Plagiarism detection software is causing great concern among students

A new survey shows that about half of high school and college students from seven European countries are concerned about the use of plagiarism detection software in education. Their concerns lead to counterproductive behavior and misdirected learning, according to the researchers behind the study.

The research, led by the University of Copenhagen’s Department of Food and Resource Economics, aimed to develop a better empirical understanding of concerns about text matching software (TMS) among students in Switzerland, Denmark, Hungary, Ireland, Lithuania, Portugal and Slovenia .

The research has been published in the International Journal of Educational Integrity.

When asked about their views on their institutions’ use of plagiarism software, 47% of high school students and 55% of undergraduate survey participants expressed concerns about scrutiny. Their concerns extend beyond the fear of being caught cheating.

According to postdoctoral researcher Mads Goddiksen, lead author of the study, these concerns largely stem from uncertainty about how the software is used and what counts as cheating.

“It is paradoxical that a technology intended to ensure academic integrity causes unnecessary concerns among students. Our research clearly shows that uncertainty about how the software works, how it is used and what constitutes plagiarism leads to concerns and counterproductive writing practices,” says the researcher. .

Write to outsmart the software, not to learn

In interviews, students described how they avoid reusing sources from previous assignments or unnecessarily rewriting texts to ‘outsmart’ the software – a development that Mads Goddiksen warns against.

“The biggest problem is not the concern itself, but that students lose focus on good and ethical writing. Instead, it is about preventing the software from flagging something as problematic. This affects the quality of both their assignments and the overall education,” he explains. .

Goddiksen emphasizes that plagiarism detection software cannot independently determine whether plagiarism has occurred; it only emphasizes text overlaps. Understanding this distinction is crucial.

“There is nothing inherently wrong with paraphrasing or reproducing content from other sources in an assignment. This is largely what academic writing entails, as long as it is done in a transparent manner. However, today such practices can show up in plagiarism checks because the software matches in the content. This makes students nervous and makes the software ineffective when used alone,” Goddiksen explains.

However, many students in the study wrongly believe that only the software determines what counts as plagiarism, even though this is not the case. This misunderstanding causes students to adapt their writing behavior to the perceived control. For example, a Danish student said, “I’m really afraid to gamble with these things. That’s why I make a lot of footnotes, about 80-90 in a twelve-page assignment. I haven’t had any criticism yet, but I don’t do that.” I don’t think this is what you should actually be doing.”

Clearer guidelines and consistent practice

The researchers emphasize that the solution does not lie in abolishing plagiarism detection software, but in using it correctly.

“The technology can be a useful tool in identifying potential problems, but requires institutions to communicate about it clearly and ensure that teachers and students understand the limitations of the software,” Goddiksen explains.

The study recommends more instruction on academic writing and good citation practices for students. Educational institutions and teachers also have a responsibility to make clear what they consider plagiarism.

“We propose a combination of instruction and clearer procedures for using the software. Teachers should step in and explain where the plagiarism boundaries are for specific assignments and how they use the software. This will not only allay concerns, but also serve to ensure that technology supports rather than hinders learning,” said Associate Professor Mikkel Willum Johansen from the Department of Science Education, co-author of the study.

Today, numerous companies offer tools that can not only identify textual overlaps, but also determine whether texts may have been generated by artificial intelligence (e.g. ChatGPT). This new form of monitoring also brings challenges for educational institutions.

“The problem is that these systems can only suggest whether something appears to be AI-generated, but they cannot say for sure. Unlike plagiarism detection software, they lack an original text for comparison. For this reason, AI detection systems are very unreliable. This underlines the importance of institutions having clear procedures and consensus on how to use the technology, so that we prevent students from being unfairly punished,” concludes Mikkel Willum Johansen.

The research is based on 3,424 survey responses and 36 interviews conducted in Switzerland, Denmark, Hungary, Ireland, Lithuania, Portugal and Slovenia.