Click photo to enlarge
PHOTO BY GARY REYES/SAN JOSE MERCURY NEWS 11/10/2002 Hoover Tower is seen through the arches of Memorial Square at Stanford University. Stanford has not been immune to the tough economic times facing valley. The University has ordered an eight percent budget cut that includes a hiring freeze and a delay of new construction projects.

Allegations of cheating at Stanford University have more than doubled in the past decade, with the largest number of violations involving computer science students.

In 10 years, the number of cases investigated by the university's Judicial Panel has climbed from 52 to 123.

Stanford, one of only 100 U.S. campuses with an "honor code," established its code in 1921 to uphold academic integrity by prohibiting plagiarism, copying work and getting outside help. Penalties for violations include denied credit for a class, a rejected thesis or a one-quarter suspension from the university. Students also pledge to report cheaters and do honest work without being policed.

"There's been a very significant increase," although the vast majority of the school's 19,000 students are honest, said Chris Griffith, chief of the Judicial Panel. More men are reported than women, and more undergraduates than graduates.

"Some of it is due to an increase in dishonesty," she said, "while some is due to an increase in reporting by faculty."

The findings came from new data presented by Griffith at a meeting of Stanford faculty at the academic senate. Although computer science students represent 6.5 percent of Stanford's student body, last year those students accounted for 23 percent of the university's honor code violators.

"My feeling is that the most important factor is the high frustration levels that typically go along with trying to get a program to run," said computer science professor Eric Roberts, who has studied the problem of academic cheating. He noted that most violations involve homework assignments rather than exams.

"The computer is an unforgiving arbiter of correctness," he said. "Imagine what would happen if every time you submitted a paper for an English course, it came back with a red circle around the first syntactic error, along with a notation saying: 'No credit — resubmit.' After a dozen attempts all meeting the same fate, the temptation to copy a paper you knew would pass might get pretty high. That situation is analogous to what happens in computing courses."

A common computer science violation occurs when students work as a team to complete an assignment, even though the rules stipulate that work must be done individually.

Also common: students obtaining someone else's code and submitting that version, after making simple edits to disguise the work. They find copies by rooting through discarded program listings taken from a recycling bin, or checking machines in public clusters to see whether previous students left solutions lying around.

"People know exactly what they're doing," Roberts said. "One student took code out of the 'recycle bin' of a laptop, changed the name of the original author and used it in six of the seven files that were submitted."

As for the problem of cheating, Stanford is by no means alone. Roberts noted that the largest cheating episode in the history of the Massachusetts Institute of Technology took place in a 1991 course titled "Introduction to Computers and Problem Solving," when 73 of 239 students were disciplined for "excessive collaboration."

Today, to reveal similarities in code, Stanford computer professors use a program called MOSS (Measure Of Software Similarity). That software is boosting the number of discovered violations.

Other violations, although fewer, were found in the departments of biology and Introduction to the Humanities. Art history had only one violation.

Universitywide, 43 percent of violations at Stanford involved "unpermitted collaboration," where students submit work that was not done independently. About 31 percent involved plagiarism, using Internet-based work that was not cited. Another 11 percent involved copying work; 5 percent, receiving outside help; 5 percent, representing others' work as their own and 5 percent, assorted violations.

The Judicial Panel's report also noted that cheating was uncommon in professional schools, such as law and medicine.

"When you're in professional school at Stanford, it is foolish to cheat. If you pass, there will be good job opportunities," said law student Eric Osborne.

"That is not as true for undergraduates in the engineering and computer science fields," said Osborne, "where in this economy, there is a lot of drive to get into grad school."

Contact Lisa M. Krieger at 408-920-5565.


HONOR CODE RULES FOR STANFORD'S COMPUTER SCIENCE CLASSES

  • You must indicate on your submission the receipt of any assistance, such as discussion of ideas and strategies "” and write the actual program code on your own.

  • You must not share program code with other students.

  • You must not look at solution sets or program code from other years.

  • You must be prepared to explain any program code you submit.

    WHY DO COMPUTER SCIENCE STUDENTS CHEAT?

    A study by Stanford computer science professor Eric Roberts concludes that:

  • Students are attracted to a computer science degree by its marketability and as a steppingstone to wealth and job security "” rather than intrinsic interest in the subject. They see stellar course work as critical to their future livelihood.

  • The material is cumulative, so students who neglect early work are completely confused by later assignments, and panic.

  • Unlike other disciplines, it is not possible to merely submit inferior work. A computer program that doesn"t work is simply rejected, which creates student anxiety.

  • It"s easy to get help from others. In terminal clusters, students work near each other, so it"s tempting to ask for help, peek over a shoulder or retrieve program listings from a recycling bin. Many students accidentally leave copies of their work on the hard disk of public machines.

  • Computer science courses often reuse past assignments, because assignments, like programs, improve with time and debugging. So students seek solution sets from past course offerings.