Large-scale collusion at Harvard

The academic world in the US is currently discussing a widespread case of collusion that Harvard announced a few days ago with the publication in the Boston Globe of a letter that the Dean of Undergraduate Education, Jay Harris, sent to all students:

"I am writing to alert you to deeply disturbing allegations of academic dishonesty involving a significant number of Harvard College students, and to remind you of every student’s duty to embrace our ideals regarding, as well as the specific rules governing, academic integrity. [...]
Harvard takes academic integrity very seriously because it goes to the heart of our educational mission [...]. Academic dishonesty cannot and will not be tolerated. I join [with others] in hoping we can all use today’s news to foster a culture of honesty and integrity in everything we do as members of the Harvard community."

This is the absolutely correct step to take -- discuss the issue of academic integrity involved here instead of trying to sweep everything under the carpet. And with a motto of veritas, truth, it is important that the university make the effort to find out what happened and perhaps use this as a teaching moment on academic integrity.

What had happened? In a course "Introduction to Congress" with 279 students (according to the student newspaper, The Harvard Crimson), the professor set a take-home exam (documented here at the Boston Globe). This pedagogical concept is not clear to many, so I will elaborate a bit here. This is not a multiple choice, short answer exam like typical proctored exams, but an open book and open Internet exam. Questions are asked that must first be researched, then written up and cleanly documented. It is not testing the regurgitation of factoids, but an investigation into how good people are at research and synthesis of own material. Usually there will be a very short time frame involved, just 2 or 3 days, so that students must begin immediately and not wait until the very last minute and hope to Google something together.

In this case Prof. Matthew B. Platt stated clearly on the exam: "students may not discuss the exam with others". Now it is clear, that if it easy to cheat -- there is no proctor, only the students' own sense of integrity is at work here. They can use the library, their notes, or any other materials, but must not discuss with anyone other than themselves.

While correcting the results, according to the New York Times, the professor noted similarities in some of the answers and contacted the university authorities. They decided to investigate and looked at all of the exams. They contacted all 125 students suspected of working with others before they made the case public.

The Crimson notes that the university can impose sanctions up to suspending a student for an entire academic year, depending on the extent of the cheating. The student newspaper also documents lots of student complaints: We didn't understand the questions, they were too hard, the class was bad, the professor was bad, the last office hours before the exam were cancelled,...., the usual excuses. But even if the professor was horrible and the class time misspent and the assigned textbook on the wrong subject: the exam was given under stated conditions, and they must apply to all students equally. There is no excuse for cheating, full stop. One of the accused complains anonymously in Salon that s/he never went to section, because it was supposed to be an easy class, and now they feel that they are being made scapegoats. Well, not cheating would have been the smart thing to do here.

There is, of course, the question of why a school that is asking its students to pay more than $35,000 in tuition per year is offering classes that are this large. But that, too, does not directly bear on the question of cheating.

Another question that arises for teachers is how to determine that collusion has happened. My research group has been working on testing collusion detection programs for the past year, we hope to be able to present the results shortly. This is an entirely different question than scouring the Internet for plagiarism sources, this is a question of checking every paper submitted against every other one, looking for commonalities.