For most journals, abstracts began to appear early in the 20th century and became popular shortly after World War II. As scientific reports became more structured and disciplined, the abstract followed suit, evolving from a simple narrative snippet to a longer, structured editorial section with headings all its own. While studies of abstracts often show some of the flaws of abstracts (more on that later), structured abstracts became longer than their predecessors, often by a significant amount.
For many journals, the abstract is the only section of a scientific article that has a heading as definitive as “Conclusions.”
The abstract was introduced as a convenience for the busy reader, an accommodation to battle information overload during an information age that in retrospect seems almost quaint.
The abstract was a nice embellishment of print, designed for the print reader. It wasn’t created because there was an existential crisis facing research publishing in print. It wasn’t created in anticipation of large-scale electronic databases. It was a nice editorial invention that proved popular with readers and was easily replicated.
Once abstracts came into being, mimicry and craft finished the job of installing it as a “given” in scientific research articles.
Abstracts have provided exactly what their creators and purveyors hoped — a short, digestible, representative version of a research article. While there are known limitations and problems with abstracts (e.g., attempts to “game” the abstract by making it more positive than the full article, the difficulties of fully representing the nuances of a full report, and selective representation of favorable data), abstracts are generally perceived as useful and desirable.
The International Committee of Medical Journal Editors (ICMJE) endorses the use of structured abstracts. Their rationale has two tenets, only one of which is still true:
Because abstracts are the only substantive portion of the article indexed in many electronic databases, and the only portion many readers read, authors need to be careful that they accurately reflect the content of the article.
Abstracts are no longer “the only substantive portion of the article indexed in many electronic databases.” Google, the main search engine used by many scholars these days, indexes the full-text of most journals. And that’s just the world’s most popular search engine. In addition, many fields have large-scale databases of millions of abstracts. Some of the resulting businesses — PsychINFO, ProQuest/CSA, and others — are quite large.
There is no denying the ICMJE’s statement that abstracts may be “the only portion many readers read.” In fact, abstracts have frequently been accused of being a little too useful, with complaints coming from editorial offices that authors cite articles purely based on the contents of the abstract; repeated findings that readers simply scan abstracts as a substitute for reading the article instead of using abstracts as the appetizer and entry point; and full-on businesses being established by harvesting abstracts — which are copyrighted in some cases — and creating databases of these summaries., this often sufficing as a review of the literature. As information overload has increased, as the time and energy available for literature review has decreased, and as each generation has become more attenuated to the utility and substitution value of abstracts, reliance on the abstract has increased.
Even with all of this, little has changed in how editors create or publisher purvey abstracts.
Now, perhaps new factors in the presentation of scientific information may merit reconsideration of the abstract’s role, especially in online databases.
The extent to which journal readers browse abstracts in PubMed or a similar database has made these familiar summaries a level of the literature for themselves. Tensions exist, however. Publishers are being increasingly judged by value calculations based on the usage they drive to their core properties. Whether it’s institutions measuring cost-per-download or advertisers measuring cost-per-impression, there are growing pressures to bring readers directly to your core journal site.
As these demands for usage-based value metrics increase, promulgating abstracts thoughtlessly may be something to reconsider. In fact, given the clarity of our article titles and other potential summarizing approaches (such as a return to brief, narrative descriptions of the hypothesis without divulging the results), is there really a need these days to save someone one click to the abstracts on our own sites?
For open access publishers, the risk of abstracts keeping readers away is similar — after all, what does it matter to the user if an article is made available at no cost when the user is satisfied by an online abstract via another source like PubMed? At the same time, open access publishers are building article usage, commenting, and other services that depend on users visiting their sites. The syndicated abstract is creating just as much of a distraction for them in this age of information snacking as it is for subscription publishers.
Some publishers are experimenting with the way they summarize articles. The Lancet has a separate summary, PLoS has synopses, and various journals have “take home message” boxes of one sort or another. But the abstract remains a mainstay of syndicated value.
The abstract tends to be taken for granted by publishers at the same time they are appreciated by readers and exploited by services. This disjunction of value is hard to fathom, especially today. Why are abstracts — arguably the most distilled, useful, structurally predictable, and desirable editorial feature of scientific articles — given away freely in an online, data-driven content economy in which usage is becoming the measure of value?
Providing the abstract freely to anyone who wants to use it has become a habit, probably a leftover of our print and “information scarcity” mindset — with this mindset, it seems harmless to promulgate abstracts as widely as possible, and doing so seems like a way to battle scarcity. But in a networked and “information abundance” world, is carelessly syndicating a valuable substitute for articles a habit we need to sustain? And how long can we afford to continue it?
24 Thoughts on "Reconsidering the Abstract — Are the Unintended Consequences Mounting?"
What a fascinating issue. It brings out the basic tension between service (providing information) and business (building readership). In that context perhaps the abstract should be viewed as a form of advertising and structured accordingly. You mention not giving away the results. This could be a new form of creative writing.
Exactly my thinking. Abstracts are rich information packets we tend to take for granted. For publications that sweat over getting them right, the amount of information given away is even greater. Why not sweat over them to fit the current information landscape and editorial/business needs?
Talk about shovelware. (I guess this is trowelware, given its smaller size, but it’s the same idea.)
