Publishers obviously like to know how many people are reading their products. It can, of course, be perfectly profitable to sell a journal that no body actually reads providing institutions are willing to continue to buy it, but being able to point to readership figures undoubtedly helps. Librarians ought to keep a close eye on usage. They are custodians of recorded knowledge, but are also responsible for the efficient expenditure of public money. Their task is more difficult nowadays because most on-line journals are sold in batches: the most economic way of getting access to half-a-dozen journals that your customers want from a publisher may well be to subscribe to a cluster of 20 or 30 journals, some of which you can be quite sure that none of your customers will want. Funding bodies need to know whether the research that they are supporting is actually being used. Aside from any personal gratification at seeing their work read by others, academic authors need evidence to demonstrate their position in the pecking order, and, increasingly, to justify keeping their jobs.
In a printed library, measuring use is very difficult. I once advocated gluing some bound volumes onto the shelves, just to see if anyone noticed over the course of a year. Short of this, the least ineffective measure is the citation rating. Academics are inveterate citers. Most scientists will hardly mention the sun rising in the morning without the obligatory (Hemingway, 1926). Even this system is not perfect. Some journals adopted a course I would strongly recommend – a request to authors to mark with an asterisk* those references they have actually read in full, to distinguish them from references they have merely picked out, by abstract or even merely by title from a PUBMED list, in order to buttress an argument.
The problem with all methods of rating is that academic authors are clever (some of them are, anyway) and rapidly learn to manipulate any scoring system to their own advantage, with anything from mutual agreements to cite each others work to reorganising authors' names on potentially successful publications. Not very long ago it was common to hear research students complaining bitterly that their supervisors had insisted on being listed as first author on publications while the poor student who had done the bulk of the work was relegated to the bottom of the list as ‘et al.’ or even omitted entirely. More recently, however I heard an eminent professor, who I would have regarded as fairly ethical, openly saying ‘we have a bright lad just come into our unit, so we are putting him as first author on our new paper, to make sure he has enough high-ranking first-authorships by the time of the next research assessment.’ Some journals insist on every author signing a statement that they have actually contributed to the work in question, but this is virtually impossible to police.
Now the electronic journal has become the main channel of use, it has become possible to measure how often a particular paper has actually been downloaded for reading purposes. Even this method is not foolproof. In addition to my humble role on the Journal of Mental Health I am also on the board of Reference Reviews – a journal aimed at librarians, advising them on the purchase of dictionaries, directories, handbooks, etc. This journal has a surprisingly high download rating by comparison with the size of its target readership. I am convinced that this is because a student seeking, say, a review article on CBT will type those terms into a database and then eagerly download what appears to be the perfect answer, only to find that what they have retrieved is a book ‘review’ by me advising public librarians on whether to buy a particular handbook on ‘Cognitive Behaviour Therapy,’ which is of little or no use to them. I suspect that the number of contributions downloaded and actually read all through is considerably smaller than the number downloaded. Sooner or later, of course, we will all be hardwired into the internet, so it will be possible to keep a completely accurate record of everything that we actually do. In the meantime, citations and download figures are probably the least inaccurate measures of use that we have.
The Journal of Mental Health is, I have to admit, a low-ranking journal. It fills a very narrow slot, with its subject-matter overlapping the interests of a range of different well-established and highly regarded specialist journals. It therefore mainly attracts papers which do not, for one reason or another, fit easily into any of these journals, or are of such a cross-disciplinary and applied nature to be on the margins of any specialist's interests. It is worth noting however, that its readership appears to be steadily increasing, either because of rigorous editing, or a change of publisher to one with a more forceful marketing policy, or because of a growing interest in cross-disciplinary evidence-based research: in 2006 our top three papers achieved 195, 141 and 121 full-text downloads, respectively. In 2007, the top three got 415, 287 and 258. In 2008, the corresponding figures were 717, 558 and 534. These figures would appear to suggest that the journal may have tripled its downloading readership over 4 years: no mean achievement. The American National Library of Medicine has finally got around to recognising the journal for its MEDLINE database, so papers will, in future, be listed in PUBMED searches, etc. It will be interesting to see what effect this has on overall downloading figures. Calculating this may not be easy as the publisher has introduced a different method of recording use, but should be possible.
In 2009 the 10 most frequently downloaded papers, in order, were:
Peter Hayward and Jennie Bright, 1997. Stigma and mental illness: A review and critique. 6(4), 345.
Ian J. Norman and Edward Peck. 1999. Working together in adult community mental health services: An interprofessional dialogue. 8(3), 219.
Laurie Davidson. 2005. Recovery, self-management and the expert patient – Changing the culture of mental health from a UK perspective. 14(1), 25.
Edward Peck and Ian J. Norman. 1999. Working together in adult community mental health services: Exploring interprofessional role relations. 8(3), 231.
Rosalie Hughes, Mark Hayward and WML Finlay. 2009. Patients' perceptions of the impact of involuntary inpatient care on self, relationships and recovery. 18(2), 152.
Stephen Rollnick, Nick Heather and Alison Bell. Negotiating behaviour change in medical settings: The development of brief motivational interviewing. 1(1), 25.
Tom Trauer, Tom Callaly and Helen Herrman. 2009. Attitudes of mental health staff to routine outcome measurement. 18(4), 288.
Hazel Bassett, Jill Lampe and Chris Lloyd. 1999. Parenting: Experiences and feelings of parents with a mental illness. 8(6), 597.
