Best Practices for Remote Examinations

Published 04/02/20. Last updated 04/21/20.

The most significant challenges with exams delivered remotely are the need to deliver asynchronous exams to accommodate students in different time zones and the issue of cheating. There is no fool-proof solution to the challenge of cheating on online exams. The campus is working to secure a browser lockdown product that locks the browser into a single screen and disables copy/paste functionality during an exam. However, this product, which is expected to be ready in time for final exams, will not provide visual monitoring. Be aware that the use of other proctoring products and Zoom proctoring are not permitted for the spring 2020 semester, meaning that visual monitoring will be unavailable. Details of the University policy on exam proctoring this semester can be found here.

You may use Zoom to make yourself or your GSIs available for questions during an exam, that is, for administering an exam, but not for proctoring. As long as video is not required, and the Zoom availability is set up in such a way that no student's DSP accommodations are transparent to other students taking the exam, using Zoom in this manner does not constitute a violation.

Faculty may administer a graded student presentation, including as an alternative assignment to a final exam when appropriate, so long as students are not required to use video or penalized for choosing not to do so, and the format is not in conflict with any DSP accommodations. Screen sharing of materials (e.g., PowerPoint slides) via Zoom during the presentation is allowed.

The most effective way to safeguard academic integrity is probably to deploy forms of assessment that are less susceptible to cheating than written exams: oral exams, papers, group projects, poster sessions, discussion boards, etc. If written exams must be administered, there are ways to discourage, if not eliminate, cheating and to mitigate its effects on other students. Conducting an oral examination via Zoom runs into many of the same problems that Zoom proctoring does. For this reason, oral examinations should be conducted by telephone.

  1. Do not curve exams – Students who cheat may still get their A’s, but in the absence of a curve, their “success” will not reduce the chances for other students to get A’s as well.

  2. Make all exams open-book, so that students who consult notes and books do not gain an unfair advantage over students who adhere to closed-book rules.

  3. Schedule multiple, short, low-stakes tests, rather than one or two lengthy, high-stakes exams. This approach is shown to be superior for promoting learning. In addition, students may be less tempted to cheat if the stakes of the exam are relatively low and less able to cheat if the exams are of short duration.

  4. Schedule exams at a specific time (unless that would negatively impact students in your course)
    • This tack would reduce the opportunity for early testers to tell their friends what questions are on the exam. The downside, however, is that with UCB students spread all over the world, it could force students in far-away countries to take exams in the middle of the night, which would be patently unfair.

    • In some smaller classes, all of the students may be within a few time zones of California, making a fixed-time exam feasible.

    • Another solution might be to offer two (or three) different versions of the exam at two (or three) different times, presumably creating a reasonable exam time for every student. The downside, obviously, is that instructors and GSIs would have to write multiple exams.

  5. Use bCourses Quiz and Assignment settings in response to the COVID-19 pandemic and the work of the Instructional Resilience Task Force. These recommendations include:
    • Shuffle answers to questions - For multiple choice-style quiz questions, you can set bCourses to shuffle the answers so that one student’s answer “A” is another student’s answer “C,” for example. This function can be set globally when you create the quiz by checking the “Shuffle Answers” setting, although this works less well for multiple-choice problems with “All of the Above” or “None of the Above” choices. Alternatively, the shuffling of answers can be selected on a per-question basis.

    • Vary the questions slightly - If appropriate, you can create a formula question that gives each user a slightly different version of the question.

    • Randomize questions - A question group can be used to randomly assign students a certain number of problems from a pool, so that every student’s quiz is different. It can also randomize the order of the questions in a student’s exam.

    • Publish an exam as a timed quiz in bcourses, with adjustments to the default time for students with accommodations. Once a student starts, they will have a certain amount of time to complete it. This way you can maintain an element of control over the structure of the take-home.
    • Turnitin can be used in Assignments to detect potential plagiarism in essay questions.
    • Alternatively, you may choose to have a remote examination that is deliberately collaborative, making use of bCourses groups to encourage students to work remotely together.

  6. Institute an honor code for exams along the lines of what is done at Princeton and Stanford.
    • Students write something on their exams along the lines of: “I swear on my honor that I have neither given nor received aid on this exam.”

    • Exams are not proctored.

    • Students are sometimes required to pledge to turn in any student they find cheating or, at least, to intervene to try to stop the cheating.

    • There are generally stiff penalties for violating the honor code (failure of exam, maybe even suspension).

    • Princeton (https://honor.princeton.edu/) and Stanford (https://undergrad.stanford.edu/academic-planning/cardinal-compass/your-questions-answered/what-stanford-honor-code) both have this kind of honor code, and instructors do not proctor exams, yet there is no indication that cheating is especially widespread on these campuses.

  1. Randomize Oral Examinations. To alleviate concerns by students on cheating and to facilitate the type of feedback that instructors benefit from when exams are proctored in person, the following procedure can be used.
    1. Prior to the examination a random set of integers is generated that are then matched to an alphabetized list of students from the course. Instructors may wish to have a third party witness this process.

    2. The number of students selected in this process is at the discretion of the instructor, but it should represent a meaningful sample size relative to the number of enrolled students.

    3. After the exam and prior to assigning final grades, at the discretion of the instructor, all of the previously selected students may be questioned about their solutions to the exam. It is advisable to select the questions in advance and ask the same questions of all of the students. 

    4. If the instructor finds evidence of cheating by a student as a result of the examination, then the student should be reported to the Office of Student Conduct or the matter should otherwise be addressed in accordance with the campus Student Code of Conduct.
    5. For this process to be utilized, it
      1. Must be announced to students sufficiently far in advance of the examination for students to self-identify to DSP and seek accommodations from that office.
      2. Before the final exam, instructors should determine the consequences for a student who fails to appear for the oral examination.
      3. The process should exhibit no bias and should be administered in such a manner as to  be defendable to other instructors.

In some instances, instructors may also want to use oral questioning to confirm written exam performance when the instructor has reason to believe that a student has cheated (but not as part of the examination process itself). Instructors are authorized by the campus Student Code of Conduct to conduct such an inquiry regarding academic integrity. For instance, instructors could orally confirm the demonstrated written performance of all students whose exam solutions suggest cheating, if this is done in a way that is based solely on their answers and is "blind" to the identity of the student. Different students could be asked about different problems on the exam, so long as the process exhibits no impermissible bias. Students could also be questioned by requesting a written response, instead of an oral response. If the subsequent inquiry causes the instructor to believe that cheating occurred, then the allegation of academic misconduct would be resolved in accordance with the requirements of the campus Student Code of Conduct

Instructors who are considering this sort of systematic review or exam performance are advised to seek guidance from their Chair and/or the Office of Legal Affairs regarding their specific post-exam protocol.  

 


Compiled by Jonah Levy and members of the Undergraduate Council and the Instructional Resilience Task Force of the Berkeley Division of the Academic Senate.