We’ve talked about inherent difficulties in grading including unclear overall goals and conflicting in-assessment goals which together help explain why you, as a TA, are likely to often be asked to grade difficult-to-grade assignments. This page is devoted to practical steps you can still take.

1 Share your work

Grading involves making judgment calls and decisions. Don’t keep those to yourself: share them with other graders. If there’s not grader communication medium in your course, create one: a simple shared document where you can all write things like when they put (1,2,3) instead of (3,2,1) I gave them half credit.

This is hugely valuable to ensuring that students who made the same mistake get the same grade, that multiple graders make the same decisions, and that when regrade requests come in you’ll not forget your previous decisions and change the rules by mistake.

2 Ambiguity

Try to get a clear guidance on how to handle ambiguous answers. Usually this is easiest by example, but word the question to elicit a rule:

If I can’t tell if they wrote 3 or 8; should I give them the benefit of the doubt, or grade based on my best guess, or assume they were hedging their bets and grade as if they wrote the wrong thing?

That’s not always easy to do, though, and sometimes your question

This student’s answer has a lot of the right words and phrases, but isn’t grammatical; I can’t tell if they’re just bad at English or if they didn’t know. What should I do?

might result in the instructor trying to figure out the answer for you. If that happens look for what their approach is: are they going for all-or-nothing, or for most likely answer, or for averaging options, or for benefit of the doubt?

Whatever the instructor’s rule, and no matter what you think about it, follow it. You’re welcome to engage them on different rules for the next assessment after grading is over if you wish, but mid-grading it is almost always the case that changing the rules is more problematic than consistently using a sub-optimal set of rules.

3 Rubrics

A rubric is a scoring guide used to evaluate the quality of students’ constructed responses¹. There is extensive work in educational research on what make a good rubric, but for many of our classes the first blush answer is:

• Having a rubric is good.
• Rubrics with several small, simple yes/no items are better than rubrics with a few big bucket-of-points items.

Rubrics are important in consistency, accuracy, and speed. Good rubrics reduce the scope for implicit biases in grading, clarify ambiguities in grading creative answers, and help manage grader cognitive load.

However, some classes have no rubrics supplied, or rubrics whose items are vague enough to need their own rubrics. What should you do in those cases?

Make your own.

I’ve never yet known an instructor or TA to be upset when another TA offers a rubric (or rubric refinement) for an assessment. Occasionally they may offer corrections to the proposed rubric, but that’s great: you still end up with a rubric you can use.

Rubric design is a complicated process, but good first steps are

• For each point available to the problem, write down a yes-no item goal for that point to be given on.
• Review half a dozen submissions looking for mistakes your list does not include.
• Add rubric items for those no on the rubric, removing or merging others to keep it from getting to big.

Rubrics needn’t be presented as such; a note in the grader-shared document like I’m giving 1 point for proof structure, 1 for sound deductions, 1 for good inductive structure, 1 for using the symbols correctly, and 1 for the overall proof being correct will encode a rubric as a clarifying note.

4 Avoid Double-Jeopardy

One of the most common threats to accuracy in grading is double-jeopardy; that is, a single mistake that propagates to impact many different rubric items. The best rubrics avoid this by construction, but making those rubrics is a difficult skill to learn and even the best rubrics won’t cover every mistake a student might make.

If you find yourself docking points again and again and again for the same mistake, we recommend you ask the instructor if repeated docking is what they want or not. If not, explain to the student that while many of their answers are wrong, you’re giving some points back because they seem to all be based on the same error.

5 Teach the student

If you are following a rubric that will be released to the student, giving a score alone is enough. If you need to diverge from that you should use whatever commenting mechanism is available to teach the student what you are removing points for and why.

You are welcome to do more than that, but we generally discourage the following kinds of comments:

• Don’t repeat the rubric they’ll see.
• Don’t repeat the key they’ll see.
• Don’t say good job unless they actually did a good job.
• Don’t be mean (you should drop the class, did you even study?, etc).

Additionally, if you think they are missing something important and need to visit with a TA or instructor, ask the instructor how that message should be communicated to them; feedback as part of grading may not always be the right medium for that kind of message.

---

1. W James Popham, https://eric.ed.gov/?id=EJ552014↩︎