[ExI] heaves a long broken psi

[Damien Broderick]

On 1/24/2010 1:28 PM, BillK wrote:

> The second 'random' was meant to refer to the fact that
> sometimes you get results above chance level and sometimes you get
> results below chance level. And, assuming all the strict controls are
> in place and actually work properly, that has nothing to do with psi
> but is due to the unpredictability of the universe we live in.

This is untrue, though. The whole point of using statistics in science 
is that with sufficient data points, random correlations can be 
disambiguated from the non-random. Do you really suppose that 
parapsychologists with PhDs and professorships in statistics or physics 
have never noticed the elementary facts you're pointing at? One of the 
most rudimentary controls is to match a string of calls not only against 
the actual target array but against strings of equal length with 
different arbitrary "targets"--and waddaya know, those control 
comparisons turn out to agree closely with chance expectation.

Psi studies don't just look for "results above or below chance level" in 
some vapid "oh look, it's not exactly 0.20 right, it must be magic!" The 
word *significant* has a technical meaning (as you surely know). If a 
subset of trials deviated significantly from m.c.e. because the mind's 
propensity to form patterns happens by chance to match the random target 
string for a few minutes, it will soon stop doing so (and hence balance 
out back within one sigma, two-tailed, or so) if there is nothing 
forcing the correspondence. That's true whether you're looking for psi 
or the effectivity of aspirin in heart attack prevention (which has a 
comparable low effect size, as demonstrated by Prof. Jessica Utts).

Damien Broderick