Harvard vs APA referencing: what actually differs, and why it matters

Harvard and APA look similar enough that students frequently assume they are interchangeable. They are not, and because both use author-date citations, the differences show up in exactly the places a marker checks: the reference list, the punctuation, and the handling of multiple authors.

Harvard is a family, not a style

The first thing to understand is that there is no single Harvard. It is a referencing approach that individual institutions have adapted, which is why your university publishes its own guide and why two Harvard guides can disagree about italics. APA, by contrast, is maintained by one body and is currently in its seventh edition, so there is a definitive answer to every question.

Where they diverge in practice

• Multiple authors: APA 7 uses et al. from three authors onward; many Harvard variants hold out until four
• Reference list order: APA alphabetises strictly by surname; some Harvard variants group by source type first
• Titles: APA sentence-cases article titles; several Harvard guides title-case them
• Publisher location: APA 7 dropped it entirely; most Harvard variants still expect it
• DOIs: APA 7 requires a full https DOI link; Harvard variants differ on whether to include one at all
• Page numbers: APA requires them for direct quotation; Harvard usually does too, but wording varies

Which disciplines use which

APA dominates psychology, education, nursing and much of the social sciences. Harvard variants are common across business, management and parts of the humanities. Engineering frequently uses IEEE, and law uses OSCOLA or AGLC, both of which are footnote-based and work on entirely different principles.

None of this is a matter of preference. Using the wrong style is a straightforward loss of marks, and the referencing band in most rubrics is worth more than students expect given how mechanical it is to get right.

Use a reference manager, but check it

Zotero, Mendeley and EndNote will save you hours, and all three produce occasional errors, particularly with edited collections, reports and anything without a conventional author. Treat the output as a draft. The most common failures are a missing edition number, an author field containing an organisation name formatted as a person, and a capitalisation scheme carried over from the source.

Whatever tool you use, the last check before submission is always the same: every citation in the text appears in the reference list, and every entry in the reference list appears somewhere in the text. Orphans in either direction are the easiest thing for a marker to spot.