The link below is a pretty awesome, counter intuitive use case of Bayes Therom that I saw which made me question statements like this. What the actual makeup of Scientific American's reader base is is probably lower than 50% but it isn't 0% either. That's a lot of wriggler room for me to be technically assertion that yeah, they probably didn't lose half of their reader base. But is it possible they lost 1/3, 1/5, 1/10? Those all affect the bottom line of a company like this in much different ways. I'd assume that the leaders made an educated decision, and weighed their pros/cons.
Also, just because you voted for Trump doesn't mean you'd unsubscribe to SA just because they endorsed the other candidate. What I'd like to read is some analysis that SA did on their readership base to determine how many people they'd lose based on this decision and then the results of their experiment. I mean what better way for a scientific journal to make a decision like this than to also follow their hypothesis up with some analysis.
I have nothing meaningful to contribute to whether or not they should/shouldn't have endorsed Biden, so I'll just say, it is not a zero consequence decision.
https://paulvanderlaken.com/2020/01/21/bayes-theorem-probabi...