I'm trying to create a card where a user can switch between different aggregation groups for their x-axis. So if they select "Race/Ethnicity" they get something like this:

and if they select gender:

I successfully set up the variable:

And the BeastMode referencing the variable:

But when I try to graph using that BeastMode, the graph won't aggregate the groups correctly:

I figure it has something do with my sort, so I've tried sorting using the BeastMode itself. I tried creating a custom sort based on the Variable:

case
when NPS Grouping='Race/Ethnicity' then COUNT(DISTINCT Response ID)*(-1)
when NPS Grouping='Gender' then COUNT(DISTINCT Response ID)*(-1)
when NPS Grouping='District' then District
when NPS Grouping='Cohort' then Cohort
when NPS Grouping='Cohort Year' then Year number
when NPS Grouping='Fiscal Year' then Fiscal Year
END

I tried using a generic aggregating BeastMode (total responses), but none of them will get the bars to aggregate.

My data table does aggregate correctly:

Any ideas on what else I could try?