Dalhousie's statistics

In early 2006, when I nicely asked to see Dalhousie's statistics on TurnItIn.com, the ones that they used to publish on their website.

They refused.

I filed a freedom of information requestion, and 2 months and $25 later, I had the information. I publish them here because one major factor missing in the TurnItIn.com debate is some actual facts.

General observations

• As of Winter 2006, only 90 of Dalhousie's instructors were registered with TurnItIn.com. The number of instructors actively using it is uncertain, as accounts are rarely deleted.
• 2000-2500 Dal students are registered with TurnItIn - again, accounts are rarely deleted and not necessarily accurate, so this is inherently inaccurate.
• Professors do use TurnItIn.com as a learning tool, so the originality scores may not be an accurate reflection of actual work submitted by students.

Originality Scores

An 'originality score' is the percentage of the paper believed to be UNoriginal. A paper with a score of 20% is, supposedly, 20% unoriginal.

As you can see from the graph below, up to 30% of papers were more than 50% unoriginal. In all cases, a majority of papers had at least some "unoriginal" text".

Obviously, you can plagiarize without copying ANY text, and you can easily have 20-30% 'unoriginal' text without plagiarizing at all, assuming proper citation.

Accuracy

I checked with the Dalhousie Senate Discipline Committee; there are no numbers for 2005-2006 yet, but for the other three academic years, I have the number of cases that involved plagiarism. My understanding is less than one third of the cases involve TurnItIn.com, but since I don't know I'll count them all.

It appears that either Dalhousie has rampant plagiarism that it never prosecutes, or TurnItIn.com flags vastly more papers as 'unoriginal' than really are.

TurnItIn.com will admit that their score by itself means nothing, but is only intended to identify papers of interest for human review. That's fine. The problem is that since TurnItIn.com is really easy to fool and catches only the low-hanging fruit on the plagiarism tree, it is pretty easy to get a plagiarised paper through with a low plagiarism score, and trust me - plagiarizers are trying.

Raw Data

Here is the raw data I received from the university.