Plagiarism by numbers
Posted December 1, 2009on:
One of the things I did prior to grading essays was to submit all 130 of them to SafeAssign (SA). This online tool is embedded within my university’s learning management system (LMS) and it compares each essay with other works in a large database.
Both my trainee teachers and I get to use SA to remove possible instances of plagiarism. My trainees get to submit a draft of their assignments which they can then edit after SA provides them with a report; I get to submit their final versions.
But there are at least two things that I do not like about its use. Don’t get me wrong. I like SA as it is a useful tool. It is the implementation of SA that bugs me.
The instructions in the LMS used to refer to the “plagiarism score” that users see in a post-submission report. It has thankfully been corrected to “matching score” because that is all it is: The report that my trainees and I can view indicates how much of a person’s work matches someone else’s. It is plagiarism only if that person opts not to take corrective action, e.g., properly citing another person’s work. But the common lingo used by many students and instructors alike is “plagiarism score”. This assumes guilt before action!
My second bugbear is my institution’s guidelines on what score indicates plagiarism and what does not. While I do not have the liberty to say that that percentage is, I think that any number is ridiculous. For the sake of argument, let us say that the number is 33%. This means that up to a third of a person’s essay can match someone else’s and that person is safe.
I think that this number does a disservice because it 1) provides a “safe zone” and 2) hides actual plagiarism. If a person has a score of 30%, then s/he is not obliged to edit his/her work further to bring the score down. This perpetuates a wrong value system.
I’d also argue that someone with a score of just 15% could have plagiarised work while someone with a score of 40% might not. For example, the first person (with the 15% matching score) could have copied an entire paragraph, but s/he is within the safe zone. The second person (with the 40% matching score) could have lots of references which were properly cited. These cited works invariably bring up the matching score, but the second person is not guilty of plagiarism. This is the numbers game. Looking only at the final matching scores assumes the first person is innocent and the second person guilty, when in reality, the opposite is true.
Now any reasonable grader will realise this and examine the SA reports to determine if a 33% matching score is or is not indicative of plagiarism. But will they do this 130 times like I did? It really does not take that much time to eyeball the reports because the matching parts are highlighted and you can make judgments immediately. Out of 130 scripts, I have detected three clear cases of plagiarism. All three were well below the limit set by my institution.
Am I terribly concerned? Not particularly. Philosophically speaking, I think practices like mashups and creative commons licences will gain ground over concepts like intellectual property and copyright.
But practically speaking, I am more concerned about the trainee teacher behaviour that stemmed from attitude. Did the three who plagiarised bother to use SA to check their drafts? If they did that and read the reports, why did they opt not to edit their work? They may be good teachers by various measures, but if they do not have the industry or integrity to check their work, they are not good teachers in my book.
On further reflection, I realise how this reinforces what I thought earlier on issues surrounding grading. The assignment and the grading itself is not likely to reveal much to me about my trainees. But their actions around it already have.