Letting students grade their own homework

Question

I am going to teach a course in algorithms to 2nd-year undergraduates. The grade in this course is composed of 90% exam and 10% homework. The problem is, this year we do not have enough personnel to grade the homework.

Theoretically, it may be possible to change the grade struture and make it 100% based on final exam, but, this will reduce the incentive of students to study and practice during the semester, which I think will harm their learning experience.

So, I thought of the following scheme: the students will have to submit each assignment before the next TA session. The TA will explain the solutions during the session, and then each student will grade his/her own homework.

Sure, some students might be tempted to give themselves 100 on all assignments, but I believe this will not be very harmful since:

• All assignments together compose only 10% of the grade, so every single assignment is less than 1%. Even if many students get 10% for free, the final exam is still sufficient to differentiate the good from the bad students.
• The group consists of excellent students (the top 20%, based on their 1st year mean grade). I believe most of them want to succeed in the exam, so they will take this as an opportunity to track their own progress during the course, rather than an opportunity to just get some points for free.

What do you think about this scheme?

Answer 1

I have used a similar technique, albeit in a course for nonmajors and at a time late in my career when I was willing to acknowledge that my grading was often pretty impressionistic even when I had a rubric.

To maximize what students learned from homework and to minimize the time I had to spend parsing their answers, I would go over the homework in class while the students still had their papers. I asked them to comment on their work, with a different writing implement than the one they used at home.

They could correct simple mistakes, flag serious misunderstandings, tell me which problems they were worried about. I encouraged comments like "I didn't understand this at home, now I do" or even "I didn't understand this at home and I still don't".

Then I collected the papers and read and graded them. That took much less time than it would have without their comments. It also gave me better feedback about problem areas.

Answer 2

This might sound a bit weird, but what if you just award points for submission? It doesn't matter if they got any of the questions right, just that they tried to answer them and turned them in. Make this clear at the start of the course.

Then you can leaf through the submissions to gauge what people understood and what still needs some clarification, and fix it in the next lecture.

Answer 3

I see two problems with this approach:

1) It is unclear how the students should grade their assignment. How many points should they deduct?

Even if you give them the right solutions, sometimes students think "oh yes, in principle my solution was right, it was only written down differently" when in fact some important parts are missing. It may very well be that very self- confident students give themselves more points than non-self-confident students.

2) Depending on how your grading structure works, 10 percent more may result in a better grade than the student actually has earned. (In countries where grades are not considered important, this may not be a problem.) While you can probably say who the good and the bad students are - does the grade you have to give the students by your syllabus say the same about them?

Two tips from me:

1) If you announce to sometimes check their solutions and control if they gave themselves the right amount of points, you can reduce "obvious cheating" a little bit. However, you cannot prevent the nonobvious cheating/ misunderstandings (which I described in (1) above).

2) You might actually not grade their homeworks, but grade their effort: You may announce that students get credits for those exercises which they reasonably tried to solve. Then the students hand in their homework and write down the number of exercises where they did reasonable approaches and sometimes you look over the exercises and check if they did not lie. This is much less work. Of course, here also students can make sometimes different claims - but much less than if they really grade their assignments. (You should make a reasonable effort to define "reasonably", though.) And you give them the solutions anyway (as you planned to do). Merry christmas!

Answer 4

So, instead of letting them grade their own homework, remove the name from the submission and put a code / number on each piece of work, hand them out randomly and they grade each others.

My experience shows they are harder graders than I am - establish a control ie check 10% every so often and they respect it.

Answer 5

You are trying to solve the wrong problem.

The problem is, this year we do not have enough personnel to grade the homework.

Note what I've emphasized. The problem is not "how to make do with not enough personnel", the problem is "not enough personnel". If you try to somehow hide that fact with some kind of a "hack" solution - which you yourself don't think very highly of - you'll just make it even harder to solve the problem.

_{(While not the focus of my answer, I'll also say that your self-grading system is unfair to students. Feedback from a trained academic is of critical importance to their education, especially the feedback about what they have not thought of themselves, or regarding alternative possible solutions which they may have stumbled upon, or errors they make trying to establish what's correct in itself, and so on.)}

"But", I can hear you saying "The dearth of homework graders is a systemic, non-technical problem - not something I can solve merely by my sharpened intellect. And it's not as though I can just tell my department 'we need more people', since other courses probably have a similar issue too."

