Evaluating altmetrics

The rise of the social web and its uptake by scholars has led to the creation of altmetrics, which are social web metrics for academic publications. These new metrics can, in theory, be used in an evaluative role, to give early estimates of the impact of publications or to give estimates of non-traditional types of impact. They can also be used as an information seeking aid: to help draw a digital library user's attention to papers that have attracted social web mentions. If altmetrics are to be trusted then they must be evaluated to see if the claims made about them are reasonable. Drawing upon previous citation analysis debates and web citation analysis research, this article discusses altmetric evaluation strategies, including correlation tests, content analyses, interviews and pragmatic analyses. It recommends that a range of methods are needed for altmetric evaluations, that the methods should focus on identifying the relative strengths of influences on altmetric creation, and that such evaluations should be prioritised in a logical order.