Bloom's famous 2 sigma tutoring study was a 3 week intervention on two sub-topics that were new to students - cartography & probability.

The intervention was assessed using unstandardised researcher-designed tests.

You cannot make big generalisations from such a narrow & small study.

I would prefer it if people said “you know what, that Roger Ascham did a great job with Elizabeth I, we should have more tutors”. That’s more intellectually honest than distorting standard deviations.