How to Stop Cheating in College

Can new technologies help counter today's ever-evolving strategies for cheating—and discourage students from doing it in the first place?

bibiphoto / Shutterstock / Zak Bickel / The Atlantic

Cheating is omnipresent in American higher education. In 2015, Dartmouth College suspended 64 students suspected of cheating in—irony of ironies—an ethics class in the fall term. The previous school year, University of Georgia administrators reported investigating 603 possible cheating incidents; nearly 70 percent of the cases concluded with a student confession. In 2012, Harvard had its turn, investigating 125 students accused of improper collaboration on a final exam in a government class. Stanford University, New York State’s Upstate Medical University, Duke University, Indiana University, the University of Central Florida and even the famously honor code-bound University of Virginia have all faced cheating scandals in recent memory. And that’s just where I stopped Googling.

The nationwide statistics are bleak, too. The International Center for Academic Integrity (ICAI), which has studied trends in academic dishonesty for more than a decade, reports that about 68 percent of undergraduate students surveyed admit to cheating on tests or in written work. Forty-three percent of graduate students do the same.

It’s easy to blame high levels of student dishonesty on new technologies, which can make cheating a matter of a swipe of a finger, rather than a stolen answer key or elaborate plot to share answers in the testing room. In a 2011 Pew study, 89 percent of college presidents blamed computers and the Internet for a perceived increase in plagiarism over the previous decade. Meanwhile, colleges are turning technology against the cheaters, using software products that proctor tests with webcams or check written work for plagiarism.

But Don McCabe, a retired professor at Rutgers University who led the ICAI student surveys for many years, is hesitant to blame today’s student cheating rates on easy access to the Internet, computers, mobile phones, and more. His survey data shows a more complicated portrait: The percentages of student cheating did begin to increase once the Internet became ubiquitous, but now are actually trending down again, toward pre-Internet levels. But he also sees a diminishing level of student participation in his surveys—fewer responses, and fewer thoughtful responses. His theory is that there’s a growing apathy toward school and cheating at school among today’s students.

What is the best way for universities to catch today’s ever-evolving cheaters—and discourage them from cheating in the first place?

One approach is to take advantage of a number of new technological tools like Turnitin that are designed to make academic dishonesty easier to sniff out. Turnitin, for its part, is a web service used at institutions around the globe to analyze written schoolwork, giving students who run their papers through it computer-generated advice on their writing’s organization and sentence structures. And it gives professors a grading platform that compares every sentence in a student essay to a big database: billions of archived web pages, millions of academic articles and—perhaps most interestingly—most of the other student papers submitted on Turnitin in the past, more than 337 million, according to the company’s website.

Some new ways of uncovering cheating may feel a little creepy. Proctortrack, a software that monitors online test-takers through a webcam, identifies slouching, stretching, shifts in lighting and picking up a dropped pencil as potentially dishonest behaviors. The company behind it, Verificent Technologies, says that Proctortrack is currently installed on 300,000 student computers, with over 1 million online exams proctored since its release.

Other data experiments happening in higher education could have implications for how schools patrol for cheating in the future: many universities are starting to use demographic data like the student’s age and family education history alongside information on classroom engagement to predict a student’s likelihood of passing a course or even of graduating in four years. It doesn’t take much to imagine how quantifying expectations for how well a student will do in class might sharpen the search for cheaters.

But some worry that over-reliance on technological methods in the fight against student cheating risks a technological arms race, with students inventing still-more-ingenious ways of beating the system and institutions using more and more invasive means of catching them.

Indeed, Teddi Fishman, the current director of ICAI, sees a link between the technological environment and the popularity of different strains of cheating. For the past decade, for instance, she’s seen cut-and-paste plagiarism increase steadily. But now, with the advent of plagiarism-checking technologies like Turnitin, cut-and-paste is falling by the wayside, replaced with what Fishman refers to as “bespoke essays or contract cheating”—services that write papers on behalf of a cheater, a much more difficult practice to police with the technologies currently available.

