Google Alerts for detection of contract cheating

On this page I briefly describe how to set up a Google Alert for efficient detection of improperly shared questions. It does not seem to be a novel idea, and I make no claim whatsoever to having originated it! I first used this setup in semester 2 of 2020 (that is, from August-November) as a tool to detect cheating for assessments in my first- and second-year mathematics classes. Although my background is in maths, the technique described here can work for any discipline.

This tip is mentioned in the following publication: Nathan Clisby & Antony Edwards (2021) Individualized summative assessments as used during COVID-19, International Journal of Mathematical Education in Science and Technology, DOI: 10.1080/0020739X.2021.1982040, pdf version.

And, the material on this webpage was prepared as a pdf presentation by Ant Edwards and me for a roundtable discussion at the Australian Academic Integrity Network Forum 2021.

I’ll add the caveat that there is clearly much more to academic integrity than this! Ideally one would work on fostering positive behaviours, and not just punishing negative behaviours. My goal in writing this up is just to add an extra tool to your arsenal.

If you find it useful, could you please let me know? My contact details are: Nathan Clisby, nclisby@swin.edu.au.

What problem does it solve?

• Detect improper sharing of test and assignment questions to Chegg and other websites where contract cheating can potentially occur.
• Minimise cognitive burden: set and forget.
• Google crawls the entire web frequently and so alerts are typically received within hours of a question appearing.
• Quick detection acts as disincentive to cheat, and reduces exposure of questions and solutions in the wild.

How does it work?

• Prior to release of the assessment, set up a Google Alert that searches for text fragments from the assessment.
• If any hits are detected you will receive an email alert from Google with the URL.

Details

• Google Alerts URL: https://google.com/alerts (you will need a Google account).
• Search text: 32 words or less.
• Options: As-it-happens, All results.
• Chain together with OR operator “rare” text snippets in double quotes.
• If assessment is provided to students as a pdf file then it is best to extract text snippets from the pdf rather than the source text (say LaTeX or Word).
• E.g.: "Sketch the graph of an even function f (x) that is defined for" OR "Sketch the graph of an odd function g(t) that is defined for" OR "function. Determine if the function"
• For individualised assessments ensure that snippets are common to all versions.
• 6 words of mathematical text, even with common words, will result in few hits.
• See screenshots at the bottom of this page for an example of the Google Alert setup page, and for an example of the messages that are sent to your account when there is a successful hit.

Summary

• We have received various alerts from Chegg that have detected cheating. (False positives came from websites including: Course Hero, Quizlet, Brainly, Numerade, Byjus, Gauthmath.)
• Chegg is indexed by Google. This is Chegg’s choice, presumably because it drives traffic to their site.
• Chegg will respond to takedown requests submitted via their “Honor Code” process.
• Chegg will remove questions within a couple of days of a request, and provide account information of question submitter, as well as details of users who have accessed the question.
• Typically this account information will not identify the student, as users can sign up with any email address.
• Ideally: generate unique questions for each student, and that way identification can occur purely based on question that has been shared.

An incomplete list of references on automatic detection of contract cheating

• Thomas Lancaster & Robert Clarke (2016) Contract Cheating: The Outsourcing of Assessed Student Work, in Bretag T. (eds) Handbook of Academic Integrity. Springer, Singapore. DOI: 10.1007/978-981-287-098-8_17
• Nathan Clisby & Antony Edwards (2021) Individualized summative assessments as used during COVID-19, International Journal of Mathematical Education in Science and Technology, DOI: 10.1080/0020739X.2021.1982040, pdf version.

(If there is any literature I should add here please let me know.)

Screenshots of Google Alert setup and results