A master’s thesis literature review contains the sentence: “According to Garcia (2005), 70% of first-year students struggle to adjust.” The supervisor, curious, finds the original. Garcia (2005) is a qualitative study with 18 participants and no 70% figure anywhere. The number came from a review paper whose author had miscopied it from a third source. The student lifted the sentence from the review, kept the citation, and inherited someone else’s error intact, now under her own name.
Studies of citation accuracy across many fields keep finding that a meaningful share of citations do not support the statement they are attached to. Most are not deliberate. They come from the habit of citing through other papers without opening the source.
A citation is a claim
When you put “(Garcia, 2005)” after a sentence you tell the reader two things: the source exists, and it says this. If you have not read it, you can only vouch for the first, and sometimes not even that. Citing an unread source as if you had read it misrepresents your own work, a mild but real form of dishonesty.
Four common citation errors
| Error | Example | How to avoid it |
|---|---|---|
| Undeclared secondary citation | Copying a citation from a review as if you had read the original | Read the original or mark it as a secondary citation |
| Misrepresenting the source | “Associated with” becomes “causes”; findings about children become findings about everyone | Open the source, find the passage, compare it with your sentence |
| Citing retracted or corrected work | Using a retracted finding as valid evidence | Check each source’s status before submitting |
| Decorative citation | An uncontroversial sentence followed by six sources nobody read closely | Each source should support one specific point |
When you genuinely cannot get the original
There are legitimate reasons: an out-of-print book, an undigitised thesis, a paper in a language you do not read. Then a secondary citation is the honest route. In APA style you write “Garcia (2005, as cited in Tran, 2018) argued that…” and list only Tran (2018) in your references, because that is what you actually read. Other styles have equivalents (“cited in”, “quoted in”).
Use secondary citations sparingly. If a source underpins your main argument, make the effort to find it: ask your library about interlibrary loan, email the author, or look for a copy in their institution’s repository. Reviewers tend to notice when a paper’s entire theoretical foundation has been read second-hand.
Checking accuracy: a routine for each important source
- Open the source and find the passage that supports your point. Record the page or section in your reference manager’s notes (Zotero, EndNote and Mendeley all have a notes field).
- Check the scope: who, where and when does the source describe? Does your sentence stretch beyond that?
- Check the certainty: does the source say “suggests”, “may” or does it assert? Do not turn someone’s hypothesis into a conclusion.
- Check the numbers: right figure, right unit, right comparison group.
- Check the type: is this their own finding, or are they citing someone else? If the latter, follow the chain to its origin.
You do not need this level of care for every entry in an 80-item reference list. Apply it to the 10 to 15 sources carrying your main claims, every number you report, and any source you cite in order to criticise someone.
When you quote directly
Direct quotation demands even more precision than paraphrase. Three rules:
- Copy the words exactly, with the page number. Mark omissions with an ellipsis and any words you add for clarity with square brackets.
- Never trim a quote in a way that shifts its meaning, such as dropping “under laboratory conditions” so a finding sounds as if it applies in the real world.
- If you translate a quote yourself, add “my translation” so readers know the wording is yours.
Retracted and corrected sources
Retracted papers keep being cited for years, mostly because citing authors do not know. Before submitting:
- Several reference managers now flag retracted items in your library; switch this on if available.
- Check the article page on the publisher’s site: retracted papers usually carry a “Retracted” label or watermark and a link to the notice.
- Search retraction-tracking databases for your key sources.
Citing a retracted paper is not always wrong: you may be discussing the retraction itself or the history of a debate. In that case state its retracted status in your sentence and in the reference list.
Citations from AI assistants and summaries
Citations taken from automated summaries, AI assistants or aggregator sites are unread citations in the fullest sense. Treat them as a list of leads to look up, not as verified sources. A simple rule: if it is not in your reference manager with a page-numbered note, it does not go in the paper.
Habits that stop you ever misciting
- Take notes as you read: a one-line summary, the page number and the claim you plan to support with it.
- Tag items in your reference manager: “read in full”, “abstract only”, “cited via…”.
- Never use an abstract-only source for a sentence carrying a number or a main claim.
- Before submission, filter by “abstract only” and deal with each item.
Something to try tonight: open your draft, pick the five sentences with the strongest claims or numbers, and reopen the exact page of each source. If even one of the five does not match, you have caught a problem before your reviewers do.
Câu hỏi thường gặp
Is it acceptable to cite a source I have not read?
Not as if you had read it. If you cannot access the original, use a secondary citation that names the work you read it in, and list only that work in your references.
How do I write a secondary citation in APA style?
Write “Garcia (2005, as cited in Tran, 2018)” in the text and include only Tran (2018) in the reference list. Use this sparingly, especially for sources that support your main argument.
How can I check whether a paper I cite has been retracted?
Check the article page on the publisher’s website, use a reference manager that flags retractions, and search retraction-tracking databases for your key sources.
Can I cite a retracted paper?
Not as valid evidence. You can cite it when discussing the retraction or the history of a debate, provided you clearly state that it has been retracted.
Should I cite references suggested by an AI assistant?
Only after you have found the source, read it and confirmed it says what you claim. AI suggestions are leads to check, not verified references.
