In a Philosophy Bites episode, John Worrall is interviewed about how trustworthy the experiments on which evidence-based medicine rests. Specifically, he discusses how suitable randomised controlled trials (RCTs) are for establishing causation. Worrall explains how RCTs are often used to make causal claims about whether an intervention works, while this might not be supported by the evidence. Ultimately, it comes down to the old problem of causation versus correlation.
It reminded me of an interview I did for Smart Drug Smarts and Jesse Lowler on causation and correlation (over skype), and I thought I’d share it here. A question discussed here is what kind of correlations we expect causal correlations to be. Should they be perfect correlations, as suggested by Hume, or just a form of regularity? Do we need to specify some ideal conditions under which the effect is guaranteed? A problem is to back this up empirically. Listen to the interview here (smartdrugsmarts) or here (vimeo).