How do researchers use online journals?

ResearchBlogging.org

Last Monday I was listening to a very interesting presentation by Ian Rowlands, reader in scholarly communication in the Department of Information Studies at University College London. He and his colleagues are interested in how researchers find and use information, and how this has changed with the internet, especially for the Google Generation (people born after 1993). If you want to be part in this research (and have some fun), you can take part in the BBC Web Behaviour Test. The test will help you discover which species of web animal you are (I'm a fox).

Flickr photo by Craig Anderson.

In another project, funded by the Research Information Network (RIN), Ian and his colleagues are studying how researchers are using electronic journals. The findings of the first part of the project were presented and discussed in a workshop last July. The presentations are available as PDF download, and as podcast with interviews of the speakers. The findings were summarized in a paper also published last July: Online use and information seeking behaviour: institutional and subject comparisons of UK researchers.

In the paper, the use of Oxford Journals by 10 major UK research institutions was analyzed in the fields of life sciences, economics and history, using the server logs for the full year 2007. Some of the key findings of the study include:

One third of users access Oxford Journals outside business hours
9.7% of uses happened on a Saturday/Sunday and 30.1% between 6 PM and 9 AM. This means that about one third of users accessed Oxford Journals outside typical business hours, either working late or from home (the study didn't distinguish between these two). These numbers indicate that remote access (from home, but probably also when traveling) is important for many users. This is obviously not an issue for Open Access journals, but institutions need to provide practical solutions (VPN, etc.) for subscription journals. From personal experience this remote access is still overly complicated. And these numbers also mean that librarians will not be available for support questions one third of the time.

Around 40% of sessions originated from a Google Search
In 2004 Oxford Journals opened up to Google for indexing. I didn't expect this important role that Google seems to play in finding scholarly papers, and I would be very interested in feedback from blog readers. Only 4% of sessions originated from Google Scholar (22% in economics). These results probably explain why Google Scholar hasn't seen that much development since it was launched. The search function at Oxford Journals was rarely used.

43% of users of history journals, but only 16% of users of life sciences journals used navigational tools (table of contents, etc.) provided by the journal. This statistic obviously doesn't look at users getting the table of contents via email or RSS, but it again shows that access via search now probably is more common than via browsing.1

Most users spend little time on journal webpages, but return often
The average number of articles viewed per session was 1.1, and the average session time was just over 4 minutes. Users rather return often, usually via a search. These numbers indicate that journal webpages are not a place where users spend a lot of time. Unless journals change this (e.g. by more active involvement of users via comments and other social networking features, etc.), they probably can't expect to generate significant revenue from online advertising. The internet has not only dramatically changed the role of libraries, but also for journals, as users are mostly interested in single articles, rather than the journal as a whole.

The median age of articles was 48 months (life sciences), 73 months (economics), and 90 months (history)
In the life sciences only 25% of the articles were no more than 16 months old, but another 25% were over 104 months old. I would have expected that the median age of articles would be much lower in the life sciences (it was two years in a similar study with ScienceDirect2). It seems as if most papers are not accessed when they are published (in the first few months after publication), but rather as the result of a search strategy, e.g. when writing a paper.

Life sciences users rarely read abstracts on publisher platforms
This should not come as a surprise, as life sciences users typically read abstracts in specialized databases, particularly PubMed. But maybe Journal publishers should stop displaying papers in an abstract view, saving users and themselves some effort. PLoS journals don't have an abstract view, but the Biomed Central journals (which are also Open Access) do. Subscription journals (including Nature) typically display the abstract instead of fulltext to users without subscription access, so there is also no need for a separate abstract view for them.

The number of PDF views was higher than the number of fulltext HTML views (178,152 vs. 106,582). This difference was much more pronounced in economics and history journals, probably indicating that here most papers were printed out and not read on the computer.

Nicholas, D., Clark, D., Rowlands, I., & Jamali, H. (2009). Online use and information seeking behaviour: institutional and subject comparisons of UK researchers Journal of Information Science, 35 (6), 660-676 DOI: 10.1177/0165551509338341

1 My July 2008 blog post Do online journals narrow science and scholarship? discussed potential consequences.

2 CIBER, Evaluating the usage and impact of e-journals in the UK. Working paper 5. Available at http://www.ucl.ac.uk/infostudies/research/ciber/

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Snippets. Bookmark the permalink.

15 Responses to How do researchers use online journals?

  1. María José Navarrete-Talloni says:

    What a nice post, Martin! It’s cool to see how internet and the virtual world have become a whole new science, including these studies of search patterns and web behaviours… very interesting!
    BTW… I’m an ostrich :-)

  2. Frank Norman says:

    Another ostrich here. I’m not sure that it is a flattering description!
    I think the omnipresence of Google helps – if you are looking for a particular paper you can just type in an author name and a couple of title words to your toolbar and off it goes to Google.
    Librarians have been telling publishers for years that they shouldn’t both with elaborate search features on their websites, but they all still feel the need to develop these features.

