Told ya so, Alex Stroup

[Ghoulish Delight]

Originally Posted by tracilicious

How an waer be armful (and wy is my keyboard malfunioning?????)

It (alone) can't beyond placebo, no more than water (alone) can be beneficial in the way homeopathy claims. Which is exactly the point Alex was making. It is one of standard distribution. Out of 400 small case studies, it is expected that through pure random chance, about 10 of those case studies will show that the people receiving the homeopathic remedy will fair worse than people taking a placebo. And in about 10 of those case studies, it's expected that people receiving the homeopathic remedy will fair better.

To put it another way, if you take some large # of people (let's say 40,000 just to put a # on it) and start randomly grouping them together to form 800 small groups 50 people (I say 800 instead of 400 since each group is roughly split in half between placebo and actual test substance), then purely by random chance, a small percentage of those groups of 50 people are going to end up being composed disproportionately of people who happen to respond better or worse to a placebo than average. So those groups of people will produce results at the far end of the spectrum. But that doesn't prove anything, it is only a single datapoint.

So you can't look at a single study and say, "aha! Here's a group of 100 people in which the homeopathic remedy performed better than placebo" and callt he homeopatic remedy confirmed. Because if you do that study 399 more times, you'll find that yes, about 9 more times you'll see a similar result, about 10 times you'll see the exact opposite result (the placebo significantly out performs the homeopathy) and the other 380 times the difference between the two would show no statistically significant benefit to either (with roughly half showing negligible tendencies toward better results within error tolerances from placebo, half toward homeopathy).