I know I might get electronically lynched for saying this, but I tend to not consider the end of the semester as some kind of hurried rush.
I know that I’m not really going to get any research done, so that’s one less thing to worry about. I have very little life outside work, so that’s mostly easy to put on hold. Except for Senior Thesis projects, I’m usually not dealing with large projects at the end of the semester so there isn’t a big grading spike, except for final exams. However, without classes being in session during finals week, grading my typical 60-80 final exams doesn’t take any longer than I would spend on normal teaching stuff. I don’t really like grading, but at least with exams I control the format of the assignment and can do little things that help, like a score page, answer boxes for the students to use, etc. The worst part about grading final exams is that they are generally 1.5-2 times as long as regular mid-terms. (This of course varies from institution to institution and from professor to professor.)
Although the physical sciences deal with “right answers”, particularly at the undergraduate level, that doesn’t mean there isn’t subjectivity in the grading. That’s what I like to call “the art of the partial credit”. More for historical reasons than any logical reasons, my school tends to prefer “high” grading scales; i.e, even all the way up to 90-80-70-60. That really doesn’t lend itself well to having the students do anything substantive on exams, but we essentially don’t do multiple choice questions in my department, either. In similar classes to which I teach, I had undergraduate classes where the grading scale was more like 80-60-40-20, but that wouldn’t fly here. If exams are to be non-trivial with a “high” scale, one has to be fairly generous with the partial credit. It’s a little more work because one has to keep notes to make sure that the same mistake is given the same score on different exams. However, it always amuses me that students tend to mostly make the same small mistakes as other students. If I give the same type of problem on final exams time after time (final exams are not returned to students at my school), I can keep the same rubrics year after year.
Despite the subjectivity already noted, I do tend towards a fairly rigid final grading metric. I set things up so one point on a homework set is the same as one point on an exam. I.e., I just add up all the points earned on assignments, and let the number of points per problem on homework and exams provide any weighting. Roughly speaking, I usually make the total number of points on homework the same as one mid-term exam. Thus, a homework problem might only be worth a few points while a mid-term problem might be worth 30 or more. I much prefer this over some system where I assign grades to assignment and then have to weight them in the end. I feel that the latter is not very transparent to the students, while my method allows students to even do things like figure out how many points they need to get on the final to get a certain grade.
One potential downside that some people would point out is that if a student does terrible in the first part of class, they will have a hard time bringing their grade up. Some people (especially in non-science fields) prefer to give some weight to improvement over the course of the semester. That probably does make sense with more subjective and non-specific material. The problem with math and science classes is that while the classes do follow a progression of topics, you have to demonstrate proficiency in all of them along the way. My final exams are semi-comprehensive so that I do pull back material from the mid-terms which can give students a second chance on the earlier stuff. (I even do this to the point where I very nearly take problems verbatim from previous exams and maybe fiddle with the numbers a bit.)
Still, doing poorly on one of the mid-terms in the first place means that even after turning in homework and having chances to master the material, the student didn’t. Our faculty are required to post and maintain quite a large quantity of office hours and we have an “open door” culture, so there are plenty of opportunities to get help. Most of us also do review sessions before an exam which often give bold hints as to what will be on the exam. Of course everyone has bad days, but my campus is also fairly touchy-feely about giving students a break if they are sick or dealing with some difficult situation. Don’t tell anyone, but I have given make-up exams for people who have overslept an exam. I find that students who are trying the game the system do poorly anyway, and in this particular scenario they may very well be sick anyway. Plus, I have had a couple students who had genuine sleep apnea diagnoses.
There are times when I would like to give someone a boost because they showed improvement after a poor mid-term. But, it is ridiculous (to me) to boost one student’s letter grade over another student’s if their total scores are nearly identical. If the situation allows, I have actually lowered the grading scale for everybody so as to bump one particular person up to the next letter grade, bringing along another student or two. This is sort of a mild form of curving, which I occasionally have to do for upper-division classes anyway. Ideally, what I like is when there is a clear break in the grade distribution near a grade boundary, but that doesn’t always happen.
The fact that a student’s final score usually dictates a specific grade (curving and such usually only affects a small fraction of students), one can have a bit of fun with the grades during the process of grading. Since my final exams are semi-comprehensive, I usually quickly run through the old problems because the students usually do well on those, and enter them into my grade spreadsheet. Thus, I can see how well a student needs to do on the new material to reach the appropriate grade level and see if they have a chance or not. Of course with a well-defined rubric they get what they get, and you have to give it to them dispassionately.