Hacker News Reader

If they were volunteers the study would have a very strong self selection bias and therefore not be scientific.

That might be true for some studies, but the generalization is not correct. It depends on what is being studied.

Care to give an example where a study with strong self selection bias is still scientific?

Nearly every drug test. E.g., the researchers ask for volunteers suffering from condition X. Those volunteers are then tested with a new drug.

Why would you think the fact that the subjects are volunteers would affect this?