How to write an abstract

What an abstract has to do

An abstract is a complete, miniature version of your study. It is not an introduction to it. A common pattern in first drafts is an abstract that sets the scene and then stops: two sentences of background, a sentence saying “the results are discussed”, and nothing a reader can use.

A reader should be able to cite your main finding from the abstract alone. If your abstract does not contain your main result, stated with a number where you have one, it is missing its most important element.

Abstracts are also how search works. Scopus, Web of Science, Google Scholar and Shodhganga all index the title and abstract. The words a searcher would type to find a study like yours should appear there, used naturally.

The four moves every abstract makes

Whatever the field and whether or not the journal uses headings, a good abstract makes four moves, roughly in this proportion:

1. Context and gap (about 20%). One or two sentences: what problem this is and what is not yet known. Skip general statements about how important the field is.
2. What you did (about 25%). Design, sample or material, setting, and the main method of analysis. Name the numbers: 412 respondents, 64 wells, three case companies, a dataset of 1.2 million transactions.
3. What you found (about 35%). The main results, with the key figures. This is the part weak abstracts leave out, and it deserves the most space.
4. What it means (about 20%). The implication for theory, practice or policy, stated no more strongly than your data support.

A simple exercise helps when you are stuck. Write one sentence for each move on a blank page, without worrying about style. Four sentences, one per move. Then expand each one to the word limit. It is much easier to expand a complete skeleton than to cut a long abstract that is missing its results.

Tense and voice

Describe your methods and results in the past tense (“we surveyed”, “adoption was higher”). Use present tense for the conclusion and for general statements (“these findings suggest”). “We” is accepted by most journals. Some universities prefer the passive in thesis abstracts; follow your department’s practice.

Structured, unstructured and thesis abstracts

The limits above are typical ranges, not rules. The author guidelines of your target journal, or your university’s thesis regulations, give the actual limit. Journals enforce it strictly, often through the submission system itself, which will not accept an abstract over the limit.

Structured abstracts

A structured abstract uses the journal’s headings, and you must use exactly those headings in exactly that order. Medical journals commonly use Background, Methods, Results and Conclusions. Emerald journals in management use Purpose, Design/methodology/approach, Findings and Originality/value, with optional headings for limitations and implications. The four moves are the same; the headings just make them visible.

Highlights and graphical abstracts

Many Elsevier journals also ask for highlights: three to five bullet points of no more than 85 characters each, including spaces. Some ask for a graphical abstract, a single image that summarises the study. These are separate from the abstract, and each journal’s guide for authors states whether they are required or optional.

Thesis abstracts

A thesis abstract covers the whole study rather than one finding, so it is longer and usually summarises the objectives, the methods across chapters, the main findings for each objective, and the contribution. Universities set their own limit. Some ask for one page, some for 500 words, and some for two pages. Where your university has a published thesis format, our thesis structure guide links to the university format guides we maintain. Keep the thesis abstract consistent with your synopsis; examiners often read both.

Example: a weak abstract and a stronger rewrite

The example below is invented for illustration; the numbers do not come from a real study. The weak version is typical of abstracts that editors return. The stronger version describes the same kind of study.

Look at what changed:

• The opening line names a specific problem in a specific place, not a statement about the world.
• “An attempt has been made to study” becomes what was actually done, with sample size, seasons and standard.
• “Various parameters were analysed” becomes a list of the parameters.
• “The results are discussed in detail” becomes the actual result, with numbers.
• “Useful for policy makers” becomes a specific recommendation the data support.

The weak version is 48 words. The stronger one is 104. Length was not the problem; content was.

What does a PhD thesis abstract look like? A worked example

The paper abstract above reports one finding. A thesis abstract has to carry four or five years of work, so its shape is different: it states the objectives, gives one sentence of method per strand of the study, answers each objective in turn, and ends with the contribution. Scholars often squeeze all of that into a paper-length paragraph, and the examiner can’t tell which finding answers which objective.

The study below is invented. The place is real; the numbers are not.

A few things to notice. Each objective gets its own finding, in the same order, so an examiner can tick them off. The methods sentence names both strands and their sizes but leaves the detail to Chapter 3. There is one clearly stated limitation, because a thesis abstract that claims nothing went wrong invites questions at the viva. It runs to about 310 words; your university’s regulations set the real limit, so check yours before you trim or expand.

Common mistakes in abstracts

• No results. Already covered, and still the most common fault.
• Too much background. Half the abstract spent on why the field matters.
• Copying sentences from the introduction. The abstract should be written fresh, after the paper is finished. Sentences copied from elsewhere in the paper also show up as matches against your own preprint or thesis.
• Claims the paper does not support. A regression on survey data does not “prove” anything. Editors read the abstract to judge whether the author overclaims.
• Abbreviations and citations. Spell terms out. Most journals do not allow references in the abstract.
• Different numbers in the abstract and the results. This happens when the analysis is rerun after the abstract was written. Check every figure at the end.
• Phrases that fill space. “In this paper”, “it was found that”, “the present study is an attempt to”. Cut them and use the words for results.

A clear abstract helps a paper through the editor’s first read; the desk rejection guide covers the other factors editors look at.

How to choose keywords and check an abstract

Most journals ask for four to six keywords. Choose terms that a researcher would search for and that are not already in your title, since the title is indexed anyway (see how to write a research title). Include the method or context if it is distinctive (“panel data”, “Kerala”, “finite element analysis”). If the journal uses a controlled vocabulary such as MeSH in medicine, or asks for JEL codes in economics, use it.

Before you submit, go through this list:

• A reader who sees only the abstract knows the problem, what you did, the main result with a number, and what it means.
• Every number matches the number in the body of the paper or thesis.
• Nothing appears in the abstract that is not in the main text.
• No citations, no undefined abbreviations, no references to figures or tables.
• The word count is within the journal’s or university’s limit. Count it; do not estimate.
• The main keywords a searcher would type appear naturally in the title or the first two sentences.
• No sentence could be moved, unchanged, into the abstract of a different study.

For the rest of the paper, see our section-by-section guide on how to write a research paper. If your university sets a style guide such as APA for the thesis, the APA 7 guide covers how abstracts are laid out on the page.