On 2 October I attended another of the OU’s professional development events. This time, it was an event organised by the OU library. Facilitated by Chris Biggs, Research Support Librarian, the session aimed to outline three things: “common bibliometrics, and their use and misuse”, “what are Altmetrics and where they can be found” and DORA, which is an abbreviation for the “Declaration on Research Assessment (DORA)” along with the “responsible use of metrics”.
I was particularly interested in this section since I’m a co-editor of an international journal called Open Learning. Bibliometrics are sometimes discussed during the annual editorial meetings between the editors, the members of the editorial board, and the publisher, who is Taylor and Frances. During these meetings, various numbers are shared and summarised.
When I saw the title of the event, my main thought was: “I should go along; I might pick up on a point or two”. What follows is a set of notes that I made during the session, along some of the useful weblinks that were shared. One thing that I should add is that the structure of these notes come from the facilitator, Chris, and his presentation. Towards the end of these notes, I share a set of reflections.
Citations
I missed the first couple of minutes, joining just at the point when Chris was talking about the ‘the social dimensions’ of citations. Citations are all about giving credit. A useful link to look at is the page Citing Sources: What are citations and why should I use them?
One view is that the more citations an article has, the more popular it is. Subsequently, some might associate popularity to quality. An interesting paper that was referenced had the title Citations, Citation Indicators, and Research Quality: An Overview of Basic Concepts and Theories.
Retuning to the notion of the social dimension of citations, one metric I noted down was that self-citations account for 12% of citations. A self-citation is where an author references their own, or earlier work. Whilst this can be used to guide authors to earlier research, it can also be used to increase the visibility of your research.
A concept that I wasn’t familiar with but immediately understood was the notion of a citation circle or cartel. Simply put, this is a group of authors, typically working in a similar field, who regularly reference each other. This may have the effect of increasing the visibility of that group of authors. Chris shared a link to an interesting article about the notion: The Emergence of a Citation Cartel.
A further notion that I hadn’t officially heard of, but was implicitly familiar with, was the notion of the honorary citation. This is where an author might cite the work of a journal editor to theoretically increase chances of their paper being accepted. As an editor, I have seen that occasionally, but not very often. On a related point, the publisher, Taylor and Francis has published some very clear ethical guidelines that editors are required to adhered to.
Something else that I hadn’t heard of is the Matthew effect. This means that if something has been published, it will continue to be cited, perhaps to the detriment of other articles. Again, we were directed to an interesting article: The Matthew Effect in Science.
It was mentioned there are interesting differences in between academic disciplines. The pace and regularity of citations in the arts and humanities can be much less than, say, a busy area of scientific research. It was also mentioned that there are differences between types of articles. For example, reviews are cited more than original research articles, and methods papers are some of the most cited papers. (It was at this point, I wondered whether there were many articles that carried out reviews of methodologies).
An interesting reflection is that articles that are considered to have societal benefit are not generally picked up by bibliometrics. This immediately reminded me about how funders require researcher to develop what is known as an impact plan. I then remembered that the STEM faculty has a couple of impact managers who are able to provide practical advice on how researchers can demonstrate the impact and the benefits of the research that they carry out.
All these points and suggestions lead to one compelling conclusion, which is that the number of citations cannot be directly considered to be a measure of the quality of an article.
An important element is all this is, of course, the peer review process. Some important points were made: that peer review can be slow, expensive, inconsistent, and prone to bias. As an editor, I recognise each of these points. One of the most frustrating elements of the peer review process is finding experienced and willing reviewers. During this session, I shared an important point: if an author holds views that are incompatible or different to the reviewers, it is okay to have a discussion with an editor. Editors are people, and we’re often happy to chat.
Bibliometrics
There are a few different sources of bibliometrics. There is Scopus, Web of Science, Dimensions, CrossRef and Google Scholar. Scopus and The Web of Science offer limited coverage for social sciences and humanities subject. In contrast, Google Scholar picks up everything, including resources that may not really be academic articles. A link to the following blog, Google Scholar, Web of Science, and Scopus: Which is best for me? was shared.
There are, of course, different types of metrics. Following on from the earlier section where citations were mentioned, there is the notion of normalised citations, percentiles, and field citation ratios. Normalised citations (if I’ve understood this correctly) is the extent of an article being over a time period. Percentiles relate to how popular, or widely cited an article is. There is, of course, a very long tail of publications. Publications that appear within, say, the top 1% or top 5% are, of course, highly popular. Finally, the field citation ratio relates to the extent to which an article is published within a particular field of research.
