How to use a Linear Mixed Models (LMMs) to compare a fixed factor with 3 levels, which differ in size

Question

I want to run an LMM with a 3-level fixed factor "block_condition" (levels: A,B,C) and 2-level block_number (1,2). SubjectID was also added as a random factor to the model.

Now the problem is that I have for block_condition A and B - two repetitions/levels (A1, A2, and B1, B2), but in condition C, I have only 1 repetition/level (C1), and I don't know how to run an LMMs with one condition that has level less when I try to set the following LMM:

lmer(DV1 ~ block_number * block_condition + (1|sub), data = dfT1).

Is there a way to run such LMM? or a legitimate way to complete/replicate the missing information in condition C - for the sake of the specific comparison?

Thanks in advance, G

Answer 1

Hi Gil,

Sorry to see no one has addressed this yet. It sounds like a difficult problem and without knowing more about the study my advice is limited. There are probably ways to force an analysis (you might put "NA" as C2 and tell the model to ignore NA values), but I would recommend reconsidering your approach. For instance you could have two models, one comparing A1,B1, and C1, and the other comparing just A2 and B2. Again, it's hard to know what is appropriate without more knowledge on the study.

Feel free to message me if you're still working on this!

Best,

Nicholas