Could scientists use Altmetric.com scores to predict longer term citation counts?

TL;DR

This study evaluates whether early Altmetric.com scores, especially Mendeley reader counts, can reliably predict long-term citation impact and reflect non-scholarly influence across various fields.

Contribution

It demonstrates that Mendeley reader counts are the most consistent early predictors of future citations, and highlights the complementary role of Altmetric scores alongside journal impact factors.

Findings

Mendeley reader counts reliably predict future citation impact.

Other Altmetric scores can predict impact in specific fields.

Altmetric scores also reflect non-scholarly impacts in some disciplines.

Abstract

Altmetrics from Altmetric.com are widely used by publishers and researchers to give earlier evidence of attention than citation counts. This article assesses whether Altmetric.com scores are reliable early indicators of likely future impact and whether they may also reflect non-scholarly impacts. A preliminary factor analysis suggests that the main altmetric indicator of scholarly impact is Mendeley reader counts, with weaker news, informational and social network discussion/promotion dimensions in some fields. Based on a regression analysis of Altmetric.com data from November 2015 and Scopus citation counts from October 2017 for articles in 30 narrow fields, only Mendeley reader counts are consistent predictors of future citation impact. Most other Altmetric.com scores can help predict future impact in some fields. Overall, the results confirm that early Altmetric.com scores can predict later citation counts, although less well than journal impact factors, and the optimal strategy is to consider both Altmetric.com scores and journal impact factors. Altmetric.com scores can also reflect dimensions of non-scholarly impact in some fields.