Science Cultures: The Uptake of Open Science and RRI in Academia
by Thomas Klebel, Nancy Pontika, Antonia Correia, David Pride, Petr Knoth, Tony Ross-Hellauer
Scientific knowledge is a key resource for achieving societal and economic goals. Open Science (OS) and Responsible Research and Innovation (RRI) promise to fundamentally transform scholarship to bring greater transparency, inclusivity and participation to research processes, and increase the academic, economic and societal impact of research outputs. These form a cross-cutting agenda that stands to contribute to most of the UN’s Sustainable Development Goals (SDGs), and are a central pillar of the European Commission’s Digital Single Market strategy. Yet access to scientific products and processes is not made uniform simply because they are made available via the Internet. How equitable is the implementation of OS and RRI across a range of stakeholder categories, and in particular for those at the peripheries? Might RRI interventions, in some cases, actually deepen socioeconomic inequalities (the digital divide) and be in conflict with wider sustainable development goals? How do geographical, socio-economic, cultural and structural conditions lead to peripheral configurations in the European knowledge landscape? What factors are at play and what can be done (at a policy level) to foster absorptive capacity and enhance OS/RRI uptake and contributions to scientific production across regions?
Such questions lie at the heart of ON-MERRIT. The work conducted in ON-MERRIT’s work package 3 “Research cultures, support and incentives” (WP3) provides rich evidence to answer these and further questions. WP3 included several tasks which aimed to study how the application of RRI and Open Science policies, principles and practices affect research, researchers and their careers across diverse contexts. In this blog post we share the results of our research activities and describe how they relate to the project’s wider aims.
Datasets on Open Science and RRI
Our first task was to gather multiple sets and sources of data, which could then be analysed and shared with the research community. We focused on two specific approaches:
A dataset on Promotion, Review and Tenure policies (PRT-policies) from institutions in seven countries (Austria, Brazil, Germany, India, Portugal, United Kingdom and the United States). The respective policies were collected by the study team and subsequently coded for a range of traditional (such as the length of one’s CV or metrics related to the journals one has published in) and non-traditional indicators (such as the rate of Open Access (OA) publishing, the sharing of research data and code, engagement with the public and citizen science). The full analysis was undertaken in ON-MERRIT WP6, and recently made available in our report D6.1 Investigating Institutional Structures of Reward & Recognition in Open Science & RRI.
A corpus of scholarly research outputs processed from Microsoft Academic Graph (MAG) and CORE, a large aggregator of Open Access (OA) content.
A full description of the methodology involved in collecting these datasets is available in Deliverable Report D3.1 “RRI and Open Science Datasets” on Zenodo. We also intend to publish the data itself via a data publication in the near future.
Cumulative Advantage in Open Science and RRI
Dynamics of cumulative advantage run deeply in science: initial success leads to beneficial circumstances later on. Multiple mechanisms related to this dynamic have been identified, such as the misallocation of rewards towards better known scientists (termed “Matthew Effect” by Robert K. Merton), increased chances to receive further funding after initial funding has been secured, and persistent under-representation of women in general. In our work, we wanted to explore how OS and RRI are related to these dynamics, whether they mitigated or potentially worsened them. To this end, we conducted four quantitative research studies addressing a range of research questions, including: Who produces and who consumes open access research literature? How is institutional performance related to the application of RRI policies and OA publishing? Does the uptake of OA publishing change existing hierarchies within academic publishing, and if so, in which ways across a subset of ON-MERRIT’s target UN Sustainable Development Goals (SDGs) - SDG 2 - Zero Hunger, SDG 3 - Good Health and Well-Being and SDG 13 - Climate Action?
Regarding our first question, we investigated levels of production and consumption of Open Access (OA) research literature globally, measured as the proportion of citations to OA literature, and tested for correlations with gross domestic product (GDP) per capita at the country and continental level. We find moderate correlations between OA production and OA consumption: institutions whose researchers publish more OA also cite more OA. Although we initially expected that countries with lower GDP (per capita) would have a higher OA consumption than production, we did not find any such correlation.
Following ON-MERRIT’s focus on three key UN Sustainable Development Goals ((SDG Zero Hunger (SDG 2), SDG Health/Well-Being (SDG 3) and SDG Climate Action (SDG 13)), we also investigated aspects of OA publishing across papers relevant to the three SDGs. Our analysis showed that well-resourced actors publish more frequently OA in the SDG areas, as well as publishing in journals with on average higher APCs, which might worsen already existing structural hierarchies within academia.
In the figure below, we display the relationship between a measure of institutional prestige (P top 10%, i.e. how many publications among the top 10% cited publications an institution has, according to the Leiden Ranking) and the average APC of journals where researchers from these institutions publish. More prestigious institutions publish more frequently in journals with higher APCs, especially in SDG Zero Hunger (SDG 2). The effect is consistent, regardless of whether institutions are assigned by first (A) or last (B) author. Importantly, this issue is actually worsening, as the gap between higher and lower ranked institutions in average APCs paid has been increasing, particularly in SDGs 2 (Zero Hunger) and 3 (Health/Well-Being).
Figure 1: Median journal APC fees for articles published by authors from institutions in the Leiden ranking. (A) First authors only (B) Last authors only. P top 10% refers to the number of a university's publications that are within the top 10% cited of their field. The four groups (p0-25, etc.) are quartiles within the distribution of P top 10%.
The research conducted in this research stream highlights that it is the higher ranked, more prosperous and more prestigious institutions that appear best able to adopt, adapt to, and benefit from the evolving landscape of Open Access publishing. Moreover, these trends hold true over time, on the global level, and when broken down to individual continents and subject areas (SDGs). We therefore have to conclude that structural inequalities in current academic publishing are not necessarily remedied by the Open Science movement, with specific trends such as APC-driven OA publishing potentially exacerbating dynamics of cumulative advantage. Importantly, if research on key global issues is only driven by well-resourced actors, it risks being oblivious to challenges faced by societies and communities less embedded into the global production of knowledge.
