CORVALLIS, Ore. - A new study on cheating by college students has found that diligent professors can virtually eliminate cheating on exams through a combination of efforts, including using multiple versions of the same test, hiring additional proctors and giving verbal warnings about cheating.
But in a worst-case scenario, the researchers found, about one out of three college students will cheat, given the opportunity.
Results of the research have been accepted for publication in a forthcoming edition of the Journal of Economic Education.
"There is a lot of information out in the popular press about cheating in academia, and most of it tends to place the blame on students," said Joe Kerkvliet, an associate professor of economics at Oregon State University and principal investigator in the study. "But our research has found that cheating is strongly dependent on what goes on in the classroom.
"As professors, we can do a lot of things to reduce cheating," he added.
The study, conducted at two different public universities in the United States, found that the biggest factor in whether students cheat is the instructor. Using tenured or tenure-track faculty tends to reduce cheating, Kerkvliet said. But when professors use teaching assistants, students were 31 percent more likely to cheat.
"There are a couple of possible explanations," Kerkvliet said. "One is that TAs have a lot less experience. But it also could be that they are more sympathetic to the students and reluctant to start the whole dirty process of accusing someone of cheating."
Professors can also reduce cheating by offering multiple versions of the same test, so that students don't share the questions and answers with friends in other class sections. The use of multiple versions of exams makes students 25 percent less likely to cheat.
Giving verbal warnings before each exams is also "surprisingly effective," Kerkvliet said.
"That surprised me," he admitted. "I didn't think it would be effective. But professors who state clearly before the exam that cheating will not be tolerated had success in reducing cheating." Those verbal warnings made it 13 percent less likely that students would cheat.
The other most effective method: using additional proctors, which made students 11 percent less likely to cheat.
"In the best classes, we estimate the likelihood of cheating to be .002 - two-tenths of one percent, or effectively, zero - if the class is taught by a full professor, using many versions of the exams, giving verbal warnings and employing additional proctors," Kerkvliet said.
Some professors may try things that don't work, the OSU professor pointed out..
Contrary to popular belief, multiple choice tests don't necessarily lead to any more cheating than other forms of exams, the study found. And physically separating students doesn't guarantee a reduction in cheating.
"We tend to focus on one kind of cheating - copying from your neighbor," Kerkvliet said. "But there are many forms of cheating: crib notes on students' hands, notes on the bills of their baseball caps, recordings on their headphones.
"And the study didn't begin to look at other classroom work like term papers and reports," he added. "My sense is that the Internet has really increased the potential for plagiarism. What's really sad is that the papers turned in that are really well-written are the ones that are most suspect."
The study used a randomized response technique, where students anonymously answered a series of survey questions that looked at a variety of classroom behaviors, though the researchers were actually focusing on cheating.
Kerkvliet found that, on average, 13 percent of students cheat on at least one exam in each class. The study also confirmed his earlier research that upperclassmen are much more likely to cheat than freshmen; that students who drink are more likely to cheat than those who don't; and that the more credit hours students take, the more likely they are to cheat.
However, the study also found that the higher the grade point average of the students, the less likely they are to cheat.
One question that arises during these studies, Kerkvliet said, is what constitutes cheating? In his own classroom, he sometimes allows students to bring materials like hand-written notes during an exam. "If you give well-conceived essay questions," he said, "you'll still find out what the student knows." Kerkvliet cited a Kent State University study which found that the use of crib sheets had no significant effect - positive or negative - on learning.
The other question that arises, he said, is why anyone should care about cheating.
"Cheating tarnishes the reputation of universities and allowing it to continue turns education into a somewhat sleazy enterprise," Kerkvliet said. "And I have seen from personal experience how it hurts the motivation of good students.
"And, personally, cheating offends me," he added. "I don't like to see it go on in my classroom."