Pick two options, declare one desirable.

Pick some metric (or maybe a group of metrics), declare them desirable.

Find one (or more) of those metrics that is higher in the desirable option. Ignore all other metrics. Declare desirable option to be superior.

Stand back and watch while journalists who never even passed basic algebra spread headlines and articles declaring your desired option to be superior.

Reminds me of a paper that Chris Martinson went over where, according to one metric (some protein in the blood or something), Hydroxychloraquine did nothing. According to other metrics that weren't mentioned in the study's abstract for some mysterious reason (these being number of days in the hospital and continuing to stay alive), hydroxychloraquine showed a positive effect.