📊 Turnitin Quick Reference Guide

Understanding Similarity and AI-Detection Reports


What is a Similarity Report?

The Similarity Report is a text-matching tool that highlights passages in a student's paper that match content from Turnitin's extensive database, which includes academic publications, websites, previously submitted student papers, and other online sources. Faculty can generate a Similarity Report when students submit assignments through Turnitin in Canvas. The report provides a percentage indicating how much of the submission matches existing sources and color-codes the matched text for easy identification.


Critical Point: Turnitin is a similarity detection tool, NOT a plagiarism detection tool. Plagiarism is using someone else's work without proper credit. Instructors must interpret the similarity report in context to determine if plagiarism has occurred.


What is an AI Detection Report?

The AI Detection Report uses machine learning algorithms to analyze student submissions and identify text that may have been generated by artificial intelligence tools such as ChatGPT, Claude, and other large language models. The report provides a probability score indicating the likelihood that portions of the text were AI-generated. However, it's important to note that AI detection has limitations; it can produce false positives and cannot distinguish between AI assistance and full AI generation. The AI Detection Report should be used as one data point in a broader investigation, not as definitive proof of academic misconduct.

📚 Turnitin Database Coverage
Turnitin compares submissions against:

  • 1.8 billion student papers,
  • 99.3 billion web pages, and
  • 89.4 million scholarly articles

The system compares submissions against all three sources simultaneously.


🚀 Quick Start: Accessing Reports

  1. Open Canvas Gradebook
  2. Click on assignment submission
  3. Click the colored percentage
  4. Review the detailed report
     

📈 Understanding Similarity Percentages
  • 0-15% Generally acceptable - Common phrases, proper citations
  • 16-24% May need review - Check proper citations
  • 25-49% Needs examination - Likely improperly cited
  • 50%+ High concern - Requires investigation

    Context matters more than the number! A 30% match might be perfectly fine with many properly cited quotes, or concerning an original analysis paper. Never make decisions based on percentage alone; always click through to see what is actually being matched.


✅ Legitimate Matches to Disregard (Safe to Exclude)

  • Reference lists and bibliographies
  • Properly cited direct quotes
  • Common phrases ("in conclusion")
  • Assignment instructions
  • Standard methodology text
  • Technical terminology
  • IRB (Institutional Review Board) statements

    How to Exclude: Click matched text → Select "Exclude" → Choose type → Recalculate. The original report is preserved; you're just filtering your view.


⚙️ Canvas Configuration Tips

      Configure these settings when creating the assignment, not after submissions.

  • ✓ Exclude bibliographic material (or exclude manually later)
  • ✓ Exclude quoted material (for quote-heavy assignments)
  • ✓ Enable submission indexing (stores in database)
  • ✓ Exclude previous drafts (for multiple submissions)

Impact: Proper configuration can reduce false positives by 15-30%.


🚩 Red Flags Beyond Percentages

These qualitative indicators are often more reliable than any percentage:

  • Sudden changes in writing style or quality
  • Inconsistent vocabulary with previous work
  • Perfect grammar from struggling students
  • Generic responses lacking assignment specifics
  • Missing examples from required readings
  • Inconsistent formatting/citation styles within the document

Best strategy: Compare with the student's previous work and discuss their writing process.


🤖 AI Detection Limitations


How it works: Analyzes writing patterns, sentence structure, and vocabulary against known AI characteristics.

Significant limitations:

  • False positives with overly formal writing
  • Cannot distinguish AI brainstorming vs. AI full generation
  • Newer AI models may evade detection
  • Not definitive proof of AI use
Use as ONE data point only. Gather additional evidence before taking action.


💬 Facilitating Student Discussions

  • Schedule a private meeting 
  • Share the report and ask them to explain the matches
  • Use open-ended questions: "Help me understand your writing process."
  • Focus on learning rather than punishment

For academic integrity concerns beyond technical interpretation, consult with your Dean or the appropriate office.


✋ Common Mistakes to Avoid

  • ❌ Relying solely on percentage
  • ❌ Assuming all matches = plagiarism
  • ❌ Ignoring context and assignment type
  • ❌ Skipping student communication
  • ❌ Making decisions without clicking through
  • ✅ Always review actual matches in context

These mistakes can lead to misinterpretation. Always review the full report before concluding.


🔗 Essential Resources & Support


💡 Best Practice: Use Reports as Teaching Tools

  • Show students how to self-check before submission
  • Discuss proper paraphrasing techniques
  • Review citation standards together
  • Provide assignment-specific similarity guidelines
  • Create learning opportunities from flagged content
  • Prevention through education > post-detection


Remember: Context matters more than percentage alone. Always investigate before making academic integrity decisions.