The relationship between open access publishing and referencing

This paper reinforces the necessity of normalizing OA adoption by discipline and country, as suggested and practiced in earlier works. Using such measures, we can identify countries for which OA has become characteristic of their publications, with OA papers informing their research as well as serving as an avenue to share their research with the world. There is, however, a danger to this normalization in that it creates both a methodological and conceptual expectation: that is, that disciplines with lower OA rates are expected to be and remain lower.

Aggregating countries by income categories also can create some distortion as there are large differences in GNI per capita within groupings. Using the GNI measure in OA studies has been criticized as GNI is not an accurate measure of internet access in a country; ICT development is arguably more related to OA access than GNI. Additionally, it has been noted that these measures both neglect the science communication practices and contexts in the countries and regions. Thus, even if we observe different behavior regarding types of OA used in lower middle- and low-income countries, other factors in these countries could be a larger contributor to the observations made.

Despite these limitations, several key contributions are made by the present analysis. We show that, globally, scholars are slightly more likely to publish OA papers than reference OA papers. This may, of course, be explained by the trend towards increasing OA over the last decade. That is, there are more contemporary OA publications, which would suggest that the OA status of references would lag. However, we find that scholars draw from OA references more than would be expected, which reinforces the citation advance of OA papers. In short, papers are increasingly OA, and scholars reference OA papers to a higher degree than would be expected; however, the proportion of OA referencing remains lower than contemporary OA publishing given the time it takes for citations to accumulate.

Marked differences are observed by disciplines and countries. For example, Law and Education are more likely to publish in OA than reference it. It is notable that these disciplines have slightly different scholarly communication practices and a strong orientation towards practice. This may suggest an intentionality about widening dissemination of their work to practitioner audiences that might not necessarily have access. Domains like medicine, health, and biological sciences draw more heavily from OA than they publish it. This could be problematic in that it suggests that the demand for openly available medical knowledge is higher than the supply. However, it could also be an artefact of the expansive OA availability of papers funded by the NIH, supported by the infrastructure of PubMed and enforced through policy mandates.

Publication behaviour, more so than referencing behaviour, is influenced by OA mandates and institutional policies. Just as institutions within countries can specialize in research areas, the science system within a country can be particularly supportive of OA practices. This includes initiatives to transform local journals into OA journals. Platforms like SciELO in Brazil may explain the higher rates of OA publishing, which is not matched by high levels of OA referencing. The publish-or-perish culture and the perception that OA journals have a faster turnaround time is also thought to drive the increase of publications in OA journals in academic systems that reward a high volume of publications. This potentially accounts for the more extreme NOAPub values as opposed to NOARef values observed. The trend towards high OA publishing and referencing for green-only and gold OA in Hungary may reflect a concerted effort to increase participation in OA publishing. Several countries display close to expected values for green-only and gold OA which may reflect a cultural acceptance for this mode of communication. China represents the other end of the spectrum with lower-than-expected levels of both publication and OA referencing for each type of OA. Such pattern is likely due to an emphasis on the journal impact factor and other journal-level indicators, which drives scholars towards closed publication environments, as well as to the high degree of country-self citation. On the whole, this paper shows that OA publications exceed OA referencing and that the success of OA dissemination of a country is not limited to their OA mandates and policies, but are affected by the entire research evaluation system.