home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.skeptic:19109 rec.org.mensa:8025
- Path: sparky!uunet!ogicse!uwm.edu!linac!att!cbnewse!gmark
- From: gmark@cbnewse.cb.att.com (gilbert.m.stewart)
- Newsgroups: sci.skeptic,rec.org.mensa
- Subject: Re: Astrology: Scientific Research
- Message-ID: <1992Nov6.154733.16809@cbnewse.cb.att.com>
- Date: 6 Nov 92 15:47:33 GMT
- Article-I.D.: cbnewse.1992Nov6.154733.16809
- References: <dhalliwe.720824224@muskwa.ucs.ualberta.ca> <1992Nov5.133607.16704@msuinfo.cl.msu.edu> <dhalliwe.720995882@muskwa.ucs.ualberta.ca>
- Organization: AT&T
- Lines: 42
-
- In article <dhalliwe.720995882@muskwa.ucs.ualberta.ca> userDHAL@mts.ucs.UAlberta.CA (David Halliwell) writes:
- >banghartr@clvax1.cl.msu.edu (Rick Banghart, IPTV) writes:
- >
- >>I still maintain that the strong correlation (if it exists) points
- >>to _something_ happening that requires further examination. Even
- >>if the mechanism is (for example) the tendency of people to
- >>conform to astrological "information" about themselves.
- >
- > ..but that was the thrust of [someone long lost in the chain]'s
- >original comment that *without* finding some logical reason why
- >one item affects the other, the statistical correlation is only
- >weak evidence. If, after looking for the link, you can't find an
- >explanation, then you would normally reject the correlation as spurious.
- >
-
- Careful, folks. Let's not get sloppy from the get-go. Remember that
- "weak" is a relative term, and the statement that "statistical correlation
- is only weak evidence," is, at best, a generalization.
-
- Secondly, the fact that "after looking for a link, you can't find
- an explanation," relies heavily on your ability to find such a link.
- If you happen to be a very good theoretician, it may be Probable
- that an explanation of a possible cause does not exist, but it does
- not prove it.
-
- Thirdly, theories are never "proven beyond a shadow of a doubt". All
- we have is levels of apparent probability, period. Theories stand
- until they fall, and every theory probably has a long list of previous
- theories that seemed highly probable until a more probable one came
- along. And all we EVER have is data, and the only measure we EVER have
- is the probability that the predicted result will indeed happen.
-
- And the basis of all of that is statistical correlation. And since
- you rarely have ALL possible data, this is still subject to inaccuracy.
-
- When we use relative terms and generalizations, we often start with sloppy
- premises and reasoning that is often difficult to clear up with subsequent
- posts.
-
- Just a note to keep us all honest.
-
- GMS
-