How to tell what citation style your thesis uses

This is the first question to answer before checking anything, because every judgement downstream depends on it. It is also a question with a slightly uncomfortable answer: your thesis probably does not have a style, and the useful thing to identify is not the style at all.

Start with the shape, not the name

Look at a citation in your text and put it in one of six shapes. This is decidable from the surface form alone — you do not need to know anything about the work, the field or the handbook.

AUTHOR–DATE        (Okafor, 2021) · Okafor (2021)
                   a name and a year

AUTHOR–PAGE        (Okafor 45)
                   a name and a NUMBER THAT IS NOT A YEAR
                   → MLA. Easy to misread as author–date.

NUMERIC, BRACKET   [12] · [13,14]
                   → Vancouver, IEEE and relatives

NUMERIC, PAREN     (12)
                   → same idea, different punctuation
                   → and indistinguishable from a year
                     that lost its author

NUMERIC, SUPER     …outcomes improved.¹²
                   → common in medicine

FOOTNOTE / LEGAL   [2020] UKSC 12 · supra n 4
                   → OSCOLA and legal citation

Two of these are routinely misread. Author–page looks like author–date if you glance: (Okafor 45) has a name and a number, and 45 is not a year. And numeric-paren looks like a truncated author–date: (12) is indistinguishable, on its own, from a citation whose author went missing.

Getting the shape right is most of the job, because the shape decides how a citation can be matched to a reference at all. An author–date citation carries an author and a year to match on. A numeric citation carries a position in a list and nothing else.

Why the shape is knowable and the style often is not

Here is the part that surprises people. Take the most ordinary citation in academic writing:

(Okafor, 2021)

  APA 7            ✓ exactly this
  Harvard          ✓ exactly this
  Chicago (a-d)    ✓ exactly this
  CSE (name-year)  ✓ exactly this
  APA-like house   ✓ exactly this

Nothing in those characters distinguishes them.

The five characters cannot name their style, and any tool claiming to have identified one from a single citation has guessed. What they can name, reliably, is the family — author–date — and that turns out to be the part that matters.

The shape is what the citation IS. The style is a best guess about which tradition it belongs to, and "unknown" is a legitimate answer rather than a failure.

Which is why we treat a wrong specific style as worse than an honest absence. A checker that decides your document is Harvard when it is APA will apply Harvard's rules to APA citations and report the mismatches as your errors. A checker that says "author–date, specific style unknown" can still match every citation to every reference correctly, because matching works on identity — the author and the year — and not on punctuation.

Your document almost certainly mixes them

A style is a property of a citation, not of a document. In practice a thesis assembled over four years contains citations in several conventions at once: material carried over from a master's dissertation, a chapter drafted with a co-author, sources pasted from a publisher's "cite this" button, and a reference manager applying a fifth thing to whatever it exported — the same mechanism that leaves stale identifiers in a list nobody re-reads.

Two citations sitting next to each other are routinely one APA and one Harvard. This is so common that a tool assuming one document-level style will misdescribe the ordinary case, and misdescribing it is how a Vancouver reference list ends up invisible to a check that decided the document was author–date.

How to diagnose your own document in ten minutes

One: sample the text, not the first page. Open three chapters at random and read ten citations in each. The first chapter is the most edited and the least representative — it is where the conventions were still being decided.

Two: classify by shape, using the six above. You are looking for whether one shape dominates, or whether you have two living together. Both answers are fine; you just need to know which you have.

Three: look at how the reference list is ordered. This is the single most reliable tell, and it is faster than reading citations.

NUMBERED, in the order the text first cites them
  → a numeric style. Vancouver and relatives.
  → the number IS the identity. Renumbering the
    list silently breaks every citation.

ALPHABETICAL by author surname
  → an author–date style.
  → identity is the author and year, so the order
    is a convenience and can be regenerated.

NUMBERED, but alphabetical
  → someone changed style partway, or a manager
    applied numbering to an author–date list.
  → worth resolving before anything else.

Four: check one entry against its citation. Take a reference you know well and find where it is cited. If the citation gives an author and year and the entry is alphabetical, you are author–date and consistent. If the citation gives [7] and the entry is seventh in a numbered list, you are numeric and consistent. Anything else is worth understanding before you check anything.

When the handbook and the document disagree

They frequently do, and the resolution is not what people expect.

Your handbook names the style you are required to submit in. Your document contains the style you have actually written in. For checking purposes, the document wins — a check has to read what is there, and applying the handbook's style to a document that does not follow it produces a report where everything is broken.

That is exactly the failure behind a case we wrote up: a Vancouver thesis read as author–date, and a candidate told that 90% of their citations had failed. Nothing was wrong with the thesis. One assumption at the start of the process had been wrong, and every judgement after it inherited the error.

For submission purposes the handbook wins, obviously. But those are two separate jobs and doing them in the wrong order — reformatting to the handbook before checking whether the references resolve — means reformatting entries that were pointing at nothing.

A mixed document is not a careless one

It is worth saying, because "your document mixes three conventions" reads like a criticism and is a description of nearly every thesis we have seen.

Mixing happens for structural reasons rather than careless ones: a chapter adapted from earlier work carries its original conventions, a co-authored section arrives in the co-author's habits, a publisher's "cite this" button produces its house format, and a reference manager restyles some entries and not others depending on when they were imported. None of that is a lapse in care. It is what a document assembled over four years by more than one process looks like.

And none of it is a misconduct question, which is worth separating explicitly because "integrity" and "citation" appearing in the same sentence makes people assume otherwise. Inconsistent punctuation is a formatting matter with a formatting fix. The only finding in this whole area that reaches beyond your own submission is a citation to a source that does not exist — and that has nothing to do with which style you wrote in, because a fabricated reference can be formatted immaculately in any of the six shapes above.

Style tells you how a citation is written. It tells you nothing whatever about whether the thing it points at is real.

What to do with a genuinely mixed document

Less than you would think, and the order matters.

Check first, convert second. A mixed document can be checked correctly, because matching runs on identity rather than format. Converting first means doing formatting work on entries you may end up deleting.

Convert in one direction, at the end. Once the list is right, restyling is a mechanical pass. Doing it earlier, chapter by chapter, is how a document ends up mixed in the first place.

And treat numeric conversion with respect. Moving between an author–date and a numeric style is not a formatting change — it changes what identifies a reference. In a numeric style the position in the list is the identity, so reordering the list rewrites the meaning of every citation in the text. If that is your situation, it deserves its own careful afternoon rather than a find-and-replace.

None of this is about doing it beautifully. A reference list with inconsistent punctuation and correct sources is in far better shape than a uniformly styled one where nine citations point at nothing — and only one of those two problems is visible to reading.