Ok. So. I have a data set of 60 thousand values, distributed thus:
They are associated with 30 different aircraft types. I wanted to use ANOM to quickly see if any aircraft groups were significantly different than the overall mean. Since the overall population distribution of these values is skewed, I decided to apply ANOM with Transformed Ranks. The original data were all negative values, so without much thought I decided to flip them to positive values before conducting the ANOM:
10 aircraft emerge from the LDL-UDL range.
Then I tried ANOM with TR with the original negative-valued data:
I would have thought that the results would have been the same. Is there a simple explanation for why I wouldn't see the same results either way?
I just went through the steps manually and I get the same scores. It is a small example than your, but the transformed ranks are identical with it is the original data or the negated data.
Are you sure that there is no other change that you might have made, such as excluding rows?
