This page is designed to help you write the abstract section of your research paper, dissertation or undergraduate project. Abstracts are a short overview of your whole paper, not so much a teaser but rather the whole story told in fewer words. This page deals primarily with times when you are reporting new findings i.e. research data, but the advice is generally applicable to all abstracts.
Good new, in my experience, most new writers do a decent job at writing the approach and results parts of their abstract but do less well at the top and tail. Remember you need to tell readers the whole story, you want to encourage them to continue reading by telling them not only your core results but also why your work matters.
Do more than summarizing the findings; show why your results are interesting and how they fit within the wider world.
Structure
Check the instructions! Before you go too far, check the journal or assignment requirements (read the syllabus). Certain journals have very different set ups and not every article type will have an abstract (e.g. Letters to the editor usually don’t).
Structured Abstracts.
Many journals are prescriptive in abstract layout, requiring specific subsections. Usually these are: background/introduction, aims, methods, results, conclusions. Structured abstracts can be easier to write, as taking away flexibility also takes away some decision making. You can’t go too far wrong. Therefore, even if you are not required to use a structured approach, you could still use those subsections (without subheading) to put together your first draft. Be aware that you could probably do a better job with a more narrative style.
Freeform abstracts
If you consider structured abstracts to be like bullet points, then freeform abstracts are more like story telling. You deliver the same core material and generally follow the same overall structure, but now you have flexibility to tweak the order to improve the flow and to make your abstract more enjoyable to read. The most obvious way to do this is to combine the methods and results for each stage of the experiment and by using better linking phrases to smooth your transitions.
Your abstract must be able to stand alone, it should tell the whole story of your manuscript
Word count
Always check your word count as they can vary a lot from journal to journal. Usually you are allowed somewhere between 200 and 350 words. In the other parts of your manuscript, word count is not usually something I worry about until later. However, in an abstract the difference between a 200 and a 350 word abstract is quite large in terms of the style and the amount of content you can include. Therefore, I recommend that you be aware of your target length before you start. Don’t know the target journal? 300 words is a good length to go for.
Tenses
- Use past tense for intro and most results (Our experiments revealed…).
- Present tense for results/conclusions that will remain true (these data demonstrate that X influences Y)
- Future tense if you make any next steps type comments.
Voice
Active voice is more enjoyable to read, delivers the message within the story more effectively and uses fewer words. The active voice puts the action in the subject of the sentence, this means that the sentence takes the strucutre agent::action::outcome. There is an editing guide soon (sign up to the blog for updates) but for now this page is quite good for describing voice.
Be aware that often your data will require passive voice e.g. the phrase in the box below might need to use the passive version if you don’t have the appropriate data to support interpretation as a direct influence.
Active voice; “X influences Y” is usually better than the passive voice of “Y is influenced by X”
Balance [for scientific abstracts]
The biggest challenge I’ve found with new writers is that they overdo the background in an abstract and include too little information about the important new stuff (methods, results and conclusions). Irrespective of whether you are using a structured or freeform abstract, the following guide to lengths should help.
Keep proportions approximately the same as you scale for word count;
- Motivation and background – 20%
- Hypothesis and Aims – 10%
- Methods – 20-30%
- Results – 30-40%
- Conclusions – 10%
- Sentence 1: Motivation – why should your readers care about the topic?
Start with one sentence of the big picture material that establishes context. Make this accessible to a scientist from a slightly wider discipline than the scope of your target journal. Keep in mind that at the end of the abstract your final conclusion will revisit this wider point. Therefore, you shouldn’t start too wide here.
- Sentence 2 and 3 – Knowledge gap / problem statement – What wasn’t known about your research area before you started?
The next one or two sentences should contain more detailed background information. Don’t make this too long! All you need to do is set up your specific research question. Deliberately highlight the knowledge gap you addressed and why that specific gap is important. Together sentences 1, 2 and 3 should establish what and why.
- Sentence 4. Aim/Objective/Question – What did you specifically want to know?
One sentence that clearly states the problem that your work has addressed. This is best if it is a clear single question that all your experiments point toward. If the goalposts moved as your work progressed, state the natural start point of the story.
- Sentences 5 to 7 – the main storyline.
- Methods – How did you do your experiments?
- Results – What results did you get?
For most journals this should be the longest section. In a structured abstract, you detail the methods and results separately. Usually about 2:1 ratio of methods to results. In freeform, I usually tie the experiment and the result together, e.g. “Confocal microscopy revealed co-distribution of X and Y”
You do not need to include everything. Instead focus on the key parts of the story that answer the main study question. Always indicate the direction and magnitude of effects as well as the strength of your inferences (i.e. p values!) and the statistical tests. As with your results section tie these together with link phrases so it is easy to read. Draw attention to anything unexpected and/or if they support/refute your hypothesis.
