Approximately 20 students were found responsible for plagiarism in COS 126: General Computer Science by the University's Faculty-Student Committee on Discipline during the 2012-13 academic year.
The number represents an increase of more than twice the number of violations that occurred in any previous academic year, Fall 2013 COS 126 lead preceptor David Pritchard said.
The spike coincides with massive increases in enrollment in recent years, and is consistent with a trend of increased cases of alleged plagiarism in introductory computer science courses across the country. The rise has increased the caseload of the Committee on Discipline and raises questions about the effectiveness of the Computer Science department's practices — including recycling the same assignments year after year and running students’ responses through a plagiarism detection software — in deterring academic violations.
“It keeps the Committee busy,” Kathleen Deignan, Chair of the Committee on Discipline and Dean of Undergraduate Students said.
The Committee currently hears between 25 and 40 cases per year. A recently published recording of a hearing, which tried to determine whether a student had intentionally submitted plagiarized code, calls into question whether the Committee always complies with the high standards it requires to conclude that a student is guilty of a violation.
Pritchard pointed to a major increase in enrollment as a principal cause of the spike in the number of plagiarism cases each year. Enrollment in COS 126 increased from 543 to 648 between the 2011-12 and 2012-13 academic years and has risen steadily since the 2004-05 school year when 131 students were enrolled in the course. Total enrollment for the 2013-14 school year currently stands at 717.
Professor of Computer Science Jennifer Rexford '91, who served as acting chair of the department last semester, explained that since COS 126 is now being taken by many students who are not computer science majors, the students taking the class have a much wider range of backgrounds than they did before. This could mean that there are more students who get in over their head and thus might be more tempted to cheat, she said.
Rexford added that while in some assignments, such as English papers, one can complete the assignment using a cursory knowledge of the subject matter and simply get a lower grade, in computer science getting the program to work at all is difficult, making cheating more tempting.
Bob Dugan, a computer science professor at Stonehill College in Massachusetts who has lectured around the country on methods of reducing plagiarism in computer science courses, explained that students find computer science assignments especially frustrating because they require students to apply their knowledge in a way they may not have been taught before.
“That’s why students find [introductory computer science] really challenging,” Dugan said. “Particularly when it’s a freshman taking the course, a lot of freshmen really haven’t had a class where they’re asked to solve novel problems every week.”
There could also be more cases coming from Computer Science courses simply because the department goes to greater lengths to detect plagiarism. While writing-based courses largely avoid plagiarism-detection software like turnitin.com, many Computer Science students’ assignments are run through a program called Measure of Software Similarity, or Moss, which was developed at Stanford in 1994. Pritchard said the department now uses a new technology in addition to the Moss software.
But the department does not rely solely on anti-plagiarism software in determining whether or not to report a case to the Committee on Discipline. Pritchard explained that after using MOSS and its new software to flag suspicious pairs of code, the department examines them manually to see if flagrant copying has taken place.
“The cases we send to the Committee on Discipline are the cases where, literally, wholesale copying has taken place,” Rexford said. She added, “the cases need to be clear-cut. This is serious business. It has a significant impact on the student, obviously, and it has a significant influence on the department because it’s a huge of amount of time and effort to bring cases to the Committee on Discipline.”
She explained that this “time and effort" entails both bringing evidence to the hearing and sitting in on it as well as going through the submissions manually.
Another factor contributing to the high number of plagiarism cases from COS 126 is the course’s use of more or less the same assignments every year, Rexford said. She noted that the majority of COS 126 violations involve students consulting submitted assignments from previous semesters. Pritchard noted that students tend to keep their old computer science programming assignments in easily accessible places for many years, as opposed to chemistry or calculus problem sets which are more likely to be thrown away.
In addition, Pritchard noted that while some assignments had not previously been checked for plagiarism, now all COS 126 assignments are checked.
Deignan, the Committee on Discipline chair, explained the Committee places a high priority on familiarizing itself with the nuances of computer code, noting that while programming may not be as open-ended as a paper, there is significant opportunity for variation between students’ assignments.
“[Code] is not like a long division problem where you and I would sit and do the same problem and we would probably have exactly the same steps,” she said. “What we’ve come to understand about computer code is there are many levels of decisions and dozens and dozens of decisions that students make.”
Deignan added that the Committee has addressed the difficulty of understanding computer code by including computer code in its training and asking the computer science faculty to be extremely detailed in their submissions when they accuse students of copying.
Princeton is by no means alone in facing plagiarism problems in its introductory computer science classes. Although computer science students represented 6.5 percent of Stanford’s student body, they accounted for 23 percent of its honor code violations in 2009. Dugan noted that at Stonehill College, where he teaches, around 10 percent of the students in each of his introductory computer science class he teaches are found guilty of plagiarism. Pritchard noted that COS 126's plagiarism rate is in fact quite low compared to plagiarism rates in introductory computer science courses in Princeton's peer institutions. He said that most of those courses have plagiarism rates of four to ten percent while Princeton's last year was around three percent with 20 students found guilty of plagiarism out of 650 enrolled.
Pritchard explained that the computer science department has taken steps to try to address the spike. The department rewrote its honor code policy over the summer to make its wording more explicit, to make the description of boundary and special cases clear and to organize the policy into sections that would be easier for people to understand. Pritchard also noted that students must pass an online quiz about the honor policy at the beginning of COS 126 before taking the class.
Rexford said the department might be able to address the plagiarism spike by varying its assignments more, but added that the department finds this option undesirable because creating a well-crafted computer science assignment is extremely difficult and students enjoy COS 126, 217 and 226 precisely because the current assignments are so interesting. She also noted that COS 126 could address its problem by making tests much more of the course grade than homework assignments, but added that she doesn’t think that’s a good solution because assignments test different things than exams.
“There are some students who do much better on the assignments than they do on the written exams,” Rexford said, “and many of them are quite good computer programmers.”
When trying to address the problem of plagiarism, Rexford said, it is important to remember that the vast majority of students aren’t plagiarizing, and that the ability of honest students to learn computer science should not be compromised in the effort to curb the number of students who do cheat.
“To stop a small number of students from cheating,” Rexford said, “how much do you contort the experiences of the vast majority of students who are good?”
This article is the second in a three-part series on the University's Committee on Discipline.