Ranking Journals Using Altmetrics

The rank of a journal based on simple citation information is a popular measure. The simplicity and availability of rankings such as Impact Factor, Eigenfactor and SciMago Journal Rank based on trusted commercial sources ensures their widespread use for many important tasks despite the well-known limitations of such rankings. In this paper we look at an alternative approach based on information on papers from social and mainstream media sources. Our data comes from altmetric.com who identify mentions of individual academic papers in sources such as Twitter, Facebook, blogs and news outlets. We consider several different methods to produce a ranking of journals from such data. We show that most (but not all) schemes produce results, which are roughly similar, suggesting that there is a basic consistency between social media based approaches and traditional citation based methods. Most ranking schemes applied to one data set produce relatively little variation and we suggest this provides a measure of the uncertainty in any journal rating. The differences we find between data sources also shows they are capturing different aspects of journal impact. We conclude a small number of such ratings will provide the best information on journal impact. Conference Topic Altmetrics The background and purpose of the study Journal metrics, such as the Thomson Reuters Journal Impact Factor, were originally developed in response to a publisher need to demonstrate the academic attention accorded to research journals. Over the intervening 50 years since Garfield’s work in the field, the Impact Factor and other metrics, such as Eigenfactor (Bergstrom, 2007), have been used and misused in a variety of contexts in academia. An oft-discussed perception is that a journal-level metric is a good proxy for the quality of the articles contained in a journal. In the evaluation and bibliometrics communities citation counting is generally understood not to be an appropriate proxy for quality but rather a measure of attention. The type of attention being measured in this case is quite specific and has particular properties. What is being measured is the attention to a paper of peers in related fields. The bar for registration of this attention is relatively high – the researcher or researchers making the citation must deem the target article to be of sufficient value that they include a citation in a work of their own that in turn is deemed publishable (e.g. see Archambault & Lariviére, 2009, and references therein). The timescale associated with citations is also long – typically being limited by the review and publication process associated with particular fields. Additionally, it is accepted that journal-level metrics say little regarding the merit of particular articles in the journal since journal-level metrics are often calculated based on thousands of articles and are often biased by the performance of the tails of the distribution of citations. These realisations have led to the recent growth in popularity of article-level metrics or altmetrics. Altmetrics have broadened the range of types of attention that we can measure and track for scholarly articles. Mostly based in social and traditional media citations, the altmetric landscape is one that is constantly changing with the introduction of different data sources all the time. While, one the one hand, altmetrics suffer from all the unevenness of traditional citations, they occur over different timescales, which provides us with a more nuanced view