Len Bowers, Rob Chaplin, Alan Quirk and Paul Lelliott. 2009. A conceptual model of the aims and functions of acute inpatient psychiatry. 18(4), 316.
Neil J. Kitchiner … (and six others). 2009. A randomised control trial comparing an adult education class using cognitive behavioural therapy (stress control), anxiety management … 18(4), 307.
The most striking aspect of this list is the extraordinary spread of publication dates. In some neuroscience or genetics journals for example, only a handful of methodological papers are likely to be of much interest within a year or two of publication. Here a paper from the very first issue of the journal is still at number six in the top 10 downloads. Paper number one, by Hayward & Bright, seems to have been in the top 10 virtually every year in the dozen years since it was published, yet on the other hand four papers managed to get into the top 10 in the year they were published, three of them from the August issue, which only gave them six months to pick up a readership. It should be emphasised, of course, that the downloading figures are still small – small enough for minor factors to make a difference. Numbers two and four in the above list, for example, are virtually two halves of the same study and were placed immediately consecutively in the same issue. I suspect that the difference in ranking may have been caused by readers downloading the first one, glancing through it, deciding that it was not relevant to their needs and therefore not bothering to download the second, rather than serious readers deciding that the first part was relevant to their needs but the second was not.
The range of topics entering the top 10 every year is quite varied. ‘Stigma’ is clearly a major topic of interest to our readers, with two or three entries in the top 10 every year. Review papers are obviously sought after. The paper by Hayward & Bright, of course, falls into both these categories, which may explain its consistently high rating. No other single topic has shown such a dominant position in recent years, but the general issue of co-ordination of care, and the ways in which service providers, carers, and the increasingly well-informed service user are working together seems to crop up regularly, possibly because this is an area which other higher-ranking but more specialised journals have neglected until recently.
It would, perhaps, be invidious to list the least frequently downloaded papers publicly. There are undoubtedly numerous scientific papers which have never been cited and seldom been read. For a highly specialised topic, when the paper has been read by an editor and three referees, and then the author has distributed the 50 e-prints which the publisher allows him, it is entirely possible that everyone who needs to read it will have already seen it before publication.
One thing that is noticeable when looking through any list of publications is that multi-authorship has spread from the hard sciences through to the whole of academic publishing. The ability to chat on-line and to exchange drafts rapidly as e-mail attachments makes collaboration ever easier. Earlier issues tended to have papers with two or three authors at most, usually all based in the same institution. Half a dozen or more is quite usual now. As an extreme, there have been letters to Nature of two or three hundred words signed by 50 authors: a clear case for ‘et al.’ if ever there was one.
Next year, comparing the 2010 figures with previous years to see what effect PUBMED has had will be a complicated task, owing to the distorting effects of the August issue 19(4). This was a special issue on diagnosis, which included contributions by various celebrated authors and politicians whose names are likely to be sought terms in their own right. The publishers very sensibly put that particular issue out for free on the web, and it also gained a fair amount of extra publicity through press releases, etc. Within 3 weeks of publication the editorial to that issue had been downloaded 1462 times, thus achieving the highest annual download figure for anything published in the journal so far, pipping Terry Pratchett's Diagnosing Clapham Junction Syndrome from the same issue into second place with 984 downloads. Congratulations are due to Til Wykes and Felicity Callard, the authors of that editorial. There are not many writers who can claim to have topped Terry Pratchett in a popularity contest. It is, I hope, no disparagement to them to point out that their contribution came first in that issue, and that, with the exception of Terry Pratchett, the download figures very broadly taper off in accordance with their position in the issue. I suspect that much of this is due to journalists and other casual readers, in those first three weeks or so, starting at the beginning of the journal and gradually losing interest as they worked their way through it. Another look in a year or two's time might give us a more balanced impression of real use.
It must be pointed out that the Journal of Mental Health is not, in terms of annual downloads, the jewel in our publisher's crown. I found it quite interesting to look through Informa Healthcare's annual figures for their most heavily used journals. In every year from 2005 to 2008, every single one of their 10 most frequently downloaded papers was from their highly successful Expert Opinion … series. Far and away the most successful was Expert Opinion on Investigational Drugs with 19 top 10 hits in 4 years, followed by Expert Opinion on Pharmacotherapy with nine, Expert Opinion on Emerging Drugs with four, and then Expert Opinion on Drug Safety, Expert Opinion on Drug Delivery, Expert Opinion on Drug Metabolism & Toxicology, Expert Opinion on Biological Therapy and Expert Opinion on Therapeutic Patents with two or three apiece. It should be noted however that, as pointed out earlier, the life-span of neuroscience papers is very limited. In 2005, 10 out of the 10 most downloaded papers were from 2005 publications. In 2006, the same. In 2007, 8 out of 10 were from 2007 publications. In 2008, 9 out of 10. ‘Slow and steady often wins the race’ (Harris, 1881). It may well be that Hayward & Bright, or even Wykes & Callard, may continue in use for some years yet, and thus eventually overtop their evanescent peers.
Acknowledgement
I am grateful to David Hunt, business manager Journal of Mental Health, and to Daniel Falatko of Informa Healthcare for giving me access to the figures quoted. All interpretations are my own.
References
- *Harris, J.C. (1881). Uncle remus stories. Atlanta, GA: Appleton & Co.
- *Hemingway, E. (1926). The sun also rises. New York: Scribner & Sons.