Thoughts on scientific abstracts also a science writing check-list

By Grant Jacobs 02/10/2012

I would strongly encourage research students, scientists and science writers to read Noah Gray’s excellent piece at the Huffington Post science section. Titled Abstract Science, it takes apart the key elements of good scientific abstracts.

Those that write science for general readers, including on blogs, should read it too. In breaking down his example abstract Gray, is a senior editor at Nature, has also offered advice that might serve as a check-list to ensure they address each of elements their readers need.

Readers are best to read his article, my condensed presentation here is limiting and you’d miss his writing and tips. He notes five key elements, presenting the value of each along with advice (the explanations here are paraphrased from his article):

  • Context – the background or history of a field establishes a framework to conceptualize the purpose of a study
  • Question or Purpose – make explicit the objectives (ultimate objectives may need to be inferred)
  • Methodology – writers addressing general audiences will usually obtain this from the body of the paper
  • Results – the ‘factual’ results, the data observed
  • Interpretation – what was inferred from the data observed
  • Conclusions – these might include pointers at future research or speculation (which should be clearly identified as such)

One additional element in presenting science to a general audience is interpretation of domain-specific phrases, casting these into more widely-understood phrasing whilst avoiding words that confuse general readers. Andrew Revkin, writing at the New York Times Opinion Pages, raised this issue using as his example “here we show”,

I think that [NASA climate scientist and Real Climate blogger, Gavin Schmidt’s] conclusion misses the reality that, particularly in the world of online communication of science, abstracts are not merely for colleagues who know the shorthand, but have different audiences who’ll have different ways of interpreting phrases such as “here we show.”

This might be re-phrased “here we argue – research papers present to the wider scientific community the research team’s case for their interpretations and conclusions. (Authors can, however, show data,* as data ‘just is’. It is the methodology, interpretations and conclusions that are argued.)


* When researchers argue over data, it would be more correct to say that they are arguing over the methodology that the data was derived from. Data itself are ‘merely’ recorded observations. You can argue how the data was generated, how it was recorded or observed, but the numbers themselves should be just whatever they are. (Baring fraud, of course… See also Ivan Oransky’s coverage of a PNAS paper arguing that the majority of retractions are due to misconduct. As I was writing this fellow sciblogger, Aimee, has put up a post about this issue.)

Other articles on science writing at Code for life:

Banished from science writing. Words, that is.

Outreach sections for research papers?

Scientists can’t write?

Three kinds of knowledge about science journalism

Media reporting of subsequent findings

Science communication bookshelf: suggestions, thoughts?

0 Responses to “Thoughts on scientific abstracts also a science writing check-list”

  • Nice Grant. Perhaps the most difficult job in science is abstract writing…getting all sorted in 250 words is very challenging. One thought, tha abstract was invented in the days of hard copy with limited space. In the online age are they as important? I know I rely a lot on searchs using keywords to pull out the papers of most relevance, I only then look at abstracts. Even than in my field of clinical studies I’d find it easier to look at a simple list wrt the methodology, eg
    cohort type and nbbumber
    RCT, observational, blind etc
    outcome measures

    I do not decide to read an article based on the results or even the conclusion. Alas, I do stop reading them when their presentation of results is dodgy!

  • Maybe it’s just me, but I find grant applications harder work in some ways. They’ve got some fierce space limitations of their own and their abstracts are ‘worth’ a fair bit. (Only your entire livelihood, etc… Unless you’re a full professor, in which case it’s your group’s income.) I suppose editors would argue that the abstract is valuable in drawing those outside the immediate field to use your work – ?

    My initial thoughts about your point about space limitations and electronic v. print-era issues, I agree – for the body of a paper. For abstracts I’d suspect the length limitation has stuck around for the practical reason that readers only have so much time and don’t want to be faced head-on with a full manuscript when trying to decide if it’s likely to be relevant.

    I know what you’re saying about selecting papers via keywords then details – you and I are looking for a very narrow range of topic and papers and we already know the gist of what’s going on in the field. (Hopefully a bit more than the gist!, but you get my meaning.)

    Perhaps abstracts can be thought of as in some ways more directed at those immediately outside the field, who might use your research as supporting evidence (or confounders) or other auxiliary information about their own interest. Specialists probably can guess from the title and few catch phrases or familiarity with the senior author’s interests and so on. (Might be worth asking Noah Gray what his opinion on that is.)

  • I’m with you there on grants ….I’m “research only” so also have a livelihood dependent on my ability to get to the top of the grant application pile (although lots of random effects are thrown in!).

    Various tools seem to be appearing to “help” researchers pull the gold from the dross. I think they have a long way to go and could probably do with some of your kind of bioinformatics expertise. I run regular automatic searchers on PubMed which send the results straight to my email and in the Papers program it is very easy to pull up latest articles by any author I want. I have occasionally found Google Scholar’s “related articles” useful, although GS has other issues. I’m not yet convinced about the usefulness of ResearcherGate to find articles, but it has allowed me to make contact with one or two people I may not have otherwise. It will be interesting to see where the future leads?

  • I know this is slightly tangential to your arguments, but I think the idea that space limitations are a product of the pre-electronic age are not correct. Scientific writing (abstracts, grants, and everything else) should convey the message with as much brevity and clarity as possible. This is very hard work, but equally valuable as actually doing the science.
    For example, lets say I write a paper that’s read by 300 people, and on average they spend 1 hour reading it. If my writing is shoddy and/or bloated, then maybe it takes an average of 1.5 hours for people to read the paper instead. That’s 150 extra hours of global research time. If I spend 50 hours (aka a lot) of extra time making this paper easy to digest, then I potentially save the global research budget 100 hours (by bringing it down to a 1hr average). Bean-counters and proponents of rational utility maximisers would say its in my benefit to transfer these time costs to others, but I guess I take the old-fashioned view that science is about finding new knowledge AND communicating it.
    Besides, the reality is the average time won’t increase to 1.5 hours: the readership will drop from 300 to 12.
    So, while I can hear the violins of the science writers, the violins of the readers ring louder to me.

  • I don’t disagree with you NZFungi … bloated writing (as I have a tendency to engage in) is not good or helpful. However, there are times when to communicate properly a complex or extensive piece of research that it requires more than 3000 words or 4 tables and figures.

  • You can play brevity thing both ways depending on clarity, I think.

    I agree with the spirit of what you’re saying, but conveying the important messages with clarity should be the bottom line I suspect. We read fairly quickly – so long as there is a summary I’m more concerned that it’s accurate and clear than if it’s very short.

    As for the body of a paper a lot time can be lost making sense of one set of authors abbreviations, developed to fit their material in the allotted space, and riddling out just what they meant by something that was stated too tersely to derive it’s full meaning on a natural reading of the text.

    Even putting that aside I’ve often found myself preferring papers from journals where the body of the paper can be long enough for full explanation and discussion. I guess I like to ruminate about my science and other’s thoughts (read: discussion, elaboration) aid that.

    Just loose thoughts.