# How should I modify my grading as a TA?

I am first year masters student. This past year, I did a first year lab with three different sections. I received the TA evaluations of the said lab and got not-so-great reviews to my surprise. I got around a 3.8-3.9 for two of them and one of them was about a 4.5 out of 5. And I also got 3 written comments. All three comments all complained about how I grade horribly, and I am disorganized. They compared the grade they got with the grades their friends from a different sections with different TAs for similar solutions (they claimed they worked with their friends). One claimed that his friend had marks in the 90s while he had marks in the 70s.

I feel like I am a fair grader. I grade based on the given grading scheme. I take away marks when students make mistakes in concepts or methods (not 2+2=5 errors). I judge how many marks to take off based on how bad the error is. I don't believe in giving students free marks or spoon feeding them since they don't pay attention to my comments (from experience) unless I take away marks. I have noticed assignments left around by other TAs in the lab rooms and seen that some TAs mark very leniently. Some mark lazily, where even if the student has the wrong method/answer, they get almost full marks, and some just act lenient because they say they feel for the students. During proofs, and multiple-step problems, I make sure to go through the steps to point out the mistakes in the concepts and make sure they're getting the full knowledge they deserve. However, I find myself feel horrible because its unfair if they see their friends ending up with much better grades than them for the same amount of work. Should I be more lenient at this stage? This is a first year course, so I can see why they would be weaker in the concepts, however, I have graded for a third year course and realized that students don't know the basics they should have learnt in the first year courses just because concepts are not reinforced to them from the very beginning.

Also, if students come to me about the grades, I either explain why they lost marks or if they have a justifiable explanation, I give them back some marks.

As for the lab part, I think I need to work on speaking with more confidence and such but in general, I think I know how to improve on it.

This year, I am teaching another first year course, and I find myself taking off points from people who use horrible notation, such as omitting limits when taking them, plugging in numbers into half the equation and leaving the variables intact in the rest, and such.

At this point I am just questioning on whether spending more of my time on checking their steps is worth it if I am just getting hated for it when there are other TAs getting by on being lazy about it and getting better reviews. I want the best for the students but they don't seem to appreciate my methods.

Sorry for the long post! Any advice is much appreciated!

Cheers!

• There will always be these problems with consistency when multiple people mark the same problems. Wherever possible, the problems should be divvied up between the markers (and scripts passed around) rather than divvying up the scripts. It sounds like you are marking conscientiously, and should keep doing what you are doing. However, this seems like a matter to be taken up with the course leader, rather than with strangers on an internet forum. Good luck! – Mark Grant Jun 5 '14 at 10:08
• Since when do freshmen have labs in mathematical degrees?? – Asaf Karagila Jun 5 '14 at 10:49
• This particular course was targeted towards engineering students. They had a lab component where they used MATLAB to solve problems in lab and had written assignments to do at home. – Merry Jun 5 '14 at 10:50
• You might want to indicate that the lab is related to mathematics (it's still off topic here, though). – Asaf Karagila Jun 5 '14 at 10:55
• When they got a different grade than their friend, did they bring it to you then? If they didn't, they really don't have a leg to stand on putting it in the review. You would need to review both the 70 paper you gave and the 90 paper the friend received and see where the disparity is. I wouldn't dismiss the claim outright, because how people feel matters to an extent, but that's the kind of thing they need to bring up right away if they really had a problem with it. – corsiKa Jun 5 '14 at 15:04

In the future prior to the course starting, all the TAs and the lead instructor should meet to discuss the grading scheme for HW. It's a good idea to set a very rough approximate median for HW. This way at the end HW averages aren't too different. I generally like to grade two or 3 HW problems in detail, and give a set number of points for "honest attempts" for the other problems. I set up a score sheet for my graders when I have multiple people grading that details exactly what a 4/4,3/4,2/4,1/4 and 0/4 HW problem looks like and include examples. When I do that the graders become much more consistent. Also I like a 3 or 4 point scale for HW problems because it usually means graders all choose the same number, once you start assigning 5 or more points you leave a lot up to interpretation.

4 - perfect, or arithmetic error (note many sign errors are conceptual)

3 - minor conceptual errors (you know what the student meant and it was correct) but their solution was unclear, or misused words in a way that made it unintentionally wrong

2 - major conceptual errors. They are clearly going about this problem completely incorrectly but they do show some evidence of going to lecture and reading the chapter

1 - states something like "this is where I am stuck on this problem ...", if this is done very well it can be worth a 2

0 - intentional BS or writing down an answer with no work

I'd then give them specific examples of past problem solutions that are archetypes of these 4 grades. It is completely reasonable to ask your instructor or have the TAs collectively come up with such a scheme. The 1 point for honest description of difficulties encourages academic honesty and gives you a lot of feedback. It saves the graders a lot of work from having to read BS answers.

