Table of Contents
The Function of an Abstract
The abstract is simultaneously the most-read and least-carefully written component of most research papers. It is the first (and frequently only) part of your paper that a reader will encounter: search engines index it, databases display it, and busy scholars use it to decide whether to read further. Yet many students write abstracts as afterthoughts — a compressed version of the introduction rather than a self-contained distillation of the entire study.
The abstract has two distinct functions: informative and indicative. An informative abstract gives enough information about the study — its purpose, method, findings, and conclusions — that a reader can understand the study's contribution without reading the full paper. An indicative abstract describes what the paper covers without revealing the findings. Indicative abstracts are appropriate for reviews and theoretical papers; for empirical research, the informative abstract is the standard.
An introduction establishes context, identifies a gap, and explains why the study is worth conducting. An abstract reports what the study actually found. Do not write an abstract that only sets up the question without revealing the answer — this is one of the most common abstract writing errors and leaves the reader with no reason to read on.
Structured vs. Unstructured Abstracts
Journals and institutions use two formats. The choice is usually determined by the target journal's requirements or your institution's guidelines — not by personal preference.
| Format | Structure | Used in | Typical length |
|---|---|---|---|
| Structured | Explicit labelled sections (Background, Objective, Methods, Results, Conclusions) | Medical, health sciences, psychology, most APA journals | 200–300 words |
| Unstructured | Continuous prose, no section labels | Humanities, social sciences, theoretical papers, many science journals | 150–250 words |
Structured abstracts are increasingly preferred in empirical disciplines because they facilitate rapid information retrieval: a reader can scan the "Results" section without reading the whole abstract. Unstructured abstracts demand greater prose control — the same information must be conveyed with coherent flow rather than labelled demarcation.
The Four Core Elements
Regardless of format, every informative abstract must address four elements. The word allocation for each will vary by discipline and paper type.
Background / Problem statement (2–3 sentences)
What is the context and what problem does the study address? Establish why the question matters and where the gap in existing knowledge lies. Do not recapitulate the literature — one or two sentences locating the study in its field is sufficient.
Objective / Research question (1–2 sentences)
State precisely what the study set out to investigate, test, or establish. In empirical studies, this often includes the hypothesis. Be specific: not "this study examines X" but "this study tests whether X is associated with Y in Z population."
Methods (2–3 sentences)
Describe the research design, the sample or data, and the analytical approach. The reader should be able to assess the methodological credibility and relevance of the evidence. Include: design type, participants (n, key characteristics), and analysis method.
Results and conclusions (3–4 sentences)
Report the main findings — with numerical precision where available — and state the conclusion: what do the results mean and what is their implication for the field or practice? This is the most important section and should receive the most words.
The IMRAD Constraint
IMRAD (Introduction, Methods, Results and Discussion) is the structural template of most empirical research papers. The abstract mirrors this structure in miniature — typically compressing a 5,000–10,000 word paper into 200–300 words. This is an exercise in rigorous precision, not merely in brevity.
The constraint works as follows: each element of IMRAD must be represented, but not equally. The distribution of words should reflect the emphasis of the paper. For most empirical studies, the results and their interpretation deserve the largest share; the methods deserve enough space for credibility assessment; the background and objective together should occupy the smallest share.
Test your abstract by asking: could a reader understand the core of this study — what was asked, how it was investigated, what was found, and why it matters — without reading a single other sentence of the paper? If the answer is no, the abstract is incomplete. Every claim, acronym, and reference must be self-contained; do not refer to "the results above" or use undefined abbreviations.
