Secondary citations ("as cited in")

A secondary citation is what you write when you want to cite something you have not read, because you encountered it inside something you did read.

(Okafor, 1987, as cited in Lindqvist, 2021)

You read Lindqvist. Lindqvist quoted Okafor. You are citing Okafor's finding while being honest that Lindqvist is where you met it. That honesty is the entire function of the form, and it has two consequences most guides do not explain.

What you are actually claiming

The form makes a narrower claim than a direct citation, and the difference is worth being precise about.

(Okafor, 1987)
  "I read Okafor. This is what Okafor says."
  → you vouch for the source
  → an examiner may ask you about it

(Okafor, 1987, as cited in Lindqvist, 2021)
  "I read Lindqvist. Lindqvist says this is what
   Okafor says."
  → you vouch for Lindqvist's reporting
  → you have not seen Okafor at all

The second is a weaker claim honestly made. Which is why the alternative — writing (Okafor, 1987) when you only ever read Lindqvist's summary of it — is not a formatting shortcut. It asserts that you consulted a source you did not consult, and every code of academic practice treats misrepresenting what you have read as academic misconduct. The "as cited in" is not pedantry; it is the sentence that keeps the claim true.

The secondary form exists so you can use a source you could not reach without claiming to have reached it.

Only one of the two goes in your reference list

This is the mechanical rule people get wrong, and the reason the whole thing is awkward.

Your reference list documents what you consulted. You consulted Lindqvist. So Lindqvist gets an entry, and Okafor does not — in APA explicitly, and in most other styles by the same logic. The 1987 work appears in your text and nowhere in your bibliography.

In your text
  …improved adherence in older cohorts (Okafor, 1987,
  as cited in Lindqvist, 2021)…

In your reference list
  Lindqvist, M. (2021). Adherence across the lifespan.
  Journal of Health Systems, 13(2), 88–104.

  ← and no Okafor entry. Correct.

Two works named. One entry. By design.

Why it looks exactly like the commonest defect

Now the practical problem. The single most frequent finding in any reference check is a citation with no matching entry — cited but not listed. A secondary citation deliberately produces one.

A DEFECT
  text has        (Duarte, 2020)
  list has        nothing
  meaning         you forgot to add the reference

CORRECT PRACTICE
  text has        (Okafor, 1987, as cited in Lindqvist, 2021)
  list has        Lindqvist only
  meaning         you followed the rule exactly

Both look like: "a citation with no entry."

This is why the phrase matters to anything reading your document. "As cited in" is a discourse marker, and it has to be recognised as one — both so it does not end up parsed as part of an author's name, and so the citation it introduces is understood as a secondary reference rather than a missing one. If you write the form correctly, the marker is there to be found. If you abbreviate it into something idiosyncratic, it is not.

Which produces a small, concrete piece of advice: write the standard phrase. "As cited in" and "cited in" are recognised conventions. "(via Lindqvist)", "(in Lindqvist)" and "(per Lindqvist 2021)" are not, and a reader — human or otherwise — has to guess what you meant.

What it does to your reference count

A small practical consequence people notice and find confusing: heavy use of secondary citations makes your text cite far more works than your bibliography lists.

That is arithmetically correct and can still be a signal worth reading. A thesis whose text names two hundred works while the reference list holds a hundred and twenty is telling you something about how the literature was gathered — a great deal of it came through other people's readings rather than from the sources themselves.

Text names          212 works
Reference list       128 entries
Difference            84

CASE 1  84 correctly written secondary citations
        → an honest document that read widely at
          second hand. A supervisor may push back
          on the depth; nothing is wrong.

CASE 2  84 citations whose entries were never added
        → 84 defects, and a viva waiting to happen.

The difference is entirely in whether the phrase
"as cited in" is present.

Which is the practical reason to write the standard form even when it feels clumsy. Without it the two documents are indistinguishable, to a checker and to an examiner — and only one of them is defensible.

The chain problem, and why supervisors dislike them

Secondary citations are permitted everywhere and discouraged almost everywhere, for a reason worth understanding rather than obeying.

When you cite Okafor via Lindqvist, you inherit Lindqvist's reading. If Lindqvist summarised the finding loosely, quoted it out of context, got the year wrong, or attributed to Okafor something Okafor was actually criticising, that error is now in your thesis with your name on it. Errors of this kind propagate remarkably well, because each person in the chain is citing a published source in good faith.

The classic shape is a number that drifts. A figure appears in a 1987 paper as a rough estimate with heavy caveats; a 2005 review reports it without the caveats; a 2021 paper cites the 2005 review as though it were the original finding; and by the time it reaches your literature review it is a settled fact with a thirty-year pedigree that nobody has re-checked.

A secondary citation is a promise that someone else read it carefully. It is worth asking how much you would like to bet on that.

The one nobody warns you about

There is a failure mode specific to the secondary form that no style guide mentions, and it catches careful people: the chain can be longer than you think.

You cite Okafor as cited in Lindqvist. But Lindqvist may not have read Okafor either — Lindqvist may have taken it from a 2005 review, without saying so, because the convention is inconsistently observed in published work. Your citation says the chain is two links long. It may be four.

There is no way to detect this from your own document, and no tool can tell you. The only signal available is age: the further the original is from the source that quotes it, the more likely the intervening links were themselves at second hand. A 2021 paper reporting a 1987 finding has had thirty-four years in which the claim could have passed through other hands.

Which is a reason to convert the ones that matter rather than a reason to panic about all of them. If a secondary citation is supporting a background sentence, the chain length is irrelevant. If it is supporting the claim your third chapter rests on, it is worth twenty minutes to reach the original and see for yourself.

When they are entirely legitimate

Four situations where a secondary citation is the right answer rather than a shortcut, and no supervisor should object.

The original is genuinely unobtainable. Out of print, held in one archive, or a document that was never published — a memo, an internal report, a manuscript. If nobody can reach it, citing the source that quotes it is the only honest option.

The original is in a language you do not read. Citing a translation or a summary, and saying so, is more honest than implying you worked from the original.

The point is historical. If your argument is about how a claim was received — that Lindqvist read Okafor a particular way — then Lindqvist's reading is the object of interest, and the secondary form is exactly right.

The source is peripheral. A single background sentence does not always justify tracking down a 1987 paper through interlibrary loan. Be honest about the provenance and move on.

How to remove one, when you should

If a secondary citation is doing real work in your argument, the fix is straightforward and worth the hour.

Find the original. More is reachable than people assume — institutional repositories, national libraries, and interlibrary loan reach a great deal of older material. Search the title rather than the citation, because the citation you have came through someone else's transcription and may carry their error — the same reason an identifier is worth re-resolving rather than trusting.

Read enough of it to confirm the claim. Not the whole paper. The passage the secondary source was reporting, and enough context to know it was not being qualified or criticised.

Then convert both. The citation becomes (Okafor, 1987), and Okafor gains a reference entry. This is the step people forget: converting the in-text form without adding the entry turns correct practice into the very defect the secondary form was avoiding — a citation with nothing behind it. If you are unsure whether that happened anywhere in your document, it is exactly what a forward tally is for.

And if you cannot find the original, that is a finding rather than a failure. Keep the secondary form, and treat the inability to locate the source as information about how far the claim can be pressed — a work nobody can reach is a work nobody can check, including you.