[wenc]

RCTs are not the only way to obtain evidence of causality.

In causal inference, limited causal claims can be made through observational studies (where no control is done), such as retrospective natural experiments [1]. These methods are of course limited by assumptions and by how closely they match reality.

Note: while a controlled experiment is the gold standard, it is not always feasible or necessary. We don't need to do RCTs on parachutes to prove their efficacy for instance. [2]

--

[1] https://en.wikipedia.org/wiki/Natural_experiment

[2] https://www.scientificamerican.com/article/volunteers-jumped...