The Uptake of Open Science and RRI in Policy and Training
Training and skills are key aspects in the uptake of Open Science and Responsible Research and Innovation practices. In our research we aimed at understanding current institutional structures for OS/RRI training and their relation to current levels of adoption of OS/RRI practices. To this end, we conducted:
An international survey to assess their practices and opinions regarding OS/RRI, as well as the institutional support for these practices
In-depth interviews with representatives responsible for training provision in 11 institutions across three continents to identify the support, drivers and barriers to OS/RRI from an institutional point of view.
For our survey, we received responses from 167 active researchers, with a slight skew towards more senior, male academics from Europe, working in the fields of Social Sciences, Engineering and Technology and the Natural sciences. Concepts of OS/RRI, such as Open Access publishing, Research Data Management, Reproducible Research, and Citizen Science, are looked upon favourably, as a different way to do research, although some reluctance and doubts were present on how to put them into practice. Research Integrity and Gender are the topics that gathered more consensus and are best implemented.
Interestingly, we found that the majority of our survey participants had not received training in any of the above OS/RRI topics (very often a large majority). The most popular topics for those who had received training were OA publishing, Research Integrity, Gender, Open Data and OS in general.
Figure 2: How many training events have you attended on these topics?
In our second study, we interviewed representatives from eleven higher education institutions responsible for implementing OS and RRI policies, from nine countries and in senior positions. Although in general researchers at these institutions are aware of and show some familiarity with OS, OA, RRI and FAIR data concepts, misunderstandings remain. Awareness differs across disciplines and levels of seniority and does not necessarily translate into practice. The main barriers to OS/RRI uptake which were mentioned by our interviewees are the lack of incentives, awareness, and time; concerns of sharing and legal issues; lack of infrastructure and services; cultural/behavioural issues; lack of funding/resources; lack of central coordination within and between institutions; and the need for OS/RRI policies. Interviewees advised conversations regarding reform on research assessment criteria and publications metrics were underway.
Training support services for OS/RRI available at the institutions is focused on institutional services or tools, RDM (data management planning, GDPR and data sharing), Data protection and handling of sensitive data, OA, Open Science funder requirements, Research Integrity and Ethics. The main challenges faced in providing training are ensuring participation from diverse audiences; lack of staff; lack of institutional support (funding or central coordination); and tailoring training to audience needs. These issues may be mitigated through the integration of training in course curricula, the development of specific and more practical training, and the improvement of online materials.
These results highlight the difficulties involved in providing OS/RRI training and support services at the institutional level and reiterate the fact that training in OS/RRI is essential for researchers to be able to perform science in a solid and transparent way and comply with most funder’s requirements and mandates worldwide. There is a need for skilled professionals and the development, normalization and integration of OS and RRI into curricula. In addition, the role of communities in reinforcing practices and promoting a real cultural change must be fully embraced. More work should be undertaken to foster interoperable infrastructures, integrated training resources and peer-to-peer training, as well as increased resources for training staff and infrastructure.
The work conducted in this WP constitutes an important step to better understand dynamics of inequality in academia and their relationship to OS and RRI. We collected relevant data on OS/RRI policies, conducted multiple large-scale quantitative studies, as well as an international survey and comprehensive interviews. Based on this evidence-base, we developed a set of preliminary recommendations, such as the following.
Support less prestigious institutions in building OS and RRI capacity and awareness. Our results show that those who produce more OA tend also to benefit more, hence it is important to create a level playing-field and close the significant difference between institutions that we observed.
Encourage publishers to waive APCs, not just for institutions in less-developed countries, but also for less prestigious institutions in more developed countries. This follows our observation that the less prestigious institutions are those that are on average slower in investing into and therefore also reaping the benefits of OA.
Support the creation of responsible metrics for OS practices that can be obtained at the granularity of individual researchers and that are related to the rigour of their scholarship. While open peer review, reproducibility, data citations, considering the meaning and motivation of citations, etc. all provide promising avenues in this direction, assessing and understanding how these are and should be linked to academic success is still in its infancy.
Research performing organisations should:
Regularly collect and make publicly available RRI data, ideally at the level of faculties/departments. The fact that we were only able to obtain RRI data at a country level and that they are not regularly collected, seriously limits the ability of monitoring progress towards RRI and reduces the potential of research into its benefits.
Publicly and annually release information about how much they spend for APCs and subscriptions to academic literature.
Dedicate, via their academic libraries, a set proportion of their budgets for the support of a range of open, shared and not-for-profit scholarly infrastructures. These could include but are not limited to funding of non-APC consortial “diamond” models of OA where neither author or reader must pay directly, promoting library publishing to combat APC stratification, supporting open citation, and new, emerging and existing scholarly scholarly data infrastructures.
Deposit their Author Accepted Manuscripts (AAMs) into repositories. This is especially important in situations where the researcher, or their institution, cannot afford to pay an APC. A strong and somewhat selfish incentive to researchers should be that by making their research OA, they are primarily helping themselves in gaining cumulative advantage over their peers who don’t. As an (important) side-effect, they also benefit other researchers, professionals and the general public.
Try to actively avoid biases, including being aware of the risk of unconscious biases, in decisions including but not limited to the selection of publication venue and institutional prestige or location of collaborating partners. Researchers should aim to take these decisions on merit, to limit the effects of cumulative advantage we studied.
As ON-MERRIT moves into its final months, we are currently working hard on further developing the recommendations via co-creation activities involving researchers, funders, and research support staff. The final recommendations will be shared with the OS/RRI community in early 2022.