Even if no 2FA was involved at all, it's a good answer to the scenario you were posing.

I think plenty of people will have second thoughts when the password doesn't go.

The comparison here is using 2FA with external device, or putting 2FA codes into a password manager.

Any kind of experiment that doesn't involve 2FA at all is not relevant for this comparison.

The anecdote provides evidence for people that are initially fooled by a phishing attack but aren't fooled enough to manually copy-paste credentials when autofill doesn't work.

Your argument about 2FA depends on how many of those people there are.

Therefore the anecdote is quite relevant, indirectly.