4 ms·
I believe that most discussions about the placebo effect are muddled by whether regression-to-the-mean effects are included or not. [1] When one is comparing a
by heisenzombie 7y ago
I believe that most discussions about the placebo effect are muddled by whether regression-to-the-mean effects are included or not. [1]
When one is comparing a treatment with a "placebo", it makes sense to include "regression-to-the-mean" effects, since you're obviously looking to do better than that with your treatment.
However, when people ask "What are the biological causes of the placebo effect?", they should really be excluding "regression-to-the-mean" effects since there's no causal link between the "placebo treatment" and the regression part of the outcome.
If one looks for a placebo effect and EXCLUDES regression-to-the-mean, the effect sizes tend to be small to nonexistent, which makes discussion about their physiological causes a lot more shaky.
[1] http://www.dcscience.net/2015/12/11/placebo-effects-are-weak-regression-to-the-mean-is-the-main-reason-ineffective-treatments-appear-to-work/ http://www.dcscience.net/2015/12/11/placebo-effects-are-weak...
- henearkr 7y agoYour concern disappears if your method is to compare, on one hand, difference between treatment and nothing, and on the other hand, difference between placebo and nothing. To only compare treatment and placebo directly is of course a methodological error. PS: By the way, I'd advocate to scrutinize the biological bases of the placebo effect, in order to 1/ use it instead of trying to ignore it 2/ predict accurately its effect and thus lighten a lot the burden of test plans (ideally the tests could not need anymore to be double blind -- would save lots of time and money)
- heisenzombie 7y agomy understanding (see references in linked page) is that when one does what you said and compare placebo to “nothing”, almost the entire effect goes away.
- fooblat 7y agoHow would you propose a blinded study will work to compare between a Placebo treatment and no treatment?
- heisenzombie 7y agoOne way is to use waiting list controls.
- henearkr 7y agoTo compare placebo vs nothing (i.e. no placebo), you don't need the double-blind requirement. Because the group receiving nothing is not receiving the placebo, and thus does not (and should not) benefit from any illusion of having being treated.
- henearkr 7y agoIf we can verify that (I really hope so) it would be great news. Because the tests would be much simpler, and maybe costs and duration would become 1/10 of what they are today or even much less. The double-blind requirements are really a pain.