[extropy-chat] Re: Damien grants psi evidence

[spike]

Brett Paatsch


Spike writes:

> 
> Think of it this way: if there are 30,000 statistics and
> math PhDs in the world, it stands to reason that there is
> about a 63% chance that at least one of them would do a
> study which has a 4 sigma outcome by pure chance, which 
> would puzzle the hapless scholar to the point of misleading
> herself.

>A good explanation for how researchers can be almost cruelly
>mislead (by sheer bad luck), but it doesn't apply in this case.
>I'm not implying that Utts is mislead or misleading in other ways
>either btw. 

>Utts was summarising the results of a field of research and its
>findings over time not just conducting a single stand alone
>exeriment herself.  

>See Damien's earliest post in this thread for the links. Brett Paatsch


Ja I read that and get it.  What I meant by a 4 sigma event
is the general outcome of an arbitrary series of experiments.
Firing up the old equations (from sooo very many years ago) I
see that getting a 4 sigma outcome in an experiment is exactly
as weird as getting a 1 sigma outcome in 5.6 trials, or a
2 sigma outcome in 2.7 trials, or a three sigma outcome in
1.57 trials.  The case which might be most effective in
convincing a mathematically sophisticated researcher that
something is way broken in the universe is a series of 8.8
consecutive trials which had an outcome of half a sigma from
the statistically expected value.  But these are but more
examples of events that are the same weirdness as a single
4 sigma event.  

Since the number of trials is ordinarily an integer,  we can
express 4 sigma events thus:


1	trials with a 	4.00	sigma outcome is equi-weird to a 4 sigma
event
2	trials with a 	2.53	sigma outcome is equi-weird to a 4 sigma
event
3	trials with a 	1.86	sigma outcome is equi-weird to a 4 sigma
event
4	trials with a 	1.44	sigma outcome is equi-weird to a 4 sigma
event
5	trials with a 	1.15	sigma outcome is equi-weird to a 4 sigma
event
6	trials with a 	0.92	sigma outcome is equi-weird to a 4 sigma
event
8	trials with a 	0.60	sigma outcome is equi-weird to a 4 sigma
event
10	trials with a 	0.37	sigma outcome is equi-weird to a 4 sigma
event
12	trials with a 	0.20	sigma outcome is equi-weird to a 4 sigma
event


The formulas are in the enclosed spreadsheet, assuming ExI is
set up to transmit enclosures.  Is it?  

If Dr. Utts had any of the above happen to her, I am not
surprised if she started casting about for weird explanations.
If we assume 30,000 mathematically sophisticated researchers
worldwide (admittedly a wiiiild ass guess on my part) we would
expect a better than even chance one of them would have one
of the above outcomes.  

spike




-------------- next part --------------
A non-text attachment was scrubbed...
Name: spike'sReckons.xls
Type: application/vnd.ms-excel
Size: 18944 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20041217/b4dd1374/attachment.xls>