Hi everyone, I am new here so I hope I post this in the right Sub! Please let me know if there is a better Sub for it- I appreciate your help!! Here's my question:

I am working on a complex study design and I am not able to find a good solution for counterbalancing my dependent variables (same instrument, different versions of it)!

Basic design is as follows:
Participants will be randomized into two groups: (I) early operation vs (II) routine conservative treatment and emergency operation. For both groups, it will look like this:

I Baseline - Intervention - Post (4 Weeks) - Fup (6 Months) - Fup (12 Months) etc
II Baseline - Baseline - Baseline - Intervention - Post (4 Weeks) - Fup (6 Months) etc

So actually, the time point of surgery is randomized (early vs emergency), the intervention itself is identical.

So now I am wondering about my dependent variables. We will measure cognitive capacities at all time points. We will use two instruments. One Instrument has 2 alternative Versions (Mini Mental State Examination). The other one has 6 (SKT, a German Instrument). I want to use all versions in order to avoid order effects. I don't know how to assign the different versions to the participants and time points. I have following options:

1.
(balanced) latin square, irrespective of randomization, starting at baseline. I could then end up with a disbalance of instrument versions in the experimental vs control group at any given time point (by chance, as groups are randomized).

2.
stratified according to condition (counterbalancing within control group and within experimental group. this would be ideal, but it's not possible because we administer the baseline test before randomization. also, it's to complicated for our partner clinics)

3.
fixed order for all participants (e.g., all starting out with version 1 at baseline, followed by version 2 at post etc. this would be perfect for between group comparison, but problematic for comparisons from time point to time point across both groups, as any effect could be due to the specific Version of the Instrument)

So, I've read loads of experimental studies, most are vague about the exact procedure in a case like this. Do you have ideas on the best solution? I am leaning towards latin square irrespective of randomization... maybe I should calculate the odds of ending up with a disbalance of versions in both groups on each time point?

THANKS a lot!!! Greetings from Berlin,
Hanni