Last semester, I was the TA for a small course on the physics of waves aimed at biophysics majors. Since I was the only TA, I was completely in charge of grading homework and writing homework solutions.
So I tried out a strategy that worked pretty well. I announced at my first recitation section that due to the inevitable errors I would make while grading homework, I was adding 5% to each homework grade (except where that would take the score over 100%). This would hopefully even out any errors I made of the course of the semester.
However, if the students wanted to find all the grading errors I made and point them out to me, I would still be happy to add back in those missing points. Of course, if that were the case, it would be clear I wouldn’t need to give that student the extra 5%, since it was only there to compensate for errors, and the student planned on catching those errors themself.
By essentially bribing students with a small bonus, I managed to go the entire semester without playing picky-points with anyone. (Although I can’t say for certain that I would have had to play it without the policy, since this was my first course here.)
This semester I’m in a 200-student introductory course and the homework is graded automatically by computer. I do have to grade lab reports, though.
At the TA organizational meeting today, they asked whether we TAs prefer labs to be graded out of 30 points or 100 points. 30 points is the clear winner. The problem with a 100-point scale is that you’re so used to it, the number is instantaneously and unconsciously compared against your expectations, but a less-common scale throws that off, is if you are hearing the temperature in Celsius or the price in yen. (For someone with my US biases, of course.)
26/30 and 87/100 are the same grade, but when you see 26/30, you say, “oh, I lost four points.” When you see 87/100 you say, “a B, what the hell! Where’s that TA?”
That’s the theory, at least. It remains to be scene if this semester will go as smoothly as the last.