Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site bonnie.UUCP
Path: utzoo!watmath!clyde!bonnie!wjh
From: wjh@bonnie.UUCP (Bill Hery)
Newsgroups: net.jokes
Subject: Re: Northwestern?/caltech
Message-ID: <598@bonnie.UUCP>
Date: Fri, 25-Oct-85 11:47:03 EDT
Article-I.D.: bonnie.598
Posted: Fri Oct 25 11:47:03 1985
Date-Received: Sat, 26-Oct-85 04:25:21 EDT
References: <1730@watdcsu.UUCP> <249@laidbak.UUCP> <990@oddjob.UUCP> <112@cher.UUCP> <3032@sdcc3.UUCP> <760@adobe.UUCP>
Organization: AT&T Bell Laboratories, Whippany NJ
Lines: 22

> Wrong.  The scores on the SAT's have nothing to do with how many you "get
> wrong".  They are standard deviations from the statistical mean.  This is to
> say that the median score of all the folks who took that particular test is
> declared to be 500........
>

Sorry but you're also wrong. The score is based on a statistical distribution,
but it not just the distribution of 'those folks who took that particular test'
as you stated.  The test scores are SUPPOSED to be normalized to remove
the effect of the particular group that took that test, and to make the
scores comparable accross different tests over several years.  For several
years in the late 60's through the '70s the scores were consistantly declining
every year; I believe the trend has reversed recently.  I'm not sure how they
normalize te scores--is there anyone out there from ETS to explain it?

Also, 'that particular test' is actually taken by only a small portion of those
taking the test at one time.  The tests are made up from a large set of questions,
with many somewhat different tests emerging--the person next to you might
have some of the same questions as you do, but not necessarily in the same
place on the exam.  At one time, some questions were used year after year, but
I think they stopped that now because of people publishing questions from
old tests.