Grading information, December 19, 2003

Here are grade distributions for the class, based on the point total
out of 600.  Since I put the letter divisions in most cases near large
natural gaps, there was not much occasion to change grades on the
grounds indicated below.  For individuals with low grades and
individuals close to the top of a range, I did carefully compare the
relevant parts of the final with in-class exam scores, to see whether
something had changed.  That process changed one B+ to an A-, and one
C to a B-. (The average point total was 465/600.

542-567 A+ (5 people)
517-535 A  (5 people)
502     A- (1 person)

488-495 B+ (2 people)
445-473 B  (7 people)
425-438 B- (5 people)

386     C  (1 person)

307-340 D  (4 people)
-----------------------------

Here's some distribution information on the final, where the average
was 166/200, and approximate letter interpretations.

180-195 A  (8 people)
157-179 B  (14 people)
140-154 C  (4 people)
115-131 D  (2 people)
97      F  (1 person)
---------------------------------------------
If you would like to look over your final, I have them in my office
2-281.  I will hang onto them, however.

Note on the final: there was a mistake in the solution I
wrote to 5(b), where the density in the integral should have had

-(s-30000)^2/2 sigma^2

as the exponent of e, and the denominator should have been sqrt(2
pi). (The first of these is forgiveable: I typed -{s-30000}^2, but TeX
considers braces to be non-printed characters.  The missing 2 in the
second was just stupid of me.)

Have a good holiday!

David Vogan

**************************************************************************

Status of grades, November 28, 2003

A number of people have asked me about the status of their grades in
18.440.  The final grade will be calculated beginning with a number
between 0 and 600 (sum of three exams, a 200 point final, and the
problem sets scaled to 100).  I'll convert that number into a letter
grade by some process that looks reasonable to me; that is, I'll make
decisions like,"people with numerical scores in the range 520-600
appear to have been doing work that deserves an A."  I'll then make
adjustments near the borders as seems appropriate.  For example, if
someone had a 30 on the second exam but did very well on the
corresponding section of the final, I might raise the numerical score
by something like thirty points, effectively changing the second exam
score from 30 to 60.  Because of such adjustments, the numbers now are
not definitive.

I calculated for each student the sum of the three exams and the nine
problem sets so far scaled to 100; this is a number from 0 to 400, and
the main thing that's missing is the final.  Here are some of the
numbers, and how I might now interpret them as letter grades.

345-380 (9 people) A
305-325 (7 people) B
270-290 (8 people) B-
220-240 (3 people) C
190-205 (3 people) D

You'll see that the adjustment system attaches a lot of importance to
doing well on the final.  So good luck with that.