I think there’s a much simpler effective synonym for “statistical arguments not based on data:” opinion. But is opinion so bad?
We cyclists who have had to make our bikes more satisfactory by adjusting our saddles, choose large and small adjustments as we see fit at the moment. For me, those are based on my opinion. Sometimes my opinion is based on canny consideration of what I feel and how my last magic fix worked or didn’t. Sometimes I just grab the wrench and change something, and deal with the result (ouch).
I really don’t see any practical way to document the process and hence produce data. We also don’t have a common language to describe comfort problems and the subsequent lack of a problem. I think asking for data is an impossible request.