[Fred Nickols (971219.1645 EST)]
Bruce Nevin (971218.1745)
Your objection to Mark's coin flip comparison trips over this ambiguity, I
think. The .5 correlation is over a population. The coin flip is one
individual at a time. "Sam's GPA will correlate with his IQ, yes or no."
(Flip) "Gail's GPA will correlate with her IQ, yes or no." (Flip). Half the
time you're right. Tough luck on the smart kid frozen out of a scholarship
because the application of statistical measures to individuals tipped the
balance on the record. But helpful to admissions officials and aid
officials who need a winnowing sieve, and wouldn't be caught dead flipping
a coin.
Bruce:
Fred Nickols here. As you might know, I work at ETS (Educational Testing
Service), although I am not a psychologist, a statistician, or a
psychometrician.
Can you say some more about how statistical measures are misused in the
admissions process? I'm especially interested in the remark about the smart
kid frozen out of a scholarship "because the application of statistical
measures to individuals tipped the balance on the record." Where and how
does this happen? Does Phil Runkel's book speak to this issue, in
particular, to admissions?
Regards,
Fred Nickols
nickols@worldnet.att.net