Hi everyone I am dealing with a tricky situation.
I will try to explain quickly my problem.
I am working with a tool comprise of 7 items. Each item have 3 status ( bad, medium and good).
In my study I use this tool to generate 3 categories in my population (BAD, MEDIUM and GOOD). The thing is I only use the number of good (out of the 7 items) to generate these 3 categories.
The rule is someone with : 0 to 2 good —> BAD; 3 to 4 good —> MEDIUM; and 5 to 7good —> GOOD !
However while doing this I focus on the number of “good” only ! I assume that “bad” and “medium” is the same at some extend. This leding to say that someone with 7 “medium” have the same risk as someone with 7 “bad”. This bring a heterogeneity in the group “BAD” and even “MEDIUM”.
So in my statistical model I computed a number of “medium” (variable number_of_mefium) for each subject and adjusted for it.
Here is the model :
Group = var1 + var2 + var3 + number_of_medium
(It’s a logistic polytomous model; BAD is the reference class)
But my point is as the number of possibility of having “medium” differ across my 3 groups. For instance in the group GOOD one can only have 0 to 2 chances to have an item “medium”, because he already have 5 out of 7 items “good”. In contrast in the group BAD one can have from 0 to 7 “medium”.
Is that a good idea to add this outcome in my model ? The aim is to consider that the more we have ideal the better we got !
I could use the tool as a score but it’s important to have 3 groups for this study.
Any help? Suggestion?
Thanks in advance