The relationship between usage and citations in an open access mega journal

TL;DR

This study investigates how early usage and subject area influence citation counts in the open access journal Scientific Reports, revealing moderate correlations and early usage's impact on citation timing and volume.

Contribution

It provides empirical analysis of usage-citation relationships across disciplines in a large open access journal, highlighting early usage as a predictor of citation impact.

Findings

Moderate correlation between early usage and subsequent citations

Early high usage linked to earlier and higher citations

Differences observed across scientific subjects

Abstract

How do the level of usage of an article, the timeframe of its usage and its subject area relate to the number of citations it accrues? This paper aims to answer this question through an observational study of usage and citation data collected about the multidisciplinary, open access mega-journal Scientific Reports. This observational study answers these questions using the following methods: an overlap analysis of most read and top-cited articles; Spearman correlation tests between total citation counts over two years and usage over various timeframes; a comparison of first months of citation for most read and all articles; a Wilcoxon test on the distribution of total citations of early cited articles and the distribution of total citations of all other articles. All analyses were performed using the programming language R. As Scientific Reports is a multidisciplinary journal covering all natural and clinical sciences, we also looked at the differences across subjects. We found a moderate correlation between usage in the first year and citations in the first two years since publication, and that articles with high usage in the first 6 months are more likely to have their first citation earlier (Wilcoxon=1811500, p < 0.0001), which is also related to higher citations in the first two years (Wilcoxon=8071200, p < 0.0001). As this final assertion is inferred based on the results of the other elements of this paper, it requires further analysis. Moreover, our choice of a 2 year window for our analysis did not consider the articles' citation half-life, and our use of Scientific Reports (a journal that is atypical compared to most academic journals) as the source of the articles analysed has likely played a role in our findings, and so analysing a longer timeframe and carrying out similar analysis on a different journal (or group of journals) may lead to different conclusions.