Xref: utzoo comp.edu:1023 rec.humor.d:622
Path: utzoo!mnetor!uunet!husc6!bloom-beacon!gatech!udel!princeton!mccc!pjh
From: pjh@mccc.UUCP (Peter J. Holsberg)
Newsgroups: comp.edu,rec.humor.d
Subject: Re: cruelty to undergrads
Message-ID: <486@mccc.UUCP>
Date: 17 Mar 88 17:28:01 GMT
References: <18618@topaz.rutgers.edu> <401@lscvax.UUCP> <3530@medusa.cs.purdue.edu> <3766@bloom-beacon.MIT.EDU> <275@ritcv.UUCP>
Reply-To: pjh@mccc.UUCP (Peter J. Holsberg)
Organization: The College on the Other Side of US Route 1
Lines: 23

|In article <3766@bloom-beacon.MIT.EDU> jfc@athena.mit.edu (John F Carr) writes:
|>In article <3530@medusa.cs.purdue.edu> tlh@cs.purdue.EDU (Thomas L. Hausmann) writes:
|>
|>>But seriously, when someone gives a satisfactory definition of just
|>>what they mean by a curve, then we can *really* know what it is to
|>>grade on a curve.
|>

Here's my $0.02 worth.  When a test has been graded, I plot the results
on a piece of graph paper.  For an "average" class & test, the median
group of grades is a C.  Then I look for clusters of grades that can be
designated A, A/B, B, B/C, C, C/D, D, and F.  However if the median is
too high, I go with straight percentages.  Sure would be nice if I knew
how to write a test with some statistical validity.  I'm awed by folks
who have so much trust in their ability to do so, that they can say that
89 is a B and 90 is an A.  I'm sure that the noise level in my tests is
around 3-6 points!

-- 
Peter Holsberg                  UUCP: {rutgers!}princeton!mccc!pjh
Technology Division             CompuServe: 70240,334
Mercer College                  GEnie: PJHOLSBERG
Trenton, NJ 08690               Voice: 1-609-586-4800