Patrick Elliott <kag### [at] gmailcom> wrote:
> I gave an example of one physicist going so far 
> off the rails it was flat nuts, because he "believed" in telepathy and 
> psychic powers, and decided to test if the incidental flashes of light, 
> produced by the body's own chemical reactions in cells, would provide a, 
> "means to explain how some people could detect other people's thoughts."

Crazy experiments are ok... as long as you follow good scientific
procedures to test them!

It's a common problem with woo "science" that the experiments they
perform are extremely faulty. They fail to perform proper double
blind testing, they fail to have controls, they often use sample
sizes that are statistically too small, and they often engage in
confirmation and publication bias.

Publication bias is something that very easily misleads, and is
easily missed by the public.

It works like this: Perform many trials (preferably with small sample
sizes to maximize variation). Even if there is no correlation between
your hypothesis and the (alleged) phenomenon you are testing, just by
statistical probability a few of the trials will show a positive
correlation between them (while likewise a few will show a negative
correlation, and the rest will show no statistically significant
correlation). Discard all the trials except the ones that showed a
positive correlation, and publish those.

For example homeopaths do this all the time.

-- 
                                                          - Warp

Post a reply to this message