8.01 Final Renormalization of Recitation Quizzes, Fall 1998
December 21, 1998
As was mentioned in the course information sheet, in 8.01 we try
to do the best we can to compensate for the fact that recitation
instructors have different styles in terms of the difficulty of
their quizzes and the manner in which they are graded. We
therefore adjust the Recitatation Quiz grades by a process that
we call "renormalization." Each instructor is calibrated by
comparing the Recitation Quiz grades of his/her students with
their Review Quiz and Final Examination grades (which are
team-graded), and the Recitation Quiz grades are corrected
accordingly. The correction formula guarantees that the
average of the Recitation Quiz grades for all 8.01 students will
either not be changed, or will perhaps go up a little.
The semifinal renormalization prescription, based on the five
Recitation Quizzes and three Review Quizzes, was posted earlier.
This document describes how to calculate your final renormalized
Recitation Quiz average, which also takes into account the
results of the Final Examination. The renormalized Recitation
Quiz grades are readjusted as a result of the Final Exam only for
those students for whom this adjustment results in an increase in
the grade.
This document will include the formulas for computing your own
renormalized Recitation Quiz grade, and also a set of grade cuts
and a histogram so that you can see where you stand. The
Recitation Quiz grade counts for 20% of your final grade in
the course.
HOW TO COMPUTE YOUR GRADE:
In our effort to be as fair as possible, we are using a fairly
complicated method of computing the Recitation Quiz average.
- STEP 1: Computation of Raw Grade
- This computation is the same as it was for the semifinal
renormalized Recitation Quiz average, but is repeated here for
convenience. The basic idea is that each student should be
allowed to drop his or her lowest quiz, and then the remaining
quizzes are averaged. The implementation of this principle is
complicated by the fact that in the Tues-Thurs sections one of
the quizzes is to count only half, and also by the fact that
some students had one or more excused absences.
- Monday-Wednesday sections:
- If
Q1 is the lowest quiz score and Q2, Q3, Q4, and Q5 are the four
others, then the desired average is:
Q2 + Q3 + Q4 + Q5
<Q> = -------------------
4
If the grades were not based on 100 as the maximum
possible grade, then you should scale them so that they
are based on 100.
- Tuesday-Thursday sections:
- For these sections the 3rd of the 5 quizzes should
count only half as much as the others. Let Q_1/2 denote
the score on the quiz that counts half, and let Q1 denote
the lowest of the others. Then let Q2, Q3, and Q4
denote the remaining scores. I am assuming that the
maximum possible score is 100 for all 5 quizzes. If Q_1/2
is the lowest score, then the average is
0.5*Q1 + Q2 + Q3 + Q4
<Q> = -----------------------
3.5
If Q_1 is the lowest score, then the average is computed by
0.5*Q_1/2 + Q2 + Q3 + Q4
<Q> = --------------------------
3.5
- All students with excused absenses:
- If you have any excused absences on the Recitation
Quizzes, then
click here
to find out how to compute your raw grade.
- STEP 2: Computation of Renormalized Grade
- As explained with the preliminary renormalizations, we
experimented with three different methods of
renormalization, but found that none of them seemed to be
fair in all cases. We have therefore decided to use all
three, and for each student the maximum of the three
calculations will be used for the renormalized grade. For
each instructor, four parameters were determined by
comparing the Recitation Quiz grades, the Review Quiz
grades, and the Final Examination grades of the instructor's
students. A table of these parameters is as follows:
TABLE OF RENORMALIZATION PARAMETERS:
| SEMI-FINAL | FINAL |
| Sections | Instructor | R | Q | M | B | R | Q | M | B |
| 1 | Busza | 1.259 | 0.610 | 0.848 | 25.3 | 1.305 | 0.564 | 0.744 | 34.3 |
| 2 & 3 | Burgess | 0.956 | 1.260 | 1.356 | -32.7 | 0.969 | 1.277 | 1.469 | -41.6 |
| 4 & 5 | Hauer | 1.072 | 0.851 | 1.097 | -1.8 | 1.044 | 0.911 | 0.985 | 4.5 |
| 6, 7, & 8 | Javan | 0.808 | 2.164 | 0.771 | 3.1 | 0.801 | 2.345 | 0.975 | -15.4 |
| 9, 10, & 11 | Joss | 1.038 | 0.900 | 0.959 | 5.7 | 1.046 | 0.878 | 0.898 | 10.7 |
| 12 | Koster | 1.015 | 1.024 | 0.575 | 34.6 | 1.007 | 1.047 | 0.549 | 36.1 |
| 12 & 14 | Falus | 0.936 | 1.192 | 1.431 | -38.6 | 0.938 | 1.154 | 1.419 | -37.1 |
| 15, 16, & 17 | Burke | 0.901 | 1.358 | 1.113 | -17.1 | 0.893 | 1.359 | 1.323 | -34.6 |
| 18 & 19 | Su | 1.052 | 0.867 | 1.002 | 3.5 | 1.065 | 0.837 | 1.206 | -9.9 |
| 20 | Wiese | 0.991 | 1.099 | 1.194 | -16.0 | 0.985 | 1.127 | 0.990 | -0.3 |
| 21 & 22 | Millar | 1.061 | 0.863 | 1.294 | -16.8 | 1.059 | 0.858 | 1.142 | -6.0 |
| 23 & 24 | Graham | 0.957 | 1.025 | 1.183 | -16.3 | 0.984 | 0.977 | 1.028 | -3.2 |
| 25 & 26 | Katz | 1.081 | 0.822 | 1.094 | -0.9 | 1.085 | 0.821 | 1.388 | -21.9 |
- Use the parameters in the above table to compute your
renormalized Recitation Quiz grades by each of the
following three methods, using both the semifinal and
final values of the parameters. Then take the maximum of
the six calculations:
- Method 1:
- Renormalized Grade = R * (Raw Grade)
- Method 2:
- Renormalized Grade = 100 - Q*(100 - Raw Grade)
- Method 3:
- Renormalized Grade = M * (Raw Grade) + B
WHERE DO YOU STAND?
The average raw grade for all 8.01 students on the Recitation
Quizzes was 74.9, which I think is very good. Each of the
renormalization methods used is designed to preserve the average
for all 8.01 students, but the ``best of six'' algorithm caused
the average renormalized Recitation Quiz grade to rise to 77.2.
The passing grade for the renormalized Recitation Quiz average
has been set at 62.0, and other numerical grades can be
translated into letter grades according to the following chart:
| NUMERICAL GRADE | LETTER GRADE |
| 87.0 - 100 | A |
| 75.0 - 86.9 | B |
| 62.0 - 74.9 | C |
| 45.0 - 61.9 | D |
| 0.0 - 44.9 | F |
A histogram of the grades is shown below:
IF YOU ARE INTERESTED:
Back to the 8.01
Home Page.
Last modified: Monday, December 21, 1998 7:46 am