I tried excluding records from respondents for which [Beer] AND [Pop_Tart] AND [Score] were empty of null using the IsEmpty() function, but this resulted in too low a record count, so I kept it as is
My solution. I added the bonus insights for Most Popular '10' rated combo and averages for each of beer, Pop Tarts, and combos. It would be interesting to do some predictive analysis with what beer someone would like based on their favorite Pop Tart.