• Turnitin calculates a “similarity score” based on the amount of verbatim and similar text matches to content found on websites, in professional publications, within UVM’s institutional student paper repository, and within a global student paper repository (depending on the settings). 
• The similarity score is not a simple indicator; it needs to be interpreted for each submission and requires experimentation on the part of the instructor, especially while learning how the settings impact the score. Turnitin notes that scores should not be taken at face value to “prove” plagiarism.
• Similarity score is dependent on instructor choices regarding the settings for the report to in/exclude quotes, bibliographies, and specific sources made when setting up the assignment and, on the fly. These are visible to both instructors and students, who can see the reports after they are generated.

Turnitin has both critics and admirers. We encourage instructors to consider Turnitin not as a simple option within Brightspace but as a tool that is deeply connected to assumptions about how to support and scaffold student learning. Transparency about why and how Turnitin is used is essential for instructors to work toward developing trusting relationships with students.

• It adds a layer of surveillance to writing assessment that some instructors find antithetical to their teaching philosophy by creating an atmosphere based on fear, not trust. 
• Other critiques of Turnitin note that the company profits from unpaid contributions of students through the addition of their work to comparison repositories. 
• Students are unable to opt-out of the Terms of Use of Turnitin if the instructor requires it. 
• Instructors choose whether to store student papers in an institutional repository or global paper repository, but students cannot make choices about what happens with their own work.
• It can be turned on for just particular assignments, specific students, or the whole class.
• From another viewpoint, Turnitin may be a useful tool for creating community accountability and reinforcing norms around independent work, particularly in larger classes with repeated assignments.

The UVM Knowledge Base contains detailed information about how to activate Turnitin and its settings.  In general, Turnitin is able to:

• display a similarity report for students upon assignment submission; instructors can choose not to have this available to students
• identify similar or plagiarized language due to its expansive repositories 
• identify the original source of similar or copied language 
• include or exclude quotes and bibliography from the report
• exclude specific sources that are identified by the similarity report 
• exclude common or short phrases under a certain number of words or similarity percentage to another source
• flag attempts to bypass similarity checkers including the use of white quotation marks, “invisible” letters, and unusual characters

Turnitin is unable to: 

• exclude footnotes, even when the “exclude quotes” and “exclude bibliography” options are checked
• distinguish between a draft and a final submission if the draft was saved to the repository, thus inflating the similarity score of the final
• rerun reports against other assignments submitted once the due date has passed (although if a later assignment is very similar to an earlier submitted assignment, the later assignment’s report will note that.)
• provide alternative options for students who do not want to submit their work to Turnitin
• positively identify plagiarism from the similarity score alone; instructors must look closely at the similarity report
• detect copied or closely paraphrased language that has been translated from one language to another

Using the student impersonator function in Brightspace, we submitted assignments using original text, copied and pasted text, and paraphrased text, revealing Turnitin’s capabilities and limitations. We also tested tactics commonly used to try to trick plagiarism checkers.We adjusted the settings for each submission and observed the impact these adjustments had on similarity scores. A summary of our observations and detailed tables are available upon request.