1 of 1 people found this helpful
There might be actually 2 parts to this problem - (1) Ranking based on the actual vote shares by candidates (rather than just raw vote counts), and (2) As you mentioned, filtering down to a candidate without breaking the % of total.
I've attached a solution that should work for both parts to this problem:
(1) I computed a RANK([Shares vote]) in order to create the suitable rank. Nesting the table calculations actually allows us to still compute each level independently with different Table Calculations: Addressing and Partitioning. For the [Share votes] we will still compute this across candidates for each state, whereas the RANK() is going to be computed across states for each candidate (since we are now ranking the states rather than the candidates). I would open up the "Edit Table Calculation" dialog to see how this is done with the field drop down.
2) Rather than Filter we want to enforce the quickfilter like a Hide function. There's a trick to doing this as a LOOKUP() function against [Candidate]: LOOKUP(MAX([Candidate]),0)
The calculation actually will be handled as a table calculation and if applied as a filter, would effectively hide each candidate rather than filter them out. This means it only affects what is displayed and not the % of total computed by our [Shares vote] field.
Thank you Wilson, this is genius!