Published online in Virtual University Journal (, February 1999.
Author’s address:
Author’s web site:

Cheating in the Virtual University

Daniel Eisenberg
Regents College

        In a utopian society, educational institutions would do nothing but educate. In Socrates' academy, there were neither examinations nor diplomas, and even today there is much non-traditional learning that takes place with minimal or no credentialing. Although the trend is very slow, testing and credentialing are slowly being taken away from universities. One factor leading to the founding of Western Governors University, where certification of competence is quite separate from instruction, is the limited validity of present university credentials.

        Nevertheless, universities are credentialing institutions, and will remain so for some time to come. Colleges and universities not only teach, we certify mastery of both of courses and degree programs. In order to carry out this function, the school must know the identity of the person whose knowledge or skill is being certified.

        In a perfect world, students would never cheat. In our quite imperfect one, given the pressures on them and the value of credentials, a certain percentage of students will cheat if they can. Traditional universities, of course, have an ongoing problem with cheating. Like some diseases, it can never be eliminated, only controlled. Colleges and universities struggle with purchased or ghostwritten term papers and even dissertations, identity fraud on exams, sharing of exam questions, and many other types of misconduct.

        Some wishful thinking, or an inner faith in the purity of the Internet, have led some to assume that distance education would be free of this concern. Faculty accustomed to small classes, who have known each of their students personally, have been surprised to find that cheating is a new concern when moving to a distance environment. In fact, it is a larger problem for the virtual university than it is for the traditional one. In the familiar classroom setting, for all but the largest classes, the instructor learns the face of each student. This aid makes it easier to learn each students' writing style. In an exam, the instructor or proctor's eyes and ears, and ability to move about the classroom, are effective controls on cheating.

        The only form of distance education that even approximates this control is interactive video. Both audio and video offer the instructor far less “information” than a face-to-face setting, and it takes two cameras to be able to observe even a small room effectively. Nevertheless, cheating is more difficult in such an environment.

        An Internet-delivered course is another thing altogether. Some instructors have an exam fantasy. At a pre-announced time they simply post the questions on a web page, or in an email message, require that the answers to be received in an hour or two, and have identity ascertained by a password. This fantasy of course ignores such practical considerations as transmission delays in either direction. (In 1995 my wife was working in a building across the street from me. When lunchtime came, I would sometimes send her an email asking if she was ready to go to lunch. Receiving no quick answer, I could walk down two flights of stairs, across the street, and up three flights to her office, and sometimes arrive there before my message.)

        The real problem, however, is the reliance on a password. The password system serves well to exclude unauthorized users from access to resources. The possessor of the password has an incentive not to divulge the means of accessing a bank account. With examinations, the reverse is the case: the password possessor has an incentive to divulge it.

        There is no practical way, nor will there be in the near future, to verify identity at a distance through any automated system. An exam administered online, for example, is inherently compromisable. Here are some ways in which cheating is possible:

Student gives password to a conspirator, who logs in and takes the exam for the student.

Student logs in, then a conspirator takes the exam. No matter what sort of identification is used—fingerprints, retinal scans, you name it—the student could validate his/her identity, and turn the keyboard over to someone else.

Student writes the exam, but a conspirator tells student what to write.

Student consults written or printed materials to find answers to exam questions.

        Mechanical remedies add as many problems as they remove. Let's suppose, for example, that we employ a video camera to watch the student taking the exam. We could thus see that the testee has, for example, no books or papers near the computer. Yet it would not detect conversations with a person off camera: for that one would need to add a sensitive microphone. Our microphone, however, would not detect a radio receiver worn in the ear. Such devices have actually been used in testing situations in the U.S.

        A person watching the testee via video could not even determine whether the keyboard used by the testee is in fact connected to the computer. A conspirator could be outside the room with another keyboard, a long cable, and binoculars to read the screen. For that matter, the student could be sitting at a dummy computer, not connected to anything, whereas the connected computer used by the conspirator is elsewhere.

        Students' resourcefulness and inventiveness in cheating are, in some cases, truly impressive. Also, a TV camera is not something that students are likely to have at home. The equipment would have to be provided at an examining location. There would have to be a human to secure the equipment, admit students, and keep the complex installation running. In addition, it takes a human to monitor the video for inappropriate activity during the test.

        To prevent cheating, or to keep it to the level that exists in the traditional university, the only way is a human supervisor in a face-to-face situation. The human has better eyes than any video, better ears than any microphone, and can interpret activity and take action with a flexibility that no automated system can approach. In other words, not just the cheapest way, but the only way to verify a testee's identity is the human test supervisor. The human can compare a student's face with the picture on an identification card. The human can also observe the testee from more angles and distances than any video camera. The mobile human can go out into the hall to be sure no one is supplying answers from outside, can identify hidden earphones, even listen for patterns of pencil taps.

        My point is that the fantasy of exams administered online, with the student logged in from home or the public library, is just that, a fantasy. There is no such thing as a verifiable on-line identity. The Internet, which began within the U.S. military, was set up on a trust basis. In some contexts, this is one of its advantages. Changing one's identity is easy on the Internet.

        Given current technology, a human is the most cost-effective means to verify identity. Passwords, fingerprints, and other more esoteric techniques being developed do establish an identity of sorts: the identity needed to access e-mail or to withdraw money from a bank teller machine. What they cannot tell is whether the person withdrawing the money has a gun pointed at her.

        This is, then, a theoretical limitation on the virtual university. Virtual testing is impossible, without so much technology that it costs more than the human monitor (and requires a human to safeguard and support). Testing is of necessity a time for face-to-face human contact. We need to build this into our systems: testing centers are part of the credentialing process. Such centers, at which the student must appear for identification and supervision, will unavoidably appear, offer the opportunity for other face-to-face activities as well.