There is also something called the h-index, which relates to the number of publications made by a researcher. During Chris’ presentation, I made the following notes: the h-index favours people who have a consistent publication record, such established academics. For example, a h index of 19 means 19 papers that have been cited 19 times.
Moving beyond metrics that relate to individual researchers, there is also something called the journal impact factor (JIF). Broadly speaking, the more popular or influential the journal, the higher its impact factor. This has the potential to influence researchers when making decisions about how and where to publish their research findings. It was mentioned that there are two versions of a JIF: a metric that includes self-citations, and another that doesn’t.
Metrics that relate to official academic journals isn’t the whole story. Outside of ‘journal world’ (which you can access through your institutional library) there are an array of ever changing social media platforms. Subsequently, there are a number of alternatives to citation based bibliometrics. Altmetrics, from Digital Science, creates something that is called an attention score, which consolidates different ‘mentions’ of research across different platform. It can only do this if there is a reference to a a persistent digital object identifier, a DOI.
Previews of AltMetric data can be seen through articles that are published on ORO, the university’s research repository. Although I’m risking accusation of self-citation here, an interesting example is the following excellent paper: Mental health in distance learning: a taxonomy of barriers and enablers to student mental wellbeing. Scrolling to the bottom of the page will reveal a summary of tweets and citations; metrics from both Altmetric and Dimensions.
There are a couple of other alternative metrics that were mentioned: PlumX, which is from Elsevier, and Overton, which is about the extent to which research may be potentially influencing policy.
Responsible metrics, the OU and DORA
Towards the end of the event, DORA, Declaration on Open Research Assessment was introduced, of which the OU is a signatory. One of the most salient point from the DORA website is this: “Do not use journal-based metrics, such as Journal Impact Factors, as a surrogate measure of the quality of individual research articles, to assess an individual scientist’s contributions, or in hiring, promotion, or funding decisions”. There is also a case study that relates to the OU which can be found on the DORA website.
This final and important part of the session, which had the title ‘The Idea of Responsible Metrics” was covered quite briefly. The topic of DORA is something that I really do need to look at in a bit more detail.
Reflections
I learnt a lot from this session. One thing that really grabbed my attention was the h-index. As soon as the session had finished, I asked myself a question: what is my h-index? It didn’t take too long to find it out.
After finding my h-index figure, I made a mistake; I wondered what the h-index for some of my immediate colleagues were. I found this unnecessarily and seductively interesting. I looked up h scores for professors, some fellow senior lecturers, and some newly recruited members of staff. Through various links, I could see who had collaborated with who. I could also see which of the professors were really high ranking professors.
I then stopped my searching. I asked myself another question, which was: “does any of these numbers really matter?”
It takes different kinds of people to run an academic department and a university. Some colleagues are excellent at research. Some colleagues are excellent at teaching, and some colleagues are even excellent at administration. In my own role as a staff tutor, I do a lot of academic administration and quite a bit of work which could be viewed as teaching. This means that I don’t have a lot of time to do any research. Broadly speaking, central academic staff have a much higher h-index metric than staff tutors, simply because they have more time. What research I do carry out is often applied research. This research can sometimes be labelled as scholarship, which can be considered to be research about the practice of teaching and learning.
It was interesting that one of the important points I took away was that societal impact can’t be directly measured through bibliometrics. I also found it interesting that different types of articles attract a greater number of citations. One of my biggest academic hits (I don’t have very many of them) has been a review paper, where I studied different ways in which a computer could be used to assess the quality of computer programming assessments. The articles that I have published that relate to pure research have certainly attracted less attention.
All this comes back to a broader question: in academia, what is valued? I think the answer is: different types of work is valued. Pure research is valued alongside effective and engaging teaching. The bit that ties the two together is, of course, scholarship.
Bibliometrics are, in my eyes, are a set of measurements that attempt to quantify academic debate. It only ever tells a part of a bigger and much more complicated story. Metrics are not, in my opinion, surrogates for quality. Also, academic fashions and trends come and go.
Acknowledgements
Many thanks are given to Chris Biggs for running such an engaging and interesting session. Many of the links shared in this article were shared during Chris' presentation.