New Research Shows (Again) That Test Proctoring Reduces Cheating

Plus, twins win $1.5 million in cheating accusation case. Plus, news from Ireland and Zimbabwe and the UAE.

Issue 169

To join the 2,611 smart people who subscribe to “The Cheat Sheet,” enter your e-mail address below:

If you enjoy “The Cheat Sheet,” please consider joining the nine amazing people who are chipping in a few bucks a month via Patreon. And thank you to those who are.

---

Another Study: Exam Proctoring Reduces Cheating

In the last Issue, I referenced this new study by Michael Henderson of Deakon University and Rebecca Awdry, Matthew Mundy, Mike Bryant, Cliff Ashford and Kris Ryan, of Monash University. Both universities are in Australia.

Having had a chance to review it now, here are some notes worth passing along.

The study includes a large sample size of nearly 8,000. But, as is often the case in academic integrity research, it’s a survey of self-reported misconduct which we know undercounts negative behavior.

The survey also has a self-selection problem and a significant population problem in that it was offered to students following the taking of an online exam, asking students whether they cheated on it. I’m not sure how many cheaters were eager to take a cheating survey right after having cheated, especially if it asks - hey, did you just cheat? The authors acknowledge these points:

Given that cheating behaviour is penalised, it is likely that some students who engaged in cheating may not have completed the survey, or did not accurately report their behaviour.

This self-selection probably partially explains why the survey reported such a low rate of cheating - just 2.8%. The affirmative response rate was to the query, “I broke the exam rules” - the exam. And since subsequent survey questions asked about cheating in “prior” exams and on “a previous exam” and separately and specifically about “this exam,” that’s probably what happened.

If that’s the case, a 2.8% admitted cheating rate for a particular, just completed test is quite high. Another 4.2% declined to answer whether they cheated. That makes 7% who either admitted to cheating or declined to answer whether they cheated on an exam they just finished. That seems alarming.

When asked a more comprehensive question of cheating behavior broadly - on “previous” exams - we see more expected results:

6.8% (n = 533) of the sample reported cheating once or twice prior, and 1.2% (n=91) reported cheating three or more times on a previous examination. The remaining 6.7% (n=525) preferred not to say.

About 8% self-reporting cheating on exams in general and nearly 15% either admitting cheating or declining to say they did not cheat, is still low. But it’s not an outlier. In their literature review, the authors say:

Research conducted in Australia, the UK, Canada and the US in the last four decades has found rates of cheating in examinations have ranged from 9.3% to 64%.

In other words, given the context in which this instrument was conducted, its results don’t really stand out. Though I am certain some folks will not understand that and incorrectly report that the rate of cheating is 2.8%, which will give me a migraine.

And now, having spent way too much time to make that point, the real finding from the survey is that - yet again - we see that proctoring remote examinations reduces cheating. Surprise, surprise.

Briefly, the survey measured remote exams conducted under four settings: passive, record and review proctoring and a live onboarding, passive proctoring with an automated self check-in, no proctoring but use of a lock-down browser and no security provisions at all - neither proctoring nor a secure browser. About which, the authors found:

that students who experienced Safe Exam Browser (SEB) reported statistically more cheating than students who were supervised online (both assisted and self-check in)

The rank of admitted cheating rates - as asked specifically about a single recent exam - was secure browser only and no proctoring (5.4%), no detection methods at all (3.3%), passive proctoring with live check-in, then passive proctoring with self check-in (2.8% and 2.3% respectively).

The authors say:

Online proctoring tools have the ability then to reduce possible instances of cheating, as well as the perception that others are cheating.

Even with all the caveats and a survey design that probably pushed all these responses into the single digits, I will say it again - when you change the risk/reward calculus by increasing the likelihood of detection, cheating goes down. When you act like you value the integrity of an exam, that message is received.

And to underscore, the proctoring methods used and studied here were entirely passive - not live proctoring. They were also the school’s own, in-house systems so we don’t have much visibility as to quality.

A few other, hopefully quick, notes on this research.

The authors found that cheating was slightly more likely, though not statistically significant, during an “open book” exam and that cheating is significantly more likely if examinees are given a broad test-taking window of say, 12 or 24 hours instead of a fixed test time. Neither is surprising.

Finally, the authors cite what they perceive as “counterintuitive” data - that those who reported cheating said they were more likely to know other students who cheated and more likely to know students who had been caught cheating. They say:

Our data confirmed research that students who cheated were statistically more likely to report knowing of others who cheated. However, in contrast with prior research – and perhaps counterintuitively - our data revealed that students who cheated were also more likely to know students who had been caught cheating.

On self-reported cheaters saying they know others who cheat, that’s pretty normal. There’s a link between people thinking others are cheating and cheating - be it actual or rationalization. Or both.

