Definition

Hypothesis

A specific, testable prediction about the relationship between variables — falsifiable in principle, stated before the data are seen.

A hypothesis converts a research question into a claim the data can contradict: "remote workers report higher productivity than office workers" (directional), or at minimum "the groups differ" (non-directional). What makes it scientific, in Popper’s sense, is that evidence could refute it.

Timing is integrity: hypotheses stated after seeing the data and presented as predictions (HARKing) corrupt the statistics that assume otherwise. Preregistration exists to keep the line between confirmatory and exploratory analysis visible — both are legitimate; blending them silently is not.