and environment_condition on my variable_of_interest.
Obviously the values are non independent (as paired by ID) so I figured linear mixed models would have been the go to with ID as random effects but I don't have multiple measures per ID per Position (this should have been thought in the sampling design but here we are).
I wonder if you guys have an idea about how to deal with this from there (working with A-B might work but I'd prefer not to).
I know that I'm pretty much asking for what could be considered as statistical non sens but wanted to know if I had miss something obvious?
Thank you for your time.
Here is an example of plotted data which might help you understand:
