Jeremy Sharapov
Cannon Editor
In my previous article reflecting on the online courses I was taking, I said that the details of how to run midterms and exams in the most student-friendly manner was too much to fit in. Well, that’s because there was enough content to fit in its own separate (and lengthy) article, so I separated my key points here into what to do and not to do in terms of online testing, in my opinion of course.
Do: Be flexible with the timing for any examinations.
This is probably the most important, as there are students in a variety of time zones, some of which would be really inconvenient to take an exam at a given time in EST. Personally, exams, where you have the entirety of, say, 24 hours to complete, are my preference, followed by those where you choose a certain time slot within a larger window. But, professors should at the very least provide one alternative sitting, with a 12-hour time difference from the standard one seeming to work best.
Don’t: Split up the test into multiple time windows to complete and submit each question.
I recently had a midterm with this format, and it was by far the worst format I’ve experienced thus far. I ran out of time on the first part that was worth the most marks, but then finished the other two that were worth way less early. However, because of this format, I could not use the extra time on that first question unlike in-person tests, and so, bombed it.
Do: Let the students go back and double-check previous questions.
Building off the last point, another more common (and only moderately less annoying) testing format some professors have been opting for is a Quercus quiz where it only displays one question at a time, where you can’t go back and double-check previous questions. This too contradicts the conventional way of solving a test: looking through all the questions first, getting a feel for what you need to solve and then solving them based on what you know best first. Now instead, students can’t figure out how much time to spend on each question, resulting in them finishing too early or not at all, and get screwed over if they suddenly have the solution to a previous question pop into their mind while reading the next. The logic that these prevent cheating is ridiculous, as forcing students to do the exact same questions in the same order only promotes collaboration for those questions because they do not feel in control of how they solve it, versus doing everything in a preferred order.
Don’t: Mislead students about the exam format.
I will never forget one exam where the professor said that there would be a singular short, conceptual question with multiple parts, in addition to three more calculation heavy questions. I completed the three questions that I thought were the calculation ones (they each had multiple parts and involved actual numbers) and spent ¾ of the time on them accordingly. When I got to the next question, which was clearly way longer and more computationally intensive, I just froze. I had just spent ¾ of the time on the exam for ¼ of the marks and had no way of knowing beforehand because this was one of those tests where you could only view one question at a time and not go back, and pretty much everyone else in the course I talked to afterwards did the same. I spoke to EngSoc’s Vice-President Academic and they confirmed that this particular exam broke several rules, however, the department didn’t do anything to remedy this, and did not curve the course to boot. Thankfully CR/NCR existed that semester, but if any prof tries this sort of thing again, I’m sure the students won’t let them off that lightly. Another time, a professor said a midterm would be in the same format as the quizzes in the course, but the test had implemented the “one question at a time, no backtracking” approach unlike the quizzes, which too infuriated me. And yet another time, a prof said to have this one piece of software we use for labs open because there will be a question on the midterm that utilizes it, but there was not and it just made my laptop lag for the entire session.
Do: Hold optional live sessions during exams where students can ask questions.
Sometimes when reading a question, I find the wording not particularly clear, so I ask a TA in the Exam Centre to clarify what they want us to do. Other times the exam may have typos, and professors announce this to everyone. That’s why it’s extremely important to hold some sort of optional session where these exact things can be done. Just be sure to make sure students can’t unmute themselves, and only ask questions in the chat.
Don’t: Make these live sessions mandatory or try to use some sort of proctoring software.
Many students don’t have a reliable internet connection, so when a prof told me that I had to stay on the BBCollaborate session the entire length of the midterm or be forced to submit then, I was quite nervous as my internet had been unreliable in the past. Thankfully it did not crash. However, this prof also insisted on having TAs try to scan TCards to verify our identity during this session, yet they did not end up getting to scan the majority of students. What is the point of doing this and checking out the physical environment when this midterm was supposed to be open book anyway? I personally found the session running on my laptop extremely distracting, and strongly prefer only hopping onto one when I myself had a question. Any corrections or clarifications should also be sent as a Quercus announcement for students not on these sessions.
Do: Be lenient on late submission penalties.
As much I understand that lateness is usually not tolerated in a physical exam room, as I mentioned earlier my own internet connection is not the most stable, and sometimes submissions can take longer than I expect them to. That’s why it’s extremely important to be more lenient on late submissions this year for students with poor internet. On some Crowdmark exams, I’ve heard that professors take off like 25% in marks just for submitting on the hour versus a minute before, even though this was technically not communicated to students beforehand. In some other courses, I’ve heard that if you’re even a second late the professor hands you a fat zero just for that. This is extremely counterintuitive; if you’re that harsh you just know students are going to petition the midterms to get their rightful marks anyway, so why hassle your department with more paperwork? More fair lateness penalties I’ve seen is stuff like 5/10% off per minute late, which can add up, but isn’t too punishing if someone underestimated their upload time by half a minute.
