>Post-hoc comparisons have essentially no value for proving things; only for
>suggesting hypotheses to be test.
Of course. Do you see drool on my lip? :)
The difficulty with the most detailed natural psi experiments such as Lotto
is, as I mentioned, that the owners of the data are usually unwilling to
disgorge it more than once. But my work so far has set severe upper limits
on any population incidence of psi (which might, of course, turn out to be
zilch).
BTW, Dr Radin (whose book I cited several times) was given the boot from
University of Las Vegas, and his research centre closed down, on dubious
grounds last year. His side of the story can be sampled at
www.psiresearch.org/crlthes.htm
Further to my critical remarks about the tacky and misleading cover/title
of his book, this exchange:
ME:
>I was disturbed when you used the shorthand (e.g.) `odds of a hundred to
>one' = `probability of one in 100'. I've been snide at Randi for doing
>this, when he meant `odds of 99 to one'.
RADIN:
< Yes, I know. I had to make lots of strategic decisions about how to
present odds. I figured that while p = .01 is not exactly the same as odds
of 100 to 1, it was close enough for the expected audience, and for any
practical purposes in my exposition. But I'll add a footnote about this on
the next edition for the purists. BTW, the title and the jacket cover were
not my choice. I was quite angry about the cover in particular. You can
probably tell this was the marketing folks' idea of what I was writing about.>
Damien Broderick