Where does hypothesis collide?

A statistical hypothesis is a precise, testable claim about a population or the numbers that describe it, framed so that data can weigh for or against it: the null hypothesis states there is no effect or no difference, and the alternative states there is one. Neyman and Pearson's 1933 On the problem of the most efficient tests of statistical hypotheses built the framework of testing one such hypothesis against another and choosing a rule that controls the two ways the test can go wrong. The referent is a formal proposition about a distribution, accepted or rejected by a decision rule applied to a sample, not a scientist's tentative explanation, a premise granted inside a proof, or a famous unsolved conjecture.[1]

Unit

One stated claim about a population parameter, cast as a null against an alternative

Recovery means

Not applicable

Measured by

Judged by a test statistic and a decision rule that fixes the rate of falsely rejecting a true null

In the methodology of empirical inquiry, a hypothesis is a proposed explanation put forward tentatively so that its consequences can be drawn out and checked against observation. Popper's 1959 The Logic of Scientific Discovery cast science as a cycle of bold conjectures and attempted refutations, in which a hypothesis earns standing not by being proved but by surviving genuine attempts to show it false. The referent is a fallible guess about how the world works, held open to empirical test, not a formal claim about a distribution decided by a rule, a premise granted for the sake of an argument, or a settled or open question of pure mathematics.[2]

Unit

One proposed explanation, stated so its observable consequences can be tested

Recovery means

A hypothesis that fails a test is rejected or revised, and one that survives is only provisionally kept

Measured by

Tested by deriving predictions and comparing them with observation or experiment, seeking a refutation