Well, to that I would say:

It's time to unionize.

(or to shake up your do-nothing shell of a union you may officially have.)

Faculty or University management has allowed itself to cut back on the number of positions beyond the necessary minimum. That has been possible for some combination of the following reasons:

1. It can make decisions without consulting with the faculty
2. It can force its own decisions on the faculty
3. The faculty do not have enough sense of professionalism / perspective on the pedagogic process to voice objections when consulted
4. Faculty representatives are co-opted by management
5. Faculty representatives are not subjected to oversight / do not really need to answer back to their public

These are all issues that a faculty union is supposed to - and can - address. Note that I didn't say you need to start the socialist revolution and take over the world, but if you had a working union, you could (in order of the above issues):

1. Demand from, and possibly force, management to only make such staffing decisions when the faculty agrees that they're reasonable
2. Allow for individual course professors to not have to accept the allocation of TAs/homework graders/lecturers they were given from above, i.e. for the individual (like you) to say "that's not enough" and for there to be a binding procedure for addressing this beforehand, rather than you being stuck with the current situation
3. Educate each other, especially new faculty, and more especially new graduate students who become junior faculty, about things like obligation vis-a-vis the students and vis-a-vis your colleagues, what the faculty believes should be the reasonable workload and why, etc.
4. Make sure that faculty representatives to decision-making processes are at least somewhat beholden to the actual faculty and its interests, e.g. through elections, and also:
5. Make your representatives report regularly to their colleagues, both in writing and in the occasional meeting, and that you all make certain decisions together rather than leaving everything to their on-the-spot personal judgement.

If the above happens, I'm sure you will have both the will and the means to rectify the lack of sufficient personnel. In fact, collective power could allow you to do things that are personally less possible, e.g. get the faculty to invest in automatized question generation and grading, at least in some courses (not algorithms probably), which could free up some of the teaching staff.

PS - Since you (seem to be) Israeli, I recommend you read a few relevant chapters of Dr. Daniel Mishori and Dr. Anat Maor's anthology about Exploitative Employment in Israel. It's available on-line for free (Hebrew only though) and has quite a few chapters on Israeli universities' teaching staff employment practices.

Answer 6

The solution I tried in a class where I had no help for grading and too many students was to not grade homework at all, but to make it clear that a portion of questions from every homework would be on the exam, nearly verbatim. Mostly I shifted numbers and/or signs around so that they couldn't just memorize what they did before and regurgitate it.

I found this worked quite well. I did give a small grade for submitting each homework on time, but the work was not reviewed and I don't think it was a particularly important part of the scheme.

Advantages:

• Low grading work
• The homework and the exam are connected in a tangible way for the students, making it clear that assignments are worth their effort (as opposed to them deciding something worth only 10% of their grade is worth only 10% of their effort)
• You've forestalled any questions about what is on the exam or will there be a study guide, because the homework always is the answer
• Less micromanagement on your part
• According to my reviews, the students liked it. They hate lack of clarity on how to get good grades
• I like solutions that involve going over the homework answers in class time, but that's very costly in terms of student face time that could be used for other things, and this solution avoids that

Disadvantages:

• Might allow less-motivated students to really fail badly if they neglect the homework, where they might have been motivated to do it by hard deadlines

Combine this with you having TAs who can go over this in sections with students, and I think it would work very well for your problem. Also, as a former TA, I applaud you for not just deciding the remaining TAs need to double their workload!

Answer 7

Self-grading is actually advocated by some education researchers (Phil Race is the one that comes to mind, but there are others).

However, the reason for doing it is not primarily about saving your marking time (although that is part of it). It's also about helping students to learn to assess their own work, rather than relying on the marker to tell them whether they've done it right or not, which is an important skill they will need when they reach the real world.

To avoid upsetting the more honest students, you need to be able to explain why it is good for them, not that it is because you haven't been given a grader (bear in mind that you can do the grading yourself, and they may believe you have little else to do all week).

Also, you need to teach the students how to assess their work. Go carefully though some worked examples with them at the start (bad ones as well as good). Show them what they need to look out for. It's pretty surprising how little they pick up on major differences between their work and the model solution (eg, in maths, random mathematical symbols versus full sentences). The rubric must also be very clear.

