Citation theories and their application to altmetrics

2015-10-16T13:59:28Z (GMT) by Rodrigo Costas Stefanie Haustein

Researchers are increasingly pressured to provide evidence that their work has impact—within the scientific community, on the economy or on society at large. Altmetrics have been suggested as indicators of various forms of impact aiming to replace citations as the sole measure of academic success. The acts on which various altmetrics are based are, however, quite heterogeneous: likes on Facebook, mentions on Twitter, saves on Mendeley, and expert recommendations on F1000 are acts that differ in terms of user community, engagement, motivation and audience. While the act of citing has been an essential part of the scholarly communication process since the beginning of modern science, it is unclear whether the acts that altmetrics are based on are relevant in scholarly communication. While empirical studies have shown that most altmetrics correlate weakly with citations—suggesting fundamental differences between these various metrics—conceptual discussions about their meaning are rare, leaving it unclear what is actually measured. We argue that the heterogeneity of user communities, levels of engagement, motivations, as well as of the audiences to which the various metrics are associated affects their respective meaning.

In order to go beyond correlation analyses and better understand the meaning of altmetrics, we discuss acts on social media from a theoretical and conceptual perspective using citation theories, chosen because of the analogy that is often made between altmetrics and citations. By discussing various acts on social media from the perspective of the normative, social constructivist, and concept symbols citation theories (Haustein, Bowman, & Costas, 2015), we show that they are more or less suitable, and that the level to which they can apply depends on the particular act considered. For example, Merton’s ethos of science can help to explain the mechanisms behind reviewing and recommending on F1000 and (to some extent) blog citations and Mendeley readership. Users on Twitter and Facebook are less likely to adhere to these norms, as can be reflected by the important tweeting intensity associated with papers that have funny titles, focus on curious topics or to retracted publications. Social constructivist theories, and in particular the Matthew effect, seem applicable to most acts on social media due to their networked nature; explaining the concentration of events on a few prominent papers or actors on social media. The concept symbols theory was helpful to explain mentions on Twitter, for example with respect to hashtags.

Theoretical discussions will be supported by concrete practical examples to begin debates surrounding the development of a framework that accommodates the broad and diverse spectrum of acts recorded by different altmetric sources.