How to write an abstract for a research paper
On this page
The abstract is the most-read part of any paper. Database users, editors, reviewers and conference committees often decide from it alone whether to read further, so it needs to stand on its own: a complete, accurate miniature of the whole study in 150 to 300 words.
What an abstract has to do
An abstract answers five questions in order, often called the five moves:
- Context: what is the problem and why does it matter? (one or two sentences)
- Gap or aim: what is unknown, and what did this study set out to do?
- Method: what design, data or sample, and main analysis?
- Results: what did you find? The most important findings, with key numbers where appropriate.
- Conclusion and implication: what does it mean, and for whom?
The most common weakness is spending half the abstract on background and leaving one vague sentence for results. As a rule of thumb, results and conclusions should take at least 40% of the words.
Word budget for a 250-word abstract
- Context and gap: 40–50 words
- Aim: 20–30 words
- Methods: 50–60 words
- Results: 70–90 words
- Conclusion and implications: 30–50 words
Always check the specific limit; journals commonly set 150, 200, 250 or 300 words, and conference systems may cut off anything beyond it.
Structured vs unstructured abstracts
Many medical, health and some social science journals require structured abstracts with labelled headings such as Background, Objective, Methods, Results and Conclusions. Others want a single unstructured paragraph. The content is the same five moves; only the presentation changes. If you write a structured version first, converting it to a paragraph is easy.
Write it last, but plan it first
Write the final abstract after the paper is complete, because results and emphasis change during drafting. But writing a rough "working abstract" before you start is a powerful planning tool: if you cannot summarize your planned study in five sentences, the plan is not clear yet. Revisit it as you work on the paper structure.
Sentence-level guidance
- Be specific. "We surveyed 412 software and consulting employees at two time points" beats "a survey was conducted".
- Report findings, not topics. "Boundary integration partially mediated the association" beats "the role of boundary management is discussed".
- Use tense consistently. Past tense for what you did and found; present tense for what the findings mean.
- Avoid citations, abbreviations and undefined jargon unless your field expects them.
- No new information. Everything in the abstract must appear in the paper.
- Match the paper. Numbers, sample sizes and claims must be identical to those in the body.
Templates you can adapt
Empirical (quantitative):
Qualitative:
Review or conceptual paper:
When your paper is still a draft without final data, keep the results sentence as an explicit placeholder, for example [insert main effect and statistic], rather than writing a plausible-sounding finding. It is easy to forget to replace invented text later.
Keywords
Most journals ask for four to six keywords. Choose terms a researcher would type into a database, include synonyms not already in the title, and consider method or population terms. If the journal uses a controlled vocabulary such as MeSH, use it.
Common mistakes
- Too much background, too little result.
- Vague statements: "results are presented and discussed".
- Claims that go beyond the data, especially causal language for correlational designs.
- Different numbers in the abstract and the paper.
- Exceeding the word limit.
- Copying sentences from the introduction instead of summarizing the whole paper.
A worked example: from weak to strong
Here is a weak opening many drafts start with:
It has no gap, no method, no finding and no implication. A stronger version of the same abstract reads:
Every sentence now does a job, and the bracketed placeholders show exactly what still has to be filled in with real results.
Final check
Read your abstract without the paper. Could a reader in a neighbouring field tell what you studied, how, what you found and why it matters? Could they decide whether to read on? If yes, it works. If a key element is missing, compare it with your discussion section, which usually contains the clearest statement of what your findings mean.