The international students flagged most often for plagiarism are frequently the ones working hardest to follow the rules. That is not a feel-good slogan, it is what the pattern in the data looks like once you separate two things that get merged constantly: how often plagiarism happens, and how often it gets caught.
The numbers on the first question are real. An analysis of five years of reported academic integrity violations at an American West Coast public university, by Bertram Gallant and colleagues in 2015, found international students more than twice as likely to have a reported violation as domestic students. Numbers like that circulate widely, and they shape how offices and instructors read the next ambiguous case. What circulates less is what sits underneath them.
Reported twice as often is not the same as cheating twice as much
A violation report measures three things at once: what a student did, how visible it was, and whether anyone filed the paperwork. Change any of the three and the count moves.
A follow-up study by Fass-Holmes in 2017, in the Journal of International Students, looked at the same reporting environment and added the missing denominators. Reported international students represented less than 7.5 percent of international enrollment, in a period when international enrollment itself was climbing every year, and the study's central hypothesis was that the rise in reports tracked the rise in enrollment. It also broke down what the reports were for: exam misconduct led in four of the five years, with plagiarism at more than one fifth of reports, not the majority people assume.
Australia's higher education regulator hosts a myth-busting overview making the sharper point directly: plagiarism is more easily detectable in the writing of students using English as an additional language, which creates the impression of more plagiarism rather than evidence of it. Detection is not evenly distributed, and the reporting statistics inherit that unevenness.
Why the same shortcut is visible in one student and invisible in another
Here is the mechanism, and it is worth being precise about because it is nobody's moral failing. A student writing in their first language who paraphrases lazily still does it fluently: the sentence structure shifts, the idiom changes, the seams disappear. The result is just as derivative, and it reads smoothly enough that neither an instructor nor a text-matcher snags on it.
A student writing in an additional language has a smaller set of alternative phrasings to reach for. Paraphrasing under that constraint pulls the text back toward the source's wording, because the source's wording is the version the student can be certain is correct English. The overlap that results is exactly what similarity software is built to catch, and exactly what an instructor's eye catches when polished borrowed sentences sit beside the student's own harder-won prose. The voice shift is the tell, in both directions.
So identical intent produces different detection rates. The domestic student's shortcut passes. The international student's version of the same shortcut gets a hearing date. Patchwriting, the word for that too-close paraphrase, was described by the researcher who coined it as a stage language learners pass through on the way to fluency, not a species of fraud, which makes it a strange thing to discipline hardest in the students for whom it is most developmental.
The conventions nobody teaches because everybody assumes them
Citation habits are learned, and they are learned differently in different academic systems. A student may arrive from a system where reproducing authoritative sources faithfully is how competence is demonstrated, where textbooks are treated as common knowledge that needs no pointer, or where collaboration on assignments is the norm rather than an offense. None of that is a deficiency. It is training for a different set of rules.
The rules actually in force here are narrower than they look from outside, and two of them do most of the damage. First, paraphrase needs citation even when every word is changed, which surprises students taught that citation belongs to quotation. Second, translation does not create original work: rendering a source from your first language into English still requires the citation, because the ideas rather than the words are what citation credits, and multilingual students are the ones with the most opportunities to get this one wrong innocently.
The fix for unstated rules is stated rules, and you are allowed to ask for them. One office-hours question, "when I paraphrase this source, do you want a citation on every sentence or one per paragraph," costs two minutes and marks you as careful rather than suspect. Campus writing centers exist for exactly this translation work between systems, they are free, and using one is never reportable. The students who get through cleanly are rarely the ones who guessed right about the conventions. They are the ones who asked.
What actually protects you
The honest version of this advice is boring craft plus paperwork awareness, and it works.
Build the paraphrase habit that survives the constraint: read the passage, close the source, write from memory, then check against the original. The closed-book method matters more for EAL writers than for anyone else, precisely because writing with the source open pulls hardest when your alternative phrasings are fewest. Where your version still lands close, quote it instead, with marks and a citation. A visible quotation is never patchwriting.
Scan your own draft before you submit, and read the report the way an instructor would: look at which passages match and how long the runs are, not the headline number. PlagiarismCheck.io is free, needs no account, and scans in Spanish, French, German and other languages as well as English, which matters when your notes and sources cross languages. It shows the source behind each match, so you can see exactly which sentence pulled too close and fix that sentence rather than guessing.
And know the process before you ever need it. If you are reported, you are owed the specific allegation, the procedure, and time to respond, and the first moves are the same for everyone. For international students there is one extra, unfair, real stake: an enrollment status that depends on remaining in good standing. That is a reason to answer carefully and get an adviser involved early, not a reason to accept a finding you believe is wrong.
If you teach or run an integrity process
The same asymmetry reads differently from the other side of the desk. If your reported cases skew heavily international, the honest first question is not what is wrong with these students but what your detection surface is sensitive to. A text-matcher plus an eye for disfluency is tuned to catch the EAL version of a shortcut and to miss the fluent version of the same shortcut.
The literature's suggestion is structural: treat close paraphrase in developing writers as a teaching moment with a rewrite attached, reserve the formal process for deception, and say the two paraphrase rules out loud in every course rather than assuming them. The students most likely to be flagged are, disproportionately, the ones no one ever told.
The pattern in the data is not a story about character. It is a story about visibility, and visibility can be fixed at the level of craft, one closed-book paraphrase at a time.