I think yummifajitas means that if done in any field, it would be a problem, not that it is necessarily observed in every field.
Yes probably what he was saying.
Also, my understanding - not expert, likely wrong - is that it is sometimes done in physics (e.g analyzing data from sensors). But haven't heard any complaints (which does not mean they don't exist). Someone else would need to weight in as this is just supposition and hearsay :)
AFAIK this isn't much of a problem in CS (my field) and never heard math or physics people complaining...
Seems to only be problem in Social "Sciences" & Bio/Med where many (most?) results are statistical significance tests.