Hi @LikelihoodFrog1,
From what I understand, you may be interested in assessing the variability of your responses for the experimental runs regarding their locations ?
One easy solution could be to add a column "Location" (Data Type : Character / Modeling Type : Nominal / 3 Levels : Top, middle, bottom) with the column properties :
- Design role : Random block
- Factor changes : Easy (unless there was a randomization restriction in the design ?)
- Value order (to order the levels if needed)
Then, you can use this column as a factor (random effect) in a mixed model analysis, to assess if the location can be accountable for a certain proportion of the responses' variability. If you want to see the change in responses' means based on this location factor, you can set the design role to "Blocking" (it will be taken into consideration as a fixed effect in the model).
If you need more guidance, you can perhaps provide more context, infos on the factors, and an anonymized dataset so that we may provide more relevant advices ?
Hope this first answer will help you,
Victor GUILLER
L'Oréal Data & Analytics
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)