"Hermes" <hermes@[EMAIL PROTECTED]
> wrote in message
> Apollia wrote:
>> On Apr 28, 5:47 pm, Apollia <xerxes...@[EMAIL PROTECTED]
> wrote:
>> > On Apr 28, 1:33 am, Hermes <her...@[EMAIL PROTECTED]
> wrote:
[....]
[Graph links]
>>>>> Statistically indistinguishable from random noise.
>>>>> (Less data than for the basketballers, so larger
>>>>> relative fluctuations are more likely to occur).
>>>
>>>>> For the record: I do not agree with Ray Murphy. :)
>>>
>>>> OK, for the record, I hope nothing I said implies that you do.
>>>> :-)
>>>> And, nothing I said was meant to imply anything like that.
>> >
>>>> As for what I agree with - well, I don't know, I don't understand
>>>> this stuff well enough yet to make up my mind, so, I'm going to
>>>> keep
>>>> trying to remain basically agnostic and neutral... :-)
>>>
>>> http://www.exactphilosophy.net/sample/sample.html
>>>
>>> A Java Applet that randomly picks one of 12 numbers
>>> 2911 times and plots the count for two arbitrary
>>> numbers in function of added count for both numbers
>>> together, i.e. same type of graph as Ray presented,
>>> but with random data instead of collected birth data.
>>>
>>> Click the graph to generate a new sample. Should not
>>> take long until you see graphs that resemble the ones
>>> presented by Ray very much.
>>>
>>> Also contains a link to the Java source code.
>>
>> Thank you! That's a really neat program. Makes me want to learn
>> Java...
>>
>> > This is not at all a border case where one suspects
>> > that maybe the presented data might be significant.
>> > This is a clear case of noise, not of any provable
>> > structure. Trust me, I have a Ph.D. in physics. :)
>> >
>> > Would be nice if it was different, but is not so.
[...]
>> > )o+
>> I definitely do trust you, but, I don't really want to accept
>> anything just on blind faith in someone's authority/credentials
>> (otherwise I'd have to dismiss astrology on the basis of zillions
>> of other scientists saying it's nonsense). :-)
>>
>> I have to reach the point of being able to understand this stuff
>> myself - then, I'll be able to make up my mind.
>>
>> That said, I highly esteem your opinion, I hope someday I'll be
>> able to understand this stuff as well as you... :-)
>
> But it is so obvious! :)
> Don't the graphs look very similar?
> Don't they occur at every so-and-so click?
> Isn't the source code very brief and simple?
[...]
> )o+
What you have done is created the illusion that because the random
graphs look similar to the ones I've presented, then mine are nothing
more than random graphs as well, and so will all others in the future
be too if they look anything like them, however your assertion is
simply not true and it can be easily confirmed by anyone who cares
to look.
The bottom line is that a close inspection will reveal that your
random graphs are not the same as mine and they are not the least bit
interesting from a statistical viewpoint because they don't get over
the magical .05 line any more often than they should. Of course that's
the way it's supposed to be, because the world of statistics would be
in complete disarray if random data WAS producing statistically
significant results more often that was expected.
The essential point is that my graphs are statistically significant
all the time and the random graphs are only slightly significant only
as often as they ought to be.
Ray
|