Don’t: Make the test significantly harder than in-person counterparts.
This one is all too common from what I’ve seen; profs expect us to perform way better because we can look at our notes and textbook during exams, however, this is not always the case. Sure, if I don’t know something, I can look at my notes, but it can take me up to five minutes to find the thing I’m looking for, which wastes a lot of precious time! Profs also claim this is to counteract cheating, but all this does is promote it because honest students end up performing extremely poorly, and only those who cheat do well, as compared to a normal exam where an honest student can still do well themselves. I’ve had a prof tell me that the average on the one midterm we had, worth 30%, was a 30/100, and by far the worst mark in the history of the course. Please have realistic expectations of us — all of this only adds up to our increased stress this year.
Do: Admit if you messed up, remedy that, and be lenient on students.
In that course I just mentioned, the prof curved everyone’s midterms by 30 marks and said he’d shift around the weighting of the midterm for people who do well on the final. In another course, after a particularly bad quiz, the profs announced that they would drop the lowest quiz mark. These sorts of accommodations go a long way in making students feel less stressed in the course, instead of relying on a large curve at the end, which does nothing for our stress during the semester. Recently, I had a midterm with multiple typos that made the questions impossible to solve and clearly contradicted the provided solutions. The prof refused to post the average, acted passive-aggressive in emails, and shut down our Piazza after students were understandably upset. Don’t be that prof.
Do: Allow students to upload their work for questions.
Last semester, I was doing really well in a course until I got back my final mark which meant that I had to have gotten a 55% on the final exam. I thought I did well, but it was all multiple choice and fill-in-the-blank, so if I got a question wrong, I got zero marks for it. My department has recently said that if any prof needs extra TA resources to mark a midterm, they would provide that to them, so please ask the department and do allow students to upload their solutions and not just final answers. While we’re on this topic, if we’re asked to upload each question individually, please separate the PDFs per question and give us enough space on them to do the solutions, for us tablet users.
Don’t: Equate a single midterm with multiple smaller “quizzes.”
This is something that is also very prevalent this semester, as instead of high weight on the final or a single midterm, many courses are instead opting to have 3-4 “quizzes” throughout the semester. I hate using the term “quizzes” because when they’re worth 15%, they’re basically the same as a midterm. The critical error many profs are making this semester is assuming students would be spending the same amount of time to study for those 4 quizzes as a single midterm in a regular semester. This is not true, we typically study about as much for each quiz as we would a midterm because of the high weight, which although it does seem to be better for retaining more information, still increases our overall workload tremendously. It can be extremely hard to juggle three quizzes and two labs in a single week, as I have experienced firsthand, and this sort of week can happen multiple times in the semester for some students. There is no ideal solution, as I personally did way better on these smaller tests than the usual big midterms this semester, but I hope this would be some food for thought.
Do: Give students timely individualized feedback on their midterms and have a clearly outlined remarking process.
What’s the point of testing students without letting them know what they did wrong? That’s part of why allowing them to upload their work is so important. I have found that Crowdmark works so much better for uploading solutions than Quercus, as you can see what you have submitted afterwards, and the TAs can comment directly on that submission. In one of my courses, it took so long (like three weeks) for the TAs to finally upload their comments on my solution that I legit forgot what I did for several questions that I would’ve liked to get a remark for. And in some courses, the procedure for remarking is extremely unclear. Remarking is especially important because I, and many of my friends, have gotten many errors fixed and our marks significantly improved.
Don’t: Make multiple versions of exams that are extremely different.
While it is more understandable to do so when there are two sittings, making them extremely different just frustrates students and makes them think the process was unfair. This happened in one of my courses, where there were two versions of the midterm (taken at the same time). All my friends who took the other version seemed to do way better than the ones who got mine. The prof, when he announced the averages, didn’t even acknowledge that the two versions might have had different averages, which is incredibly frustrating. Other courses with multiple versions did show the different averages, and the profs promised to curve differently, yet, there is no avoiding a sour taste in one’s mouth. One solution that I found especially smart was asking students to use some digits of their student number in part of the question, making them unique to each student to prevent cheating. However, it is important to only consider one or two digits at a time, because having to solve a circuit with 10-ohm resistors is very different from having to solve on with 827-ohm resistors.
Just to wrap up my thoughts here, I do want to reiterate that all of these recommendations are of my own opinion, no matter how objective some of them seem to me. If the tone of this article was a bit passive aggressive, it was because I had personally experienced everything I wrote about, both the good and the bad. While those exams that hit a lot of my ‘don’ts’ frustrated me to no end, there were many that went quite well in terms of format, even though I may not have done particularly well on them myself. I hope that this shows that there are indeed ways to make online testing less cumbersome for both professors and students, and really do think it would be in everyone’s best interest if some of these recommendations were implemented on a larger scale. And to other students reading this, if a test hits a lot of these don’ts and feel unfair, do reach out to your class reps so they can at least express that frustration to your professors.