← All posts

Citing conference papers and proceedings — APA, Harvard, Chicago

Conference papers, proceedings, posters and abstracts done right: how APA, Harvard and Chicago handle them, when they count as sources — with examples and the common mistakes.

9 min read CitingConferenceProceedingsThesis

Conference papers are one of the most underrated source categories in bachelor’s and master’s theses. In computer science they replace journal publication across whole subfields; in economics they sit alongside working papers; in the humanities they often carry the most recent material on a niche topic. Cited correctly, they can sharpen a thesis’s evidence noticeably more than books and journals alone. Cited badly, they produce bibliography entries that instantly read as shallow research. This piece shows you which formats exist, when each one holds up as a source, and how to cite them cleanly in APA 7, Harvard and Chicago.

The four formats: talk, paper, proceedings, poster

When you say “I found this at a conference,” different disciplines mean different things. Four formats show up most often, and each is cited differently:

Is this even a scholarly source?

The one question you must answer first: was the paper peer-reviewed, and what does “reviewed” actually mean at this conference? Three patterns come up:

  1. Full peer review before acceptance. Reviewers read the submitted paper, evaluate method and contribution, often double-blind. Examples: NeurIPS, ICML, ACL, CVPR, ICSE, SIGCOMM. Papers here are fully valid sources — in many CS fields more citable than later journal versions.
  2. Abstract review before acceptance. Only the abstract (300–500 words) is reviewed. The paper itself is written after the conference, without anyone having read the manuscript. Typical for many social-sciences and business conferences.
  3. Editorial screening. Conference organisers decide whether the topic fits — nothing more. Common at smaller specialist meetings, doctoral colloquia and almost all industry-facing conferences.

Search the conference call for papers or author’s guide for the phrases “peer review”, “double-blind”, “program committee”. If none appear, the paper is closer to a working-paper draft than to a journal publication — and the evidence carries less weight accordingly.

APA 7 — conference contributions

APA 7 separates three cases: a published conference paper in proceedings, an unpublished talk, and a contribution to an electronic conference volume.

In-text (published paper):

Transformer models scale nearly linearly with data volume and compute (Kaplan et al., 2020, p. 5).

Reference list — paper in proceedings with editors:

Kaplan, J., McCandlish, S., & Henighan, T. (2020). Scaling laws for neural language models. In A. Beygelzimer, Y. Dauphin, P. Liang, & J. W. Vaughan (Eds.), Advances in Neural Information Processing Systems 33 (NeurIPS 2020) (pp. 1877–1901). Curran Associates.

Reference list — paper from an online conference with DOI:

Weber, K., & Schmid, L. (2023). Automated citation verification with large language models. In Proceedings of the 45th DGfS Annual Conference (pp. 112–124). German Linguistic Society. https://doi.org/10.5555/dgfs.2023.045

In-text (unpublished talk):

(Müller, 2024, paper presentation)

Reference list — accepted but unpublished talk:

Müller, H. (2024, March 14–16). Self-efficacy in examination situations [Paper presentation]. 62nd Congress of the German Psychological Association, Vienna, Austria.

Two details APA 7 usually requires: the date range of the conference (not just the year) and the format tag in square brackets ([Paper presentation], [Poster session], [Conference session]). Without the brackets, APA conformity won’t survive a strict format check.

Harvard — shorter, equally precise

Harvard variants differ across universities. The core structure stays the same: author, year, title, volume details, publisher.

In-text:

(Kaplan et al. 2020, p. 5)

Reference list — paper in proceedings:

Kaplan, J., McCandlish, S. and Henighan, T. (2020) ‘Scaling laws for neural language models’, in Beygelzimer, A. et al. (eds) Advances in Neural Information Processing Systems 33. Curran Associates, pp. 1877–1901.

Reference list — talk without proceedings publication:

Müller, H. (2024) ‘Self-efficacy in examination situations’. Paper presented at the 62nd Congress of the German Psychological Association, Vienna, 14–16 March.

If your department demands Harvard but has no explicit rule for conference contributions, you are safe borrowing the structure of an edited-book chapter and treating the conference as the “convening event”.

Chicago — footnote and bibliography

Chicago Notes separates published and unpublished conference contributions cleanly.

Footnote (first mention, published):

¹ Jared Kaplan, Sam McCandlish, and Tom Henighan, “Scaling Laws for Neural Language Models,” in Advances in Neural Information Processing Systems 33, ed. Alina Beygelzimer et al. (Curran Associates, 2020), 1877–1901.

