Confounding cross-correlation: A meta-analysis of the association between citations and altmetrics

This paper presents initial results from an ongoing meta-analysis of the correlation between altmetrics and citations. Since the Altmetric Manifesto was published in 2010, a large number of studies have examined altmetric-citation associations, resulting in a diverse array of observations. Very few studies have attempted to clarify this diversity with a meta-analysis. As such, we collect a large number of existing studies quantifying the correlation between altmetrics and citations, which for the first time will facilitate a meta-regression to establish how study characteristics such as discipline, non-zero values, and citation sources moderate the altmetric-citation relationship. To date, we have identified relevant literature examining altmetric-citation correlations in the Web of Science and then reviewed this literature against several inclusion criteria before extracting the correlation coefficient, sample size, and several other variables from each study. The final sample for the meta-analysis consists of 111 studies with a total of 914 correlation coefficients. Initial results show that the strength of the correlation with citations varies substantially by the source of altmetric data but also within each altmetric source. The meta-analysis strives to explain this variation and results of the moderating study characteristics will be presented during the conference.