Discipline-Specific Conventions
| Discipline | Abstract conventions |
|---|---|
| Medical / Health sciences | Structured format; specific numerical results required (effect sizes, p-values, confidence intervals); CONSORT or PRISMA compliance where applicable |
| Psychology / Social sciences | APA format; structured for empirical; continuous for review or theoretical; participant demographics required |
| Natural sciences | Unstructured but IMRAD-aligned; specific quantitative findings; brief methods |
| Humanities | Unstructured; argument-focused rather than findings-focused; may include central claim and contribution statement rather than "results" |
| Business / Management | Usually structured for journals; continuous for dissertations; practical implications section often expected |
Keyword Strategy for Discoverability
Keywords are the metadata that determine whether your paper appears in database searches. Most journals request 4–6 keywords; most institutional repositories index the abstract and keywords together. Poorly chosen keywords mean your paper is invisible to precisely the readers who most need it.
Principles of keyword selection
- Do not repeat words already in the title — database algorithms already weight titles highly; keywords are the opportunity to add additional search pathways.
- Use the vocabulary of your field's indexing systems — MeSH terms for medical research, PsycINFO thesaurus terms for psychology, JEL codes for economics. These are the terms database curators use; matching them increases retrieval.
- Include both specific and general terms — a specific term captures dedicated searches; a more general term captures adjacent researchers who might find your work relevant.
- Consider multi-word phrases — "mindfulness-based cognitive therapy" is more useful than either "mindfulness" alone or "cognitive therapy" alone, because it captures the distinctive combination.
- Anticipate the language of your target readers — if the primary audience is policy-makers rather than academics, include the policy vocabulary they use to describe the phenomenon.
Write the Abstract Last
This instruction appears obvious but is routinely ignored. The abstract must accurately represent the paper you actually wrote, not the paper you intended to write at the outset. Research changes: hypotheses are revised, unexpected findings emerge, the contribution shifts. An abstract drafted before the paper is complete typically fails to represent the study accurately and must be substantially rewritten.
Practically: draft the abstract after the final revision of the paper, when you know exactly what the study found and what you are claiming. Then read the abstract alone, without looking at the paper, and verify that it stands independently.
Annotated Abstract Example
ObjectiveThis study investigates how undergraduate humanities students in three UK universities perceive the formative value of AI-generated written feedback and identifies institutional factors that shape these perceptions.
MethodA qualitative, interpretive study using semi-structured interviews with 36 undergraduates across three institutions, analysed through reflexive thematic analysis.
ResultsFour themes were identified: (1) perceived epistemic authority of AI feedback; (2) genre-literacy mismatch in humanities assessment; (3) the moderating role of tutor transparency; and (4) institutional digital literacy provision as a boundary condition. Participants with higher AI literacy appraised feedback more critically and incorporated it more selectively.
ContributionThe findings challenge assumptions of discipline-neutral AI feedback efficacy and propose that humanities assessment cultures require domain-specific implementation frameworks. Implications for institutional AI policy and tutor professional development are discussed.
Revision Checklist
- Does the abstract stand alone — no unexplained terms, acronyms, or references to "the paper"?
- Is the research question or objective stated explicitly and specifically?
- Are the methods described with enough detail for credibility assessment?
- Are the main findings reported — with numbers where applicable — not merely hinted at?
- Is the contribution or implication of the findings stated?
- Is the abstract within the word limit specified by the journal or institution?
- Are the keywords non-redundant with the title and field-appropriate?
- Is the tense consistent? (Past tense for what was done; present for established findings; present or future for implications)
Common Abstract Errors
| Error | Effect | Fix |
|---|---|---|
| No findings reported | Reader cannot assess the study's contribution | Always state at least the direction and character of findings — never "results will be discussed" |
| Background inflated, results minimal | Emphasis inversion — the least important element dominates | Limit background to 2–3 sentences; give results 40–50% of the word count |
| Method absent | Reader cannot assess evidence quality | Include at minimum: design type, n (or corpus size), and analysis method |
| Overpromising in the conclusion | Credibility-damaging mismatch when reader reads the full paper | Match the strength of the conclusion to the strength of the evidence reported |
| Undefined abbreviations | Excludes readers outside the immediate sub-field | Spell out all abbreviations on first use, even if spelled out in the paper |
| Repetition of the title | Wastes words; provides no additional information | Begin with the context or significance; never restate the title as the first sentence |