Footnote (shortened):

² Kaplan, McCandlish, and Henighan, “Scaling Laws,” 1885.

Bibliography:

Kaplan, Jared, Sam McCandlish, and Tom Henighan. “Scaling Laws for Neural Language Models.” In Advances in Neural Information Processing Systems 33, edited by Alina Beygelzimer, Yann Dauphin, Percy Liang, and Jennifer Wortman Vaughan, 1877–1901. Curran Associates, 2020.

Footnote (unpublished talk):

³ Hans Müller, “Self-Efficacy in Examination Situations” (paper presented at the 62nd Congress of the German Psychological Association, Vienna, March 14–16, 2024).

Chicago always requires the label “paper presented” or “lecture” in the bibliography for unpublished contributions, so it is unambiguous that the source is grey literature.

Special case: the preprint of a conference paper

Many conference papers also live on arXiv, SSRN or OSF as preprints — often weeks before the official publication. For your reader, the version you cite matters:

A classic mistake: authors quote different numbers in the conference version than in the arXiv preprint because they added experiments between drafts. Anyone who reads the preprint and cites the proceedings produces quiet miscitations — the claim in the text matches the preprint, but the citation points to the diverged proceedings version.

Special case: LNCS, LNAI, ACM and IEEE papers

Four series and two publishers cover a large share of CS-adjacent conference literature:

All three export clean BibTeX entries. Zotero and Citavi import them almost error-free — but manual review still pays off, because database BibTeX often comes with the wrong capitalisation, duplicated author lists or inconsistent page ranges.

The pre-submission checklist

The most common mistakes

Citing a preprint as proceedings. An arXiv entry is not a proceedings contribution. If your reference list reads “Kaplan et al. (2020). Advances in Neural Information Processing Systems.” but the URL points to arXiv, it is inconsistent. Examiners spot it immediately.

Dropping the volume title. For chapters from proceedings, the volume title belongs in the entry — not just the conference name. “Proceedings of NeurIPS 2020” is precise; a bare “NeurIPS 2020” is not specific enough.

Citing a talk without a date. Conferences happen on specific days. Giving the year but not the date range makes the entry read like a journal article — which it isn’t.

Treating a poster like a paper. A conference poster is rarely peer-reviewed and typically contains only a short summary without solid method details. Using a poster as a central piece of empirical evidence means you shortened the research phase — and that stands out in the viva.

Personal communication in the reference list. APA 7 explicitly keeps unpublished talks out of the reference list — they go in the running text only. Listing them anyway breaks the rule and pads the bibliography with a source no one can verify.

The real test: does the citation actually match your text?

Conference papers are especially prone to miscitations — not because the claims are weak, but because students often skim them. A NeurIPS paper is 10 dense pages of text, equations and tables; anyone quoting a number often pulls it from the abstract or the introduction without checking the relevant results section. The number matches the text — but the page number in the citation points to the abstract, and the claim in context reads differently in the paper.

That is exactly the step Acurio checks that no format checker performs: you upload your thesis together with the PDFs of the conference papers you cite, and Acurio verifies for each citation whether the claim in your text really appears in the cited source — and in the cited passage. Conference papers are one of the source types where this check delivers the most.

Conference papers are stronger sources than their reputation from school days suggests — as long as you take the publication format seriously. Before every citation, ask the three questions: How strict is the review at this conference? Is the version I’m reading the version that should be cited? And does the specific page number match the claim I’m supporting? Anyone who works through those questions mechanically ends up with reference entries that hold what they promise.

Frequently asked questions

Does a conference paper count as a peer-reviewed source?
Not automatically. Some conferences (in computer science, NeurIPS, ACL, CVPR or ICSE, for example) run strict peer review and rank above journal articles in their field. Others review only the abstract, others don't review at all. Before citing, spend two minutes on the conference's call for papers — that's where 'review' is defined.
How do I tell a conference paper apart from proceedings and a preprint?
A conference paper is a single talk or manuscript. The proceedings are the volume that collects all accepted papers of one conference — the equivalent of an edited book with many chapters. A preprint is the author's early version, usually on arXiv or SSRN, independent of whether a journal or conference review actually happened.
Can I cite an unpublished talk I listened to live?
Yes, but as personal communication. APA 7 puts an unpublished talk in the running text only — no bibliography entry, because other readers can't retrieve it. As soon as a handout, a slide deck or a public video recording exists, the talk becomes a citable source.

Continue reading

The 12 most common citation mistakes

Get the free PDF guide: examples from real bachelor's and master's theses.