[comp.edu] Grading on a curve

rjung@sal1.usc.edu (Robert Jung) (03/17/88)

(Okay, this probably doesn't belong in rec.humor.d anymore, but I wanted to throw
my $0.02 in...)

  I always thought "grading on a curve" meant the following:

1. Design test.
2. Grade tests (numeric scores).
3. Get highest numeric score (call it x)
4. Work out grades as follows:
	1.00x - 0.9x		= A
	0.89x - 0.8x		= B
	0.79x - 0.7x		= C
	0.69x - 0.6x		= D
	anything worse		= Fail

  This way, if the test is exceptionally hard, the student's aren't
penalized. The numeric breakdown separates the chaff from the wheat, and
your smart students *are* rewarded...

  DISCLAIMER: I am in no way, shape, or form an educator of some kind. This is
just my concept of "fair grading" (HA!)


						--R.J.
						B-)

P.S. Okay, a joke for all the rec.humorists out there:
	How do you determine the sex of a chromosome?
	You pull down its genes.

<=====================================><=====================================>
Disclaimer: These ideas are all mine! Mineminemineminemineminemineminemine!

Send e-junk-mail through Bitnet to rjung@castor.usc.edu

mchin@homxc.UUCP (M.CHIN) (03/17/88)

Personally, I always used the following definition of a curve.

	1)	Grade all the tests.
	2) 	Find the mean.
	3)	Find the standard deviation.
	4)	For your standard 'C' centered course, a C is the mean
		B is one standard deviation up.
		A is everything above two standard deviations.
		D is one stand deviation down.
		F is everything below two standard deviations.

Although this didn't always hold true, it gave me a rough estimate of what
my grade was going to be.  I liked this definition because it allowed me to
figure out the curve for non-C centered courses.  A B/C centered courses
placed the mean at the B/C level.  Everything above within one standard
deviation was a B.  Made A centered courses real nice.

Michael Chin
ihnp4!homxc!mchin

cmr@m2-net.UUCP (Chuck Rader) (03/20/88)

[Note that the only attempt at humor is in the summary line and follow-ups
are directed to comp.edu]

When I was an undergrad here at U of D in the late seventies, several 
Management Science instructors ran their students' test score through
a computer program to calculate the mean, variance and standard deviation.
The cutoffs for each grade were determined by the standard deviation, with
C centered on the mean.