Giving good feedback to students is a big challenge. When a student submits some assignment, it usually doesn’t take me that long to get a good idea of what an appropriate grade would be. I’ve seen enough assignments that I might not even have to read every in detail or everything at all. Writing it up, however, takes time that I don’t always have. Paradoxically, the best students suffer the most from this. I see little meaning in adding dozens of “well done” remarks on arbitrary pieces of the text when basically the entire text is well done. So, good students might end up thinking that I didn’t fully appreciate their work and did not give it the credit by highlighting what they did right and wrong.
This semester, I’m using peer-grading for the very first time. For that purpose, I’m making use of peergrade.io, which I find to be a quite comfortable online platform. I’m getting back to that in a moment. Peergrade.io markets peer grading as a tool to safe time and suggests to outsource feedback for assignments to students. I am opposed to this idea. Student grading cannot replace teacher grading. There is more authority in the grade of a teacher, and there could always be group dynamics in classes with adverse consequences. Bullying may not have an influence on grades ever.
I see two other advantages in peer grading. First, students (to some extent) learn from seeing alternative solutions to the same problem. Second, in principle, student grades should correlate strongly with the grades that I would give to the same assignment. If that is not the case, something has gone wrong. That in itself could be a helpful diagnostic. If things go right, it actually adds credibility to the grades that they receive by authority.
It still helps me to save time in some way. I now can write shorter feedback, highlighting the key issues, and in a more standardized and equal fashion across groups. Students will have detailed feedback from their peers on top of that. However, I now also need to administer the peer-grading. Any time that I saved, I’ll spend on that. So, nothing to gain there other than a more effective use of time.
It won’t work out of the box. If you want your students to grade, you need to give them guidance. Hence, you need to develop grading rubrics that potentially are more explicit than the ones you’re using for yourself. That likely depends much on how extensive your feedback has been in the past. When students are supposed to assign five points to task, I refrain from giving a clear, verbal explanation for all five steps. Instead, I label the bottom, the top, and the mid-tier. If there are more steps, not just five, I add a small number of mid-tier labels. It reduces effort for students, as they have to process less information, and it reduces my effort to develop unambiguous labels that don’t overlap. I’m very fine with accepting some additional degrees of freedom on the other hand. For me, it’s part of the idea. Peer-grading cannot be a perfect match with my evaluation. I’m aiming for a high correlation only.
I noticed that it works better with shorter assignments. It’s a bit strange, as less text intuitively gives less room to differentiate students when they prepare their assignments. I cannot confirm this intuition; I can confirm that shorter assignments lead to better outcomes. I firmly believe that we make students write too much. (This is a long post, which is kind of ironic.) Weaker students won’t deliver better assignments if they have more space to fill but don’t know how. I also find them more difficult to grade and I feel more depleted afterwards. This is my profession. If I feel depleted, I should not expect my students to have higher stamina.
One more thing. Students don’t always get teaching innovation that have no influence on their grade. My way around this is to turn the three assignments of my course into a contest. For each assignment, the best three teams can earn one extra credit for their exam grade. (Bocconi’s exam grades are on a scale from 0 to 30, and my exams take an hour. One extra credit seems to have about the right impact.) You do bad in the first assignment? You have another chance the next time around.
In order to facilitate the administrative part, I’m using an internet platform called peergrade.io. There are free plans for smaller courses and that has been enough for me so far. I could manually enter all student e-mail addresses, or I could give them a sign-up link plus a course-specific code. Students do not need a private e-mail address to set up, but should be instructed to register with a username that allows to identify them.
My university has a Blackboard subscription. Blackboard has a peer-assessment function built-in, but it’s not available at my institution. What’s more, from what I understand, it’s based on a short-question-short-answer model, while peergrade.io can handle file submissions. And more. Videos, podcasts, links, presentations, and virtually any type of media that could come to my creative mind. You could give absolute freedom to your students. I have not gone that far, but I intend to include a video-based exercise in a course in the next semester. Preparation of an assignment includes defining grading rubrics. Some copy/paste functions make it easier to work with repetitive rubrics. I’ve noticed that peergrade.io will not allow more than six steps per scale when assessment uses discrete scales. (Open-ended text boxes and various other modes are available as well.)
Once submitted, teachers can allow students who forgot to submit to submit late. Once the peer-grading period starts, students get randomly assigned to other assignments. I can customize how many assignments they shall evaluate. A dashboard for each assignment allows me to monitor who has given assessments, who has received them, and how good or bad this assessment scores overall. I can also see how high agreement is on each assignment, the average within-assignment agreement on each question across assignments, and how long students spend grading. I find that immensely helpful to improve my own grading rubrics (also for myself) and in order to check that things are not running out of hand.
One more thing. Two, actually. Teachers can use the platform to grade assignments as well. They can select whether that shall overwrite peer-evaluations or how students grades and teacher grades should be weighted. Students can flag feedback from other students that they deem unfair. If you give full authority to your students to determine grades, that’s an important of arbitration.
I’m very positive about the platform, but I’m not all happy about the group submission function. Currently, students can select to submit as individuals or as a group. If they select the group option, they next need to manually select their group members. While that is great for changing group compositions, it’s an invitation for students to mess up things. I’ve had regular instances in which group members submitted separately as individuals. What can happen is that student C from group 2 gets assigned the submissions from student A and B, who are both from group 1. Or that student A is assigned the submission of student B. Fixed groups as an additional option could help with that.
Finally, as a teacher, I cannot delete assignments that were submitted by students. Only the student who submitted can. I’ve had a case where student A submitted for group 1, but forgot to add his group member student B. I contacted the group and asked them to correct their submission, but student A had gone on a weekend trip and was offline. Student B and his fellow team members could not delete and/or update the submission, and neither could I. Would I have the power to delete submissions myself, this and other problems could find faster, more student/teacher-friendly solutions.
That said, I haven’t encountered any other problems with the core functions. I use it as a free tool. Paying attention to the above, it works like a charm. I only can’t get the summary statistics per students working. (That’s a different data table, not the Dashboard I mentioned above.)