Excellent – touches on a big problem: “complaints coming from editorial offices that authors cite articles purely based on the contents of the abstract; ” And b/c of this, authors misinterpret findings and recommendations – they don’t read the details that might indicate that a recommendation or finding only pertains to a sub-group, or was insignificant is some cases. When writing about clinical practice, those details matter!
As a researcher at a small institution with limited library resources, abstracts are invaluable for deciding whether or not to pursue a particular article (whether by web query, purchase – $35+ ain’t cheap!, or contacting the author). This is particularly important for those cases where the topic of an article is only vaguely defined by the title (this happens more frequently than acknowledged in the above post), and there’s a question as to whether or not the article actually includes the information I want in the appropriate depth.
I agree that it’s bad to rely upon the abstract too much, but will making them scarce really solve the problem? In many cases, it’s a problem of time management (too many papers, too little time) rather than a problem of information access. And how would this scheme of limiting abstracts benefit the authors (those people providing the content for free or paying page charges)?
I think the abstract form could be replaced by something just as helpful in determining the content, but something perhaps less prone to problems and less framed by print conceits.
What might you propose? Any replacements for abstracts would have to be carefully devised in order to avoid replicating the problems of their predecessor. Without author/reader buy-in (we love our abstracts), it could be an uphill battle. One option to fix part of the problem would be for publishers to prohibit indexers from grabbing full abstracts, replacing them with a link to the abstract at the journal website. This would be akin to what most publishers do with Google, where you can search the full text, but not easily view it without going to the journal itself.
I think there are plenty of options. I think the Lancet has other summaries that are more online-friendly. Many journals have “take home” message boxes. Something that gives an idea of what an article contains without giving into the foibles of the traditional abstract seems easy to accomplish. A title already says a lot.
“In a world, where cells divide and die. One cell dared to dream… ‘Johnny! The mitochondria is going to blow!’ From the writers of ‘The Good, the Bad, and the Cell Type-Specific Roles of Hypoxia Inducible Factor-1 in Neurons and Astrocytes’ comes the story of the little cell that fought for his freedom of expression. ‘I just feel… different.’ ‘You are different, Kip.’ [explosion, anguish] ‘No!’ [black out] Coming soon, Nature, Volume 472 Number 7347.”
I’d read that.
Great post on this issue, thank you – I second Shawn’s concerns about misinterpreted findings based on “appetizers” rather than a well-balanced meal leading to inaccurate writing, especially in the health sciences.
I think David’s point about the tension between service and business is at the crux of the matter here. This tension plays out differently in different institutions – publishers will understandably try to make the shrewdest business decisions, and libraries, while primarily service providers, are becoming increasingly invested in the research done at their institutions.
And I think those tensions can be resolved without anyone getting snubbed. It just takes some rethinking and experiments. The optimal solution would give readers both enough of an idea of what a paper is about while enticing them to read it. Now, the abstract has become a substitute, and there are downsides both intellectually and commercially to that.
I just want to point out that I’m starting to see (finally!) that _book_ publishers are realizing the value of abstracts. (The intellectual and commercial dynamics are very different in the book side of the world — especially since books are so much more varied and more inconsistently structured than journal articles, and since books are not commonly available as open access.) I have one client who routinely creates an abstract for every chapter of every book they publish. These abstracts are of course not “in the book” in a formal sense, and don’t appear in the print book; they’re metadata. They are freely available online, whereas the full book content is only available for purchase (print or eBook). I think this is an _excellent_ practice, and I hope more book publishers come to realize this.
Our academic imprint publishes its first book at the end of this month and it will include an abstract in the book (along with a list of key terms), in the prelims. Often the only summary available for a book, even on databases,is the blurb. We’ve written one of those as well, but blurbs are written for a specific purpose (to encourage purchase)and are not usually suitable for other purposes such as supplying metadata. Having taken this decision, I’m glad to see some enthusiasm above for abstracts in book publishing!
As a systematic reviewer of qualitative research reports, I find many published abstracts to be wholely lacking. The description of qualitative methods is often inadequate and the entire article must be secured soley to determine what methods were used to conduct a study. If publishers are concerned about “giving away the farm,” perhaps more emphasis should be placed on writing clear titles, research questions, and methods versus explicating the findings and conclusions.
“is there really a need these days to save someone one click to the abstracts on our own sites?”
Isn’t this the entire rationale behind the existence of PubGet, that there’s incredible market value in reducing researcher effort by one click (which, from my extrapolation of their numbers means saving a whopping 8 seconds per day)? Speaking of whom, it should be noted that, at least as of the last SSP Meeting, PubGet’s search engine is based on abstracts and metadata, and not the full text of articles so there are indeed some instances where both tenets above are true.
An interesting side-concept has emerged in the Twittersphere about this post — that abstracts are open access. While they are free, appropriating them opportunistically into the OA movement seems like revisionist history — so OA has existed since the 1980s?
Your post reminds me of a comment I made at an SSP meeting early in the digital revolution. I suggested that publishers sell the abstract and give away the article, since as you said, the abstract is often times all that is read by the reader.
The business value of abstracts may well be changing in our digital world. For books, abstracts might help generate additional sales as PDA begins to replace traditional approval plans. For journals, by contrast, the abstract might lead to cancellation of subscriptions and replacement of them by “pay per article” schemes.