How to speed up grading and clarify why students lose points (without spending days grading)
When a group of students takes a test, they will predictably make categorical mistakes. Even in a learning situation such as Physics 131 at the University of Maryland, where the questions are challenging, often unique (compared to, say, standard 20th century physics instruction), or downright open-ended, the same mistakes are often replicated by multiple students. This presents an opportunity to speed up the test grading process by a large factor; you may find it useful even during homework grading.
For instance, if a student trips over units, the grader can just write "U" on the side, deduct whatever points reflect the severity of the mistake and move on (see "But what about point values?" below). The grader does not need to waste time writing "your units are off here" or "units wrong" or even "units." It is UP TO THE STUDENT to interpret the code and find the error (though obviously the graders and instructors can help in office hours if the student is still stumped). The student can then look on this table (if the student has never gotten a "U" before such as in our example) to translate the code and figure out how to improve next time. Or, to challenge the grader later in the case that the grader has made a mistake. This prevents, "Why did I get two points off?" type questions and at least reduces the volume of confusion down to the occasional, "But I didn't make the mistake you claimed I made!" which, of course, due to human error, will happen once in a while and is a great physics conversation starter.
In order for this to work, a relatively universal code should be used and accessible to the graders and the students alike. One can also generate unique codes for specific assignments or test problems. A common error might get its own category (but do not use any of the reserved codes below, and avoid letters that look like numbers); make sure you record whatever codes you come up with (and what kind of point deductions you think they're worth). But, beware code proliferation (which starts to waste everyone's time) and avoid "symbolosis": using the same letter for two totally different kinds of mistakes. Sometimes it is worth a grader using multiple codes when more than one kind of mistake has been made. However, once the quality of the student's effort has suffered with three or more mistakes, sometimes an "X" makes more sense than specifying every kind of mistake.
Here, then, is a list of "Universal Grading Codes" which can be used right away. We expect the list may morph and grow. But, please also see "What about point values?" below!
Universal Grading Codes |
The most common codes |
A |
All correct (you should still look over it carefully) |
B |
Blank - always worth 0 points |
X |
Off-the-wall or random answer, grasping at straws, etc.; sometimes worth points for trying
|
Other common codes |
Cf |
Error carried forward from previous part (but otherwise correct)* |
E |
Explanation weak - missing key elements |
Ex
|
Missing requested example(s) - typically for a definition question |
L |
diagram or graph poorly labeled (e.g., missing an axis title and units) |
L+ |
diagram or graph very poorly labeled (multiple forces mislabeled or missing labels) |
M |
minor arithmetic error |
M+ |
major arithmetic error |
N# |
Newton's laws (e.g., N1 is an error regarding the first law) |
P |
Mostly right, but incorrect extraneous piece added** |
Q |
Not answering the entire question (often because you didn't read the entire question) |
Sf |
Significant figures too few or too many |
U |
something wrong with the units |
*People should not be penalized twice for Cf; presumably they lost points in the previous section. Also, Cf should ONLY be used if the subsequent answer is otherwise correct.
**If the extraneous part(s) contradict(s) the correct part, an "X" may make more sense than a P!
But what about point values?
"A" is all the points, of course, and "B" should be 0. The category "X" is often a pity point issue: if the problem is worth 5, and they have written a passionate (but completely wrong) explanation, a point or two avoids a bimodal effect on the distribution. For a 10 point problem, maybe 2 or 3 points. The idea is that even a wrong answer is more valuable than a blank one.
Other categories are at the discretion of the instructor or, sometimes, the grader. On tests, there is always the danger of "burn out" grading wherein after two hours of seeing wrong answers, the grader might start taking off even more points than earlier. To avoid this, the ideal method of grading is to categorize a large portion of the class (at least 30 tests or so in a class of 100+) first into piles for each category. Then decide on the point values for the categorical mistakes, jot them down to be consistent with everyone and continue. This is very useful also if for some reason a few tests are graded much later on and the grader needs to remember what a "E" is worth on problem 2b to match whatever was done for "E" on 2b on other tests, especially if, for some reason, it isn't the same grader!
Comments (0)