CAN WE PREVENT COLLABORATIVE CHEATING?
When students sit exams at different times, there is a significant potential for cheating, but researchers show how to prevent it in a paper to be published at the London Festival of Learning today.
Unsurprisingly, if you offer an exam over a period of days and let students choose when to sit for it, some students will take it later so that they can learn about what is on the exam from those that have already taken it. The researchers refer to this behaviour as “collaborative cheating” and find in a large enrolment engineering course that around 5 per cent of students studied in ways that indicated that they had been told information about what was on the exam.
They find that these students gained a 12 percentage point advantage (on average) for any questions where all students are given the same question, even if the questions are modified in simple ways like changing the numbers in a maths question.
Prof. Craig Zilles, one of three researchers from the University of Illinois in the United States on the project, said: “Even with class policies that indicate that students shouldn’t communicate about exams during the exam period, we know students do communicate.”
The solution is to not give the same questions to all of the students, according to the researchers. Instead, for each question on the exam, pick randomly from a pool of equally difficult questions. In practice, the researchers find that having three to four questions to choose from in each slot of an exam is sufficient to make the advantage from collaborative cheating to be negligible.
The researchers believe that two factors contribute to this effect. First, it is harder to get complete information about what questions are on the exam because a potential cheater needs to talk to more fellow students. Second, a potential cheater needs to remember a lot more material.
You might think it might just be easier to give all of the students the exam at the same time, but running exams asynchronously is key to a strategy to make exams better for both students and faculty in large classes. This research was done in the context of the University of Illinois’s Computer-based Testing Facility (CBTF), which is used by a number of its largest engineering classes.
“It’s a whole different paradigm for running exams in university classes,” says Prof. Zilles, “Using computers in a proctored facility, we can run more frequent, shorter exams which are less stressful for students and lead to better learning outcomes, and we give students the flexibility to take exams at times that are convenient for them. It’s a win-win.” Asynchronous exams are necessary because classes with hundreds of students are many times the size of the facility.
The researchers’ findings also apply to Massively Online Open Courses (MOOCs) which allow students to take exams at their convenience. Preventing cheating in MOOCs, however, is even harder, because a single student can register for the same MOOC multiple times meaning that they can perform “collaborative cheating” without having to involve other students.
How much randomization is needed to deter collaborative cheating in asynchronous exams by Binglin Chen, Matthew West and Craig Zilles.