  3. Eva Amsen says:

    I’m an octupus, which is like a fox, but less social! Pshaw!
    I didn’t play the games, though. I clicked past them because I don’t have time for all that. So it wasn’t a surprise that I ended up with one of the “fast-moving” animals. Who has time to play games!?

  4. Richard Wintle says:

    Hm. BBC Web Behaviour Test requires (a) sign-in, and (b) 20 minutes. I think not.
    Outside business hours – not surprised. If I need to look up a journal article on a weekend, I look it up, I download it, done and dusted. If I have a problem it’s either because (a) my institution doesn’t subscribe to the journal, which nobody can do anything about (at least not on any kind of useful timescale), or (b) something’s broken in the institutional IT, which complaining to a librarian will not fix.
    Not surprised that users don’t read abstracts on journal websites. PubMed is for reading abstracts, journal sites are for downloading, in my view.
    Regarding Google, I too am surprised. I would have thought that the vast majority of entries to publisher websites would come via curated search engines (PubMed or similar). And I think website exits should be checked as well – how many of those entries via Google resulted in someone immediately exiting to an institutional library portal, or to PubMed to find related literature?
    Finally, I’m not surprised that Google Scholar isn’t more widely used. It’s not exactly on Google’s front page, is it? And although useful for full-text searching of article content (in a leaky, too-many-duplicates kind of way), I don’t buy for a second that it’s better for keyword and concept searches than good old PubMed (or similar).
    But I’m a Luddite, so take all the above with an appropriately-sized grain of salt.

  5. Martin Fenner says:

    Eva, I also didn’t play the games, ended up with a fox. I don’t know whether any of the animals is better than the others. Octopus sounds tasty, but that’s a different story.
    Maria, doing internet research would be something really interesting to do. And it helps to better understand how is science online stuff works (or doesn’t).
    Frank, in a way the results of the project are really frustrating. Most researchers don’t spend much time with libraries or journal websites, just quickly go to the internet and get their stuff. They would probably be *web leopards* in the BBC test.

  6. Eva Amsen says:

    _”20 minutes”_
    Not if you skip the games =D
    I did it in 5.

  7. Vinod Babu Damodaran says:

    Hi Martin, nice to read the article.
    It is true and we have to accept the fact that people are now more depending on Google search, including both “google generation” and old “non-generation”, rather than depending on more authentic database searches.
    (Why you restrict to Oxford journals only?!)

  8. Martin Fenner says:

    Vinod, the study used the Oxford Journals server logs for a full year. That already was a lot of data to handle. Footnote 2 is about a similar study with ScienceDirect data. Oxford Journals was used because they have quality journals in all three subjects covered: the life sciences, economics and history.

  9. Vinod Babu Damodaran says:

    Hi Martin, thanks for your reply.
    Sorry I forget to convey you that I identified myself as a web fox, but the survey was too long with 18 steps………………. , really very long.

  10. Nicolas Fanget says:

    Fox here as well, but I think the ostrich seems tastier than the octopus…
    Google scholar is not very well advertised (or developed as far as I can tell) but does allow you to search the full text of articles. If I remember well, you might be able to see a snippet of a paper you do not have subscription for, helping decide whether to buy it.
    Otherwise, Google/Pubmed search all the way, unless I know exactly where to find what I’m looking for, in which case I _might_ go to the journal page. I find that some journals are very poorly indexed by the *G*, for example plant journals.

  11. Donald Berkholz says:

    I think there’s another explanation for PDF vs HTML that you’re neglecting. The PDFs are just easier to read regardless of format because they’re better laid out. Given a choice, I will never read a webpage with its poorly placed, poorly formatted figures, if I can open a PDF instead.

  12. Alyssa Gilbert says:

    Great post, Martin. I tried to take the BBC test, but it wouldn’t let me.
    I never really used Google Scholar until I started my new post-doc. Since it’s in a new research area, I am unfamiliar with the journals, and have found it to be a good resource.

  13. Martin Fenner says:

    Donald, I agree that PDF usually is easier to read than HTML, but it’s getting more complicated with things like the “Article of the Future”:http://blogs.nature.com/mfenner/2009/07/26/how-does-the-article-of-the-future-look-like format.
    The _PLoS_ journals show the number of HTML views and PDF downloads for every paper as part of their “article-level metrics”:http://blogs.nature.com/mfenner/2009/08/15/plos-one-interview-with-peter-binfield. HTML views usually far outnumber PDF downloads.

  14. Nicolas Fanget says:

    Martin, I think that HTML vs PDF metrics are skewed by the fact you have to access an HTML page before you can download the PDF, unless you have a deep link direct to the PDF. Also, I am sure many people will have a quick look at the HTML to decide whether to download to their library or not (1 view), then move on or download. So the PDF d/l number (keep) really should be subtracted from the HTML view number (taste).

  15. Martin Fenner says:

    Nicolas, I haven’t done a systematic analysis, but if I look at the _PLoS_ article-level metrics, HTML view seem to outnumber PDF downloads by a factor of 10 or more.