But to me, the finding that self-reported cheaters say they are more likely to know someone who’d been caught cheating is not necessarily counterintuitive. To be against presumption, you must assume that the observed consequences of cheating were significant enough to be a deterrent. If you instead assume, as I do, that even the few students who are caught cheating rarely face any significant consequence, then a link between knowing someone who’d been caught and cheating is perfectly intuitive.

And finally, finally - I just have to share this quote from this research as it’s another very strong contender for the 2022 academic integrity quote of year:

Students who cheated placed lower importance on academic integrity than students who did not cheat

There you go.

Twins Awarded $1.5 Million in Cheating Accusation Law Suit

Several news outlets reported on a jury award of $1.5 million to identical twins who said they were falsely accused of cheating on an exam in medical school.

According to the news coverage, the students attended the public Medical University of South Carolina, and were accused of cheating during an exam. They ultimately dropped out and sued, winning the award in part by convincing a jury that their similar answers were not due to cheating but because, as twins, “their minds were linked.”

A few things here briefly.

One, the accusation was not just similar answers but observed signaling during the exam, which the twins denied. The cheating accusation was upheld by a school review but ultimately overturned by the Dean. The pair dropped out anyway, saying the ordeal had damaged their reputation. After quitting medical school, they went to law school and are now lawyers who handle defamation cases. So, clearly, their damaged reputations as cheaters didn’t keep them out of law school. They are also daughters of a South Carolina state legislator.

So there’s that.

I’m not sure what academic integrity experts the school or state used to defend the case but the “minds linked” defense to cheating is an interesting one.

UAE Professor Writes on Academic Integrity

Zeenath Reza Khan, an assistant professor at the University of Wollongong in Dubai, recently wrote about improving academic integrity in - where else? - Times Higher Ed.

At the top, Khan gets credit for setting the table accurately:

The academic world has been grappling with issues of misconduct for decades. Student cheating has been recorded to be as high as 75 per cent in classrooms worldwide, oscillating between the figures based on student self-reporting. Whether it was William Bowers’ study in the 1960s or Donald McCabe, Linda Trevino and Kenneth Butterfield’s in the 1990s, or even the late Tracey Bretag’s more recent string of publications, statistics remain similar.

Yup. It is a problem.

But he also gets it upside down when he later writes:

Prior to the pandemic, studies suggested that there were either no real differences between cheating levels offline and online, or in fact that students online were less likely to engage in misconduct than their face-to-face counterparts.

True, a few studies did suggest that cheating online was comparable to cheating during in person classes or assessments. But like three. One had a sample size of 30 and another is so frequently misquoted it drives me crazy - see Issue 23. But there are dozens of others showing the opposite plus a nearly unanimous consensus regarding the prevalence of cheating online since the pandemic. As he said, whatever divergent views there may have been on this point, they were “prior to the pandemic.”

The professor raises that point to ask why anti-cheating technology in online classes isn’t stopping cheating. I say, that’s because it can’t. Nothing can stop cheating.

But Kahn gets great kudos for acknowledging the baked-in distance between teachers and students in online environments and its role in misconduct:

During the Covid period, we no longer had the luxury of looking into our students’ eyes when discussing concepts, which meant that there were enough times when we couldn’t decipher whether they understood us or not. It also meant that we were unable to establish a relationship with our students, to build trust and mutual respect.

I agree.

Later, the professor stresses the importance of culture in ensuring integrity:

Technology has led to an invariable cat-and-mouse game in classrooms – virtual or otherwise. We know that it is not enough to simply implement software to help “stop” cheating. Academic integrity cannot be a byline, afterthought or by-product of a different conversation in board meetings. Instead, to develop and uphold academic integrity, we need to make this a conversation that takes centre stage in all aspects of university life.

Again, I agree. Though I don’t know anyone who said technology alone was “enough” to stop cheating. It’s required, but not sufficient.

Anyway, good for the professor for speaking up about the challenges in meeting academic misconduct. And good - again - for Times Higher Ed for opening its pages to these voices.

Ireland Says it Withheld Certification Results from 10 Students for Suspected Cheating

Reporting in Ireland says that, on suspicion of cheating, authorities have withheld certification results from 10 students in their “junior cycle” exams. The paper says those subjects include “Irish, Maths, History, Business Studies and Religious Education.”

I don’t know what to make of the number - 10. The coverage says 67,130 students took the exams. So, 10 seems implausible. Authorities withheld eight results on suspected cheating the previous year.

Something seems off.

Zimbabwe Arrests Police Officer for Having a Leaked Law School Exam

Coverage in Zimbabwe says a police officer has been arrested after he was:

allegedly found in possession of leaked University of Zimbabwe Faculty of Law examination paper.

Courts have denied the officer’s request for bail as he’s also been accused of contacting witnesses in the case and urging them to delete evidence.