Would anybody more knowledgeable in the topic help me understand the results of the first graph? Paper is here: https://sci-hub.se/https://psycnet.apa.org/doi/10.1037/hea00... (I haven't finished reading it) Seems the authors claim it's statistically significant, but does it matter if it's clinically significant? 1mm difference seems pretty small, and they conclude that it suggests that it 'can harnessed to improve treatment outcomes'. I'd have understood if the conclusion were that effects exist at all, but not if the claim is that it's useful for something else.
If you want to create a genuinely effective treatment, you want to maximize the placebo effect so much as possible - so that anything above that is 100% driven by the medicine.
Though in practice I expect it is probably already exploited in the opposite direction. By minimizing the effect of the placebo (or even pushing it into nocebo territory) pharmaceutical companies can construct studies that are much more likely to show statistically significant effects that do not exist. There's some seriously perverse incentives in play when a few percent this way or that is the difference between losing millions, and earning billions.
