Where should a citation go when only one clause in a sentence needs support?

The reason this matters is that a citation is a factual assertion, and its position is part of the assertion. Writing "adoption rose sharply after the policy change, which suggests the incentive scheme worked (Okonjo, 2019)" tells a reader that Okonjo said both things. If Okonjo only reported the rise, you have attributed your interpretation to her.

The fix is mechanical. Split the sentence at the point where her evidence ends and your reasoning begins, and put the marker there: "adoption rose sharply after the policy change (Okonjo, 2019), which suggests the incentive scheme worked." Nothing else changes. The claim is now yours and the evidence is still hers.

Penders states the rule behind the split. A source should get credit for its exact contribution, not simply appear somewhere near the argument it is supposed to support (Penders, 2018, p. 2). The same paper notes that one source often has to be cited more than once, because it supports different claims at different points in an argument. So if you move a marker into the middle of a sentence and the end of that sentence still rests on the same work, cite it twice rather than assuming the first marker carries across.

Two habits make this easy. First, draft with the marker in the middle rather than the end, even when it feels ungainly, because moving it later requires rereading the source. Second, when a sentence resists splitting, the sentence is usually doing two jobs and should be two sentences.

Different fields tolerate different densities. Numbered styles in engineering and the sciences carry markers so lightly that mid-sentence placement is normal. Author-date styles in the humanities and social sciences interrupt more, which tempts writers to park every citation at the end of the sentence. Resist it. Precision costs a reader half a second and buys you the benefit of the doubt for the rest of the chapter.

If mid-sentence placement still feels fussy, the measured error rates argue for it. In a systematic review of quotation accuracy, Jergas and Baethge found total errors ranging from 6.7% to 83.0% across 28 studies, with a median of 22.5%, and major errors, meaning the reference does not support the citing statement at all, ranging from 2.2% to 55.0% across 27 studies, with a median of 11.5% (Jergas and Baethge, 2015, p. 5). Read those numbers with their limit attached. The review covered medical journal articles and excluded other reference types (Jergas and Baethge, 2015, p. 2), so they document that misattribution is common in published work, not that a fifth of your own citations are wrong.

Do I repeat the same citation in every sentence of a paragraph?

No. Cite on first mention, then signal continuing attribution with the author's name in your own prose, as in "Kessler goes on to argue" or "her data show". Repeat the full citation when you move to a different page range, when another source intervenes, or when a new paragraph continues the same source.

Repeating a parenthetical five times in six lines is the most common symptom of anxious drafting, and it reads as though you do not trust your own paragraph. The alternative is not to drop attribution but to carry it in the grammar. Verbs do this work: "Kessler finds", "she attributes", "the study reports". Each one keeps the source live in the reader's mind without another parenthesis.

Carrying attribution in the grammar puts the weight on the verb, so pick the verb against what the source actually did. Lingard warns that reporting verbs are not interchangeable, and contrasts writing that an author "argues" a point with the more accurate "suggests" or "explored" when the source was offering a tentative reflection (Lingard, 2018, Part II, p. 134). Penders makes the same point from the other side, contrasting "the authors suggest" with "the authors establish", and "this pilot study shows" with "this excellent study shows", wording that inflates or shrinks the contribution (Penders, 2018, p. 3). A correctly placed marker does not rescue a verb that overstates the finding.

The rule to keep is that a reader must never have to guess whose claim they are reading. If a paragraph runs for eight sentences on one source and your name markers thin out in the middle, add one. If you write a sentence of your own analysis inside that paragraph, make the shift explicit before you make it.

How do I attribute a whole paragraph to one source?

Name the source in the first sentence, keep every sentence in the paragraph inside that source's argument, and put the citation at the first mention. Do not open a paragraph with your own claim and attach a citation at the end. A reader cannot tell where the source stopped and you started.

Some style guides permit a single citation at the end of a paragraph that is entirely drawn from one work. It is allowed, and it is still a weaker choice, because it forces the reader to reinterpret eight sentences retroactively. Leading with the attribution costs you three words and removes the ambiguity completely.

A looser standard is in circulation, and it is worth knowing so you do not adopt it. Jergas and Baethge describe an earlier study that counted a reference as correct when it supported the general meaning of a sentence such as "for a long time psychiatrists have debated about X", even when the source did not establish every element of that wording (Jergas and Baethge, 2015, p. 11). That tolerance suits an error audit, not a thesis chapter. If your source supports only the narrower claim that psychiatrists debated X, write the narrower claim, put the citation directly after it, and find separate evidence for "for a long time".

The test to apply before moving on: cover the citation with your thumb and reread the paragraph. If a stranger could not now say which sentences are reported and which are argued, the attribution is in the wrong place.