The abstract is the most-read part of a paper and often the most hurriedly written. Editors use it to decide whether to send a manuscript out for review. Searchers use it to decide whether to download the paper. Databases use it for indexing. Yet a remarkable number of abstracts end with “The findings and their implications are discussed” — a sentence that tells the reader nothing at all.
A good abstract must stand alone: someone who reads only the abstract should know what the study asked, how, what it found and what that means. Below is a structure, a word budget and a worked rewrite of a weak draft.
The five moves of an abstract
| Move | Answers | Suggested words (of 250) |
|---|---|---|
| 1. Background | What is the problem and why it matters | 25–40 |
| 2. Gap and aim | What is unknown; what this study does | 25–40 |
| 3. Methods | Design, sample, setting, main analysis | 40–60 |
| 4. Results | Main findings, with numbers | 60–90 |
| 5. Conclusion and implications | What it means, and for whom | 30–50 |
Note that Results should take the most words. The common failure is the reverse: three sentences on why the topic matters and one vague sentence on findings.
Results need numbers
“Results show that colleague support positively affects teacher engagement” carries little information: the reader cannot tell how strong the effect is or how certain. Compare: “Colleague support was associated with higher engagement (β = 0.31, 95% CI 0.22–0.40), more strongly than principal support (β = 0.12).” The second sentence gives magnitude and comparison, and creates a reason to read on.
For qualitative studies, replace numbers with the substance of the findings: not “three main themes were identified” but what those three themes actually say.
Structured or unstructured
Structured abstracts use headings (Background, Methods, Results, Conclusions) and are standard in biomedicine, public health and parts of education research. Unstructured abstracts are a single paragraph, typical in the social sciences, humanities and engineering. Both follow the five moves; only the format differs. Always follow the journal’s author guidelines, including the word limit, which may be 150 or 300.
A worked rewrite
Draft (shortened): “In today’s era of globalisation and digital transformation, online learning plays an increasingly important role. This study aims to explore factors influencing student satisfaction with online learning. Data were collected using a questionnaire. The results show that many factors have an influence. The study offers several implications for managers.”
Problems: the opening sentence could head a thousand papers; no gap is stated; the methods omit sample, setting and analysis; the results name no factors; the implications say nothing specific.
Rewrite: “Online courses suffer high dropout, yet most research on learner satisfaction examines content quality while overlooking interaction among learners. This study tests the relative contribution of content quality, instructor feedback and peer interaction to satisfaction and intention to continue. We surveyed 612 undergraduates who had completed at least one online module at three universities and analysed the data with structural equation modelling. Instructor feedback showed the strongest association with satisfaction (β = 0.42), followed by peer interaction (β = 0.27); content quality was no longer significant once these were included. Satisfaction explained 38% of the variance in intention to continue. Institutions designing online courses may gain more from investing in timely feedback and group activities than from producing additional materials.”
The rewrite runs to about 120 words, yet every sentence carries information the reader could not have guessed.
What does not belong in an abstract
- Citations: most journals do not allow them, except when the paper responds to a specific study.
- Undefined abbreviations, or abbreviations used only once.
- Information absent from the paper: every number and conclusion must match the main text.
- Promises: “will be discussed”, “are presented in the paper”.
- Overclaiming: “impact” or “determines” when the design supports only association.
Keywords: do not repeat the title
Search systems already index your title and abstract. The most useful keywords are terms not already there that searchers might type: synonyms, method names, geographical setting, population. If the title says “online learning”, add “e-learning”, “higher education”, the country, and “structural equation modelling”. Three to six keywords is typical.
A pre-submission checklist
- Read the abstract in isolation: would an outsider know what was asked, done and found?
- Results contain at least one number or one concrete finding.
- No opening sentence that could introduce someone else’s paper.
- Conclusion verbs match the design.
- The word limit and format match the journal’s requirements.
- Every number matches the tables in the paper.
Next step: count the words in each move of your current abstract. If Background is longer than Results, cut Background to one sentence and spend the words you save on your main finding and its number.
Câu hỏi thường gặp
What should a research abstract include?
Usually five moves: background, gap and aim, methods, main results and conclusions with implications. Results should take the largest share and include concrete numbers or findings.
Can you cite references in an abstract?
Most journals do not allow citations in the abstract. Check your target journal’s author guidelines to be sure.
What is a structured abstract?
A structured abstract uses headings such as Background, Methods, Results and Conclusions and is common in biomedical journals. An unstructured abstract is a single paragraph following the same logic.
Should I write the abstract first or last?
Write it last, once you know exactly what the paper says. Writing it first often leaves mismatches between the abstract and the main text.
How do I choose keywords for a journal article?
Pick three to six terms readers might search for that are not already in your title, such as synonyms, methods, setting and population. Avoid repeating title phrases verbatim.