And even as existing algorithms get scary good at identifying student behavior that deviates from the expected norm, Turnitin, Proctortrack and other tech tools can’t pass nuanced ethical judgements on human actions. Particularly in cases where a student’s intentions aren’t clear, deciding whether a student did their work with integrity is a tricky business.

A quick read of case notes from the Williams College Honor and Discipline Committee’s rulings on cheating accusations bears this out. In one situation, “[t]he student was a first year student from a high school abroad in which citation was not taught at all,” but still was punished by flunking a paper that didn’t have appropriate citations. In another, a student who had taken a test early tried to respond in a “vague yet supportive” way to a classmate who wanted to know if her notes had been useful to him in studying for the exam—and lost a letter grade on his test for his trouble. In a third, a freshman claimed that he hadn’t cited ideas taken from footnotes in the course’s text because he had thought that they were his own. The committee failed him on the assignment, but not the course, “because some felt that he genuinely was unaware that the ideas had their origin outside of his own thinking.”

Fishman points out that while students usually understand the “gross boundaries” of cheating, the specifics are much fuzzier, especially when it comes to paraphrasing and citation. “Frankly, I’ve been in many, many groups of teachers who are discussing where the borders are of plagiarism, and most of the time [they] can’t agree on where the exact boundaries are,” she told me. The definition of common knowledge—which determines what information needs attribution, and what doesn’t—is one such point of contention. “That’s a really complicated idea,” she explained. “There’s no one box of stuff that we can say, ‘Okay, this is common knowledge,’ because it varies from community to community. What’s common knowledge amongst a group of medical students, and what’s common knowledge amongst a group of engineering students is going to be different.”

Elizabeth Kiss, the president of Agnes Scott College, says that one way to achieve this is an honor code. “Fundamentally, it’s about creating a culture which focuses on membership in the community being connected with academic honesty, and then also having the critical conversations that reinforce that culture,” she says. The student handbook echoes her: “The cornerstone of the entire structure of Agnes Scott life is the Honor System,” it states.

At Agnes Scott and around a hundred other schools across the country, students sign an honor code, a promise to act with integrity on campus. In exchange for their trustworthiness, the students receive privileges that indicate the community’s trust in them—in Agnes Scott’s case, these even include self-scheduled, non-proctored exams and a leadership role in the judicial process when a student violates the code.

There’s evidence that honor codes do, in fact, deter cheating. Behavioral research shows that people who were reminded of moral expectations—by writing out or signing an honor code, or copying down the Ten Commandments—before they took a test reduced cheating. McCabe’s surveys have found that honor code schools have lower rates of cheating than other institutions by around a quarter, provided that honor code was made a central part of campus culture.

At Agnes Scott, that translates to a number of things: Students under investigation for honor code violations can request a public hearing, open to the whole community. The student-run Honor Court and the faculty, administrators and students who serve on the Judicial Review Board work to have guilty students reflect meaningfully on their behavior before dispensing a punishment. Kiss recalls a Judicial Board hearing where a student had copied an incorrect answer off of a neighboring student—despite the fact that her own calculations were correct. “She compounded it by trying to come up with more and more ridiculous and outlandish reasons for why she would have gotten that answer. So our job was to get her to have a breakthrough.”

Fishman argues that these kind of academic community-wide discussions about what constitutes integrity reverberate beyond the classroom. “What we hear from employers is that when they get students from a Bachelor’s degree, they’re really good at doing what they’re told to do, but they’re not necessarily good at looking at the situation and figuring out what needs to be done. So that points to the idea that instead of more structure and more consistency, what we need to do is provide a range of problem-based scenarios and let the students try to figure it out.”

Even McCabe, who thinks that today’s students are apathetic about school, is convinced that honor codes are universities’ last best hope. “The only reason I imagine students stop cheating is because they’re being trusted,” he says. In other words, chicken or egg?