Ctrl-C, Ctrl-V may no longer be an effective way to cheat.
Computer science students at Stanford University account for a disproportionately large percentage of honor code violations, according to findings by Stanford computer science professor Eric Roberts that were presented to Stanford faculty members this month.
The level of academic dishonesty is traditionally higher in computer science than in other disciplines because it is easy to copy code, but both the Stanford and Duke computer science departments are using a variety of methods to lower the prevalence of cheating.
The violations primarily occur in work completed outside the classroom and by students new to the discipline. In a 2002 paper, “Strategies for Promoting Academic Integrity in CS Courses,” Roberts notes that motivating factors to cheat include the cumulative nature of the courses, the reuse of assignments by professors and the ease of obtaining homework solutions from other students, the Internet or unemptied computer recycle bins.
Roberts also points out the unforgiving nature of the computer, which will not run a program containing errors, as a factor that can lead to high levels of frustration for students.
Although the nature of computer programs makes cheating as easy as copy and pasting, it also makes detection easier.
“[Copied code] is not obvious to someone in a non-technical field, but it’s completely obvious to someone in the field,” Roberts said.
But both schools’ departments rely on more than their expertise to identify potential cheating—they also use the program Measure of Software Similarity, to compare student assignments to the work of current and past students to determine similarities in submissions. The program yields a percentage that reveals the level of similarity between pieces of work.
Professor Owen Astrachan, co-director of undergraduate studies for Duke computer science, tells students he reserves the right to run MOSS.
“There’s an atmosphere you are [impressing] with your students by saying I’m catching cheaters, I’m catching cheaters, I’m catching cheaters,” he said.
Sophomore Caitlin Ryan believes the threat of being caught is all the incentive most students need.
“They have instilled enough fear to encourage us to not cheat,” Ryan said.
Duke’s department does not aim to rely solely on threats, but also implements measures to prevent students from reaching the point where they feel cheating is the only option. These measures include providing options for which assignments to complete and requiring students in introductory courses to submit a “readme” file in conjunction with their work that details their time spent and anyone they talked to about it. This information would provide justification for some significant similarity between students, Astrachan said.
Junior Kosta Kostadinov, a computer science major, noted that academic integrity reminders discourage cheating.
“I think if you are reminded about the honor code, people cheat significantly less,” Kostadinov said. “[The readme files] trigger a sense of that.”
These discussions have led Astrachan and the department to reconsider their policy of only small point deductions for a student failing to submit a readme file. He noted that the department should implement a more stringent policy of refusing to grade a student’s assignment if a readme file is not turned in alongside the work.
He noted that one of the biggest challenges for professors is the fine line between cheating and collaborating.
“Collaboration is good, copying is not,” Astrachan said. “But there is a slope between talking and taking.”
Given the cloudy boundary between collaborating and cheating, both Astrachan and Roberts noted the situation requires careful consideration.
Ryan, who is not a computer science major, said she is taking a computer science course to learn about the subject.
“I’m in the class to learn how to program,” she said. “If I copy, then I’m not learning.”
Get The Chronicle straight to your inbox
Signup for our weekly newsletter. Cancel at any time.