Article downloads and citations: is there any relationship?

It is frequently assumed and perceived that a high number of downloads reflects great interest in scientific publication, which would also be reflected by an increased number of citations in the short and the medium term. In a previous analysis, Davis et al. showed that article citation counts gave no citation advantage for open access articles [1], thus raising some doubts as to whether the interest in the readership is really mirrored by citations of individual contributions in further articles. Even more interestingly, controversial data were provided on the plausible link between downloads and citations, wherein Jahandideh et al. found that more downloads at a limited period of time is an indicator of more citations [2], whereas Appell reported a very poor correlation between number of downloaded full papers (both PDF and HTML) and number of citations (r = 0.29) [3]. In an earlier evaluation by one of us, the relationship between citations and downloads was not clear cut [4]. The suspicion here is that downloads reflect broader interest, comprising that from other researchers who may then cite the work, as well as end user practitioners who may utilise the work for their practice, but who are unlikely to publish material themselves [5]. After all, the number of downloads in practice reflects many-fold excesses of published works.