So they prefer values in the range 11 - 20? Does this also mean that they maybe prefer a single value like 15? Then they could use a quadratic loss function like: (y-15)^2. That would mean that deviations from 15 would be penalized more.
Do they have explanatory variables that could help...
Did you ask me? :)
I have seen a standard method is to take the middle of the interval but I don’t remember how to treat the open interval (the 5+ interval).
I was (like Dason) thinking about using the censored Poisson distribution. But maybe it does not fit.
Skip an other level of the three levels and you will have just two levels. That is your two factor experiment is redused to one factor.
To do factorial experiments is to do compromises. You can't get everything.
I am sorry but I don't understand exactly what Obh mean. My point was that when two data sets have different "spread" (but the same mean) then the WMW does not have the promised quality of an error rate of 0.05. The error rate will be larger. And I believe that most people interpret a...