Edit: I was a bit vague about what setting an "approximate" median, prior to the start of the course, means. It doesn't mean make all TAs hit an "exact" median score. What it means is have all TAs have an "approximate" median in mind while grading, it is only used to guide the TAs who are grading way too easily or harshly. You would tell your TAs something like this, "In the past, 95% percent of sections in Course X, when I have been the lead instructor, have had HW medians between 71 and 79 percent, please come talk to me if you think your HW score distribution is very different from this."

• "I grade based on the given grading scheme." It seems that there already exist a grading scheme. The problem seems to be some other TAs being "too nice". – Taladris Jun 6 '14 at 10:40
• what I said is there needs to be an in person meeting to discuss the grading scheme, where TAs agree upon a preset approximate median, ... not just the existence of a grading scheme, because as you have experience the grading schemes existence doesn't mean it will be followed. – WetlabStudent Jun 6 '14 at 20:55
• Asking for a specific median is wrong... But I agree on the rest – vonbrand Jun 7 '14 at 12:47
• @vonbrand although I see where you got that, I was pretty vague. I have clarified. – WetlabStudent Jun 8 '14 at 1:32
• Yes. I agree that if the median/average/... is out of whack, something strange is going on, and that should be investigated. But that is different from "desirable median." – vonbrand Jun 8 '14 at 2:51

Once I got quite good evaluation from a lab class but for my grading approach, which received exactly the same comments as you, e.g. that I like to cut points and things which I consider bad weren't even mentioned in other groups.

This is despite the fact that we had really detailed rules on how to grade each type of mistake. Not only each author of the problem prepared a list of common mistakes, but he would grade his group first, also skim over other groups solutions and then adjust the rules. Finally the rest would grade their groups, and yet there would be an update if someone found an error without an appropriate rule. It was almost as one person grading it all.

Still, there was huge inconsistency between graders. I was curious and went over other groups submissions and what I found was that the scores depended on how lazy or meticulous the grader was. It was easy to tell from the number and level of detail of comments the TAs left.

The actual problem was that we would start from some top possible score and cut points for errors. Lazy graders would find less mistakes while scrupulous TAs would find a lot. Hence, the inconsistency.

I think I did a good job in spite of comments about grading, and in my opinion so did you.

Instead, imagine that if you would give positive points for all nice things you found, the result would be quite different. Unfortunately, it's easier to find mistakes rather than correct pieces. One solution could be to include both kinds of points, but then there might be pressure to consider the solution good unless one finds a mistake, which would turn back into previous, flawed approach. Personally I have no idea how to fix this, without some person doing much more work than the others.

Concluding, don't worry too much about it and keep up the eagerness you have (it's great that you found this site and asked about your problem).

I hope this helps $\ddot\smile$

• + points for good things, or a holistic problem grade are far better than subtracting for errors! +1 – WetlabStudent Jun 5 '14 at 22:37
• Curiously, it is easier to find what is correct (just check if certain key steps are present, that an intermediate result is derived, and so on) than to detect errors (by definition, there are infinite ways of getting it wrong). I'll also give extra credit for solutions using different approaches, that we didn't anticipate. – vonbrand Jun 7 '14 at 12:55
• @vonbrand Well, if your student would write something correct, but irrelevant (e.g. true formula with some things multiplied by zero, etc.), would you give him additional points? – dtldarek Jun 7 '14 at 13:07
• @dtldarek no you wouldn't. You reward points only for correct and relevant steps. – WetlabStudent Jun 8 '14 at 1:30
• @dtldarek true, but its generally not too tricky. If the formula is 0=0, or if there are a ton of formulas written down, in a way fishing for points, I give zero. If the formula is "plausibly relevant" I do give something. This has really never been so much of a problem for me, but I can definitely see where it is tricky. – WetlabStudent Jun 8 '14 at 1:38

When I worked in the lab, we usually either tried to have the same person grade everyone for a given lab, or at least round-robin who grades which students on different assignments, so each student got an equal taste of the stickler/gimme guy.

The best strategy for consistent grading (and being able to explain later) is to solve in detail, and add points for things done right. Publish the solution and grading criteria. If there are several sections, coordinate with the other TAs to use consistent grading, for example by rotating the problem sets or having every one solve (and define grading) for one problem, or even grade one each. For programming/laboratory assignments, perhaps set up a general schema (and publish it beforehand!) for grading, with some overall criteria like orderliness, use of the tools, clear result.