One way to reduce the cheating aspect would be to moderate the work - either take in a random sample each week, or get them to hand in all their marked work, and remark one sheet chosen at the end (I chose to use a die to decide which to mark, so they can see you're not deliberately picking the one they did badly on).

Answer 8

Instead of having the students grade their own paper why not have them grade their peer's paper. Peer editing will hopefully reduce the risk of students giving themselves an "A". In addition, peer editing allows students to see how others have attempted to solve the problem and it also allows them to discuss how they came to their solution. This cooperative dialog should hopefully heighten understanding of the topic.

If there is a great concern in terms of dishonesty in using peer editing. I recommend having each student write their name at the bottom of the paper they are peer editing. Then I would also encourage you or the TA to randomly mark a few papers each week just to check the honesty of the grades. It keeps everyone on their toes while keeping the workload to a minimum. This also allows you to identify common mistakes to address during lectures.

Answer 9

Does the homework include different parts (for instance, different questions/problems)? If so, you can say that, for every homework, you will only grade one or two questions chosen at random. Because they are chosen at random, they will still need to reply all of them. You can have the additional requirement that incomplete homeworks will not receive any score at all. Then, in class, roll a die to choose the question, so they see it is truly random.

Another approach is that you randomize the homework papers and distribute them, so everyone grades someone else's. That might be delicate if it's a large class and don't know each other very well, or there are frictions that you may not be aware of.

A third approach is that everyone grades their own homework, using a different color/ink than the solution they wrote, and then give them to you. You pick a few papers at random (~10% of the class), or not at random but the ones you think may have inflated the grades, and double-check their grading.

In all these approaches, the probability that their homework will be graded by someone other than themselves should deter most of them from inflating their grades.

And, as someone mentioned, if this is not a consequence of a well-thought teaching decision, but of a lack of personnel, join your union and fight for an improvement of your working conditions!

Answer 10

It seems to me that the amount of work, both on the part of the students preparing and the staff grading, involved with the homework assignments is disproportionate to their share on the grade. 10 assignments worth 1 percentage point each seem a bit too much buck for too little bang, so to speak.

I agree with you about the assignments being useful to improve the course experience and quality; therefore, I think that could be reflected in how they're weighted. If you have 90% weight on the exam and 1% each on 10 home assignments, that seems to me to communicate to students that the assignments aren't important. In trying to improve course experience, I'd try to fix that first.

I realize this may be outside your authority to change, but an avenue that occurs to me is to increase the homework share on the grade. Perhaps to 70-30 for exam-homework? I've had a few assignment-heavy courses where the grade balance ran to 50-50 or even higher (i.e., exam worth less than half the final grade). I'm not saying you need to go that far, especially if the assignments are individually small. But still.

If the balance was changed this way, this, together with presenting the course differently than before (i.e., a course where the homework matters) might be a fairly good argument to get additional staff assigned to grade it (at least if that's a realistic possibility at your school), thereby solving your original problem and, IMHO, improving the course in the process.

Answer 11

I don't know if this would help. But at my uni, we had to submit homework, but it had no grade in itself. It did, however, allow us to participate on a sort of test at the end of the semester (as in, we couldn't participate in it if we hadn't submitted 70% of our homework), in which some exercises from our homework were selected and depending how good we did on these exercises the grade we got. 10% is maybe too little to implement this, but maybe you can think of something similar. Hope this helped

Answer 12

One approach is to collect the entire homework assignment, but then only grade one or two problems (chosen at random). Since students won't know which problems will end up being graded beforehand, they are still responsible for completing the entire assignment. However, it drastically reduces the total number of problems that you need to grade.

The system remains fair, as all students have the same problems graded on each assignment. There will, however, be a small amount of griping early on about how "I did the rest of the problems perfectly, but you graded the only problem I didn't do." As long as you make sure to note that the problems you selected to grade were chosen randomly and remind them that sometimes they might get graded on the only problem they did perfectly, students will generally buy into the system. I would further advocate for this system as homework accounts for a smaller portion of the final grade under your syllabus, so any random variations should be reasonably minimized.

I have also tried doing student grading in the past, but as other answers have noted, this can end up consuming a large amount of precious classroom time and you are forgoing your ability to give feedback---one of the most important reasons to do homework.