Can we talk about science? Very generally? I teach a 4th year undergraduate honours seminar in which we discuss this, but recent experience suggests it bears wider treatment.

1. There are two types of studies: observational and manipulative. In the first you look, but don’t touch. In the 2nd, you change something, preferably with a control or to compare among treatment levels.

2. There are also two kinds of science: descriptive and hypothesis-testing (HT). Hypothesis here refs to biological (or scientific more generally) hypothesis (i.e. not statistical).

3. Type of study is crossed with type of science; all four combinations exist. You can observe a system to describe it, or to test a prediction of an hypothesis. And you can manipulate something to describe what happens, or to test a hypothesis.

Aside: Note I used both 'a' and 'an' hypothesis on purpose so nobody (or everybody) can get worked up about that one.

4. In both types of study, you want to know if any ‘pattern’ in the results is real (i.e. is something beyond random sampling variation). This is statistical hypothesis testing, which is very important but not to be confused with HT science.

...That is, both descriptive and HT studies can, and usually do, employ statistical hypothesis testing, so the latter does not therefore denote something is HT science.

5. Descriptive and HT science are both valuable; hypotheses often originate from descriptions, and more generally science proceeds via a back-and-forth between these. Descriptive science should neither be denigrated nor dressed up to appear as HT science.

Finally, sometimes i find it so frustrating that you cannot edit a tweet to correct a typo.