Sentence 8 Conclusions
DON’T FORGET THIS. A single sentence wrap to put your work back into a real world context. Make your conclusions a as general as your data can support while making sure it doesn’t an overreach. As noted above, your last line of your conclusion should go back to the same width as your opening sentence. Your conclusion should also agree with your title and the conclusions section of your discussion. Use different phrasing in the two sections but keep the same message.
Note that depending on the journal, this part of the abstract can be of different length. The proportion also depends on how big/impactful your findings are. Journals like Nature expect this to be the major part of your abstract.
Writing a student paper? I think the conclusion is most often missing from the abstract I have marked. Don’t forget it, it is required to round out your abstract
Thesis / Dissertations abstract
The concept of an abstract for long-format writing like a PhD thesis or Masters dissertation is the same as for a paper, except you will have a lot more work to cover and a proportionally bigger word count to play with. You likely will still not be able to discuss everything; therefore, the abstract is an opportunity tell your examiners which parts are the most important. You can set the scene for all the work to follow. Most importantly, this is a good chance to very briefly highlight the value, extent and novelty of your work as a whole.
As always, step one is to check the instructions. If there are no word limits, then set yourself a limit of one single-spaced page, any longer than that, and it’s a sign that you haven’t been very discerning about what you have selected to highlight! You don’t want to frustrate your examiners before they have even seen a single data point. In terms of structure, long-format writing abstracts are almost always freeform, so you have flexibility to do whatever works.
As usual, start with the motivation and rationale, then a clear statement of the problem that your work addressed. Limit this to about 20% of the words, usually one paragraph. For methods and results, you have two main options. If you used a single model system or a limited range of approaches throughout the work then describing the experimental system using its own paragraph might make sense, with the results then described in order. This can work, for example, if you have generated a new transgenic mouse model or a genome-edited line, and then characterised it in the rest of the thesis.
If instead, you have results chapters that are a bit less connected, then an option is to use one paragraph per chapter with a brief overview of the methods and major results together in each paragraph. Together the methods and results should make up about 60% of the abstract.
The final 20% should be dedicated to conclusions. This might sound like a lot, but you should cover what the data mean, whether they answer the problem statement, and how they have advanced the field.
I don’t recommend including any future directions in the abstract, you don’t need anything to distract from what you have done. End with a sentence that wraps up the story by saying what your work means in relation to the original motivation.
Other considerations
No references
Usually the referencing rules are relaxed for an abstract. There are exceptions, if your whole story is hung on the back of a single previous finding or if you are publishing as part of a linked series of papers. In these rare occasions, be aware that there is usually different rules for how to cite in the abstract compared with the rest of your in text citations. Most times you must write the whole citation including title and full author list. As usual check “the instructions for authors” page of the journal.
Keywords and search engine optimisation
You can ignore this if writing a student paper but it is something you should consider if aiming for a journal.
It’s all very well writing a great abstract but if no one reads the abstract in the first place then you are no better off. You need your work to be found when people are searching. Think about who you want to read it and how they will use search engines (not just pubmed/scopus but google and Bing are often used to locate papers). Test some search terms and see what comes up.
I’m not expert in SEO but what I understand is:
- Place your main keywords in the first two sentences
- Be consistent in terminology throughout manuscript including subheadings, figure legends etc.
- Don’t overuse keywords, too much repetition (‘keyword stuffing’) may result in search engines ‘un-indexing’ your article.
For biomedical work you can use MeSH keywords, headings in National Library of Medicines-controlled library thesaurus to help you find good keywords. Google Adwords keyword planner or Google trends can also be good sources of information. But…
Don’t let the quest for adding keywords ruin your message!
Multiple drafts
Like all the rest of your document, your abstract will need multiple drafts before it is perfect. Abstracts are usually the last thing I write in terms of first draft. However, they then get read every subsequent time I am reading and editing the paper. Looking back, I think I actually think I should have spent more time perfecting the abstract of my papers and will be consciously focusing more energy on them in the future.
Before you submit
Get someone who is not an expert in your field to read your abstract! If they can’t understand it, go back and edit more.
Some useful links
Nature summary paragraph explained – pdf
Our writing guide section for the rest of your paper.
Pdf version of this page
I’ve written a book… “Life in the Lab; the essential guide for researchers”. It has everything in it including applying for positions, conducting lab work, experimental design, data analysis and loads on writing. Find out more here – or click the pic to go to the listing on Amazon.
LaNts and Laminins 