My interpretation of this description (and what would be useful) is that TOTAL( MEDIAN([val]) ) would calculate the MEDIAN of val across all records in the source dataset.

If that was the case, I should be able to plot:

MEDIAN([val]) / TOTAL( MEDIAN([val]) )

and get a result for each row that is the row’s MEDIAN(val) scaled by the entire dataset’s MEDIAN(val). That would be quite useful, to normalize all the row values.

But instead I get tiny values from the above formula – the largest is 0.017.

So either TOTAL() is broken, or its description is wrong, or I don’t understand something else…

I believe so, but I’m not actually sure myself what the definition of ‘the entire dataset’ should be (e.g. when more than one db is in the source and or filters are being used)