home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!hri.com!noc.near.net!ns.draper.com!newsgate
- From: skh4161@mvs.draper.com (Kjeld Hvatum)
- Newsgroups: rec.org.mensa
- Subject: Re: The Mega Test
- Message-ID: <01GSL94RNTAQADCJUZ@ccfvx3.draper.com>
- Date: 21 Dec 92 15:11:00 GMT
- Sender: mmdf@ns.draper.com (MMDF Master)
- Organization: Draper Laboratory
- Lines: 39
-
-
- >From: Chris Long <clong@romulus.rutgers.edu>
- >Message-ID: <Dec.16.17.29.17.1992.3447@romulus.rutgers.edu>
- >
- >> That is, they could all be solved quickly if you could write fairly
- >> simple programs for an infinitely fast computer.
- >
- >So could #25 and #26, assuming you define the correct answer in a way
- --------
- But that's the problem.^^^^^^^^
-
- >> Without going into a lot of psychometric mumbo-jumbo, the theory is that
- >> the two types of questions measure slightly different things.
- >
- >But if you're only trying to measure the underlying "g", what difference
- >would it make?
-
- Why do you think the SAT uses antonyms as well as analogies to
- measure vocabulary? Why does the Wechsler have 11 different
- subtests? Shouldn't just lots of questions of some desirable
- type be best. Well, in fact, vocabulary is one of the best
- single measures, and some IQ tests are entirely vocabulary tests.
-
- But, as you might know, when you have a noisy measure, you can
- do better by combining two noisy measurements which aren't
- as closely correlated as the components of a single measure.
- This is, for example, why professional investment
- managers add a little gold to a conservative portfolio of
- stocks, bonds, etc., in order to make it even more stable,
- as contradictory as that may sound. Regression analysis,
- and more advanced forms of it, like Kalman filtering, implement
- this observation to the hilt.
-
- >As far as I can see, the only role very ambiguous questions
- >serve is to punish those who think differently.
-
- I think the reasons you sent me for picking the other choices
- miss a number of observations about each group of choices.
- I'll respond to you further on this via e-mail.
-