[mannschott]

> In school you do “experiments” and if the answer doesn’t match what is in the book you are marked down.

This is unsurprising and probably obvious. The purpose of the exercise is learning how to do experiments reliably so that at some later point when doing experiments to test a hypothesis you can have some confidence that the results you produce are actual data and not just noise.

When doing science class "experiments" of this sort perhaps it would illuminate things to students better by going up one level: the hypothesis under test is "student can perform experiment reliably". The putative subject of the "experiment" isn't the actual subject in inquiry: you are. On the other hand maybe this would just confuse students. (I've observed wildly varying capacity for abstraction and meta-cognition among the people I work with as a software developer so I suppose this is something that varies among individuals generally.)