Open science reforms have contributed to a more rigorous and robust psychological science, but there is still much to improve. “Current best practices aim to improve transparency via methodological and statistical practices. But we aren’t covering many other important norms”—those that can further improve the science’s quality—according to Janet Taylor Spence Award recipient Eiko Fried (Leiden University) in his introduction to the symposium Open Science 2.0: Improving Psychological Science by Focusing on Diversity, Grassroots Communities, Education, and Theory Formation at the 2023 APS Annual Convention in Washington, D.C.
Fried singled out two norms that open science reforms may have overlooked so far:
- Communalism: the common ownership of scientific discoveries, enabling all scientists to access and contribute to them equally.
- Universalism: the focus on logic and arguments rather than attributes of scientists.
Incorporating these norms into open science reforms is fundamental to creating a psychological science that is truly open, transparent, and equitable, Fried said.
Four researchers followed Fried’s introduction with findings and tools in support of these norms.
Sakshi Ghai (University of Cambridge) talked about the role of diversity in open science practices, which can both hinder and contribute to more globally representative science. For instance, most people in the world live in the Global South, but only a minority psychological research features researchers and participants from the Global South. One issue that must be addressed to overcome this gap is how diversity in science is defined: Diversity is “not just diverse samples, methods, and researchers, but we also need to be asking culturally informed research questions that are relevant to diverse populations,” Ghai explained.
Further, intersectional identities mean that human diversity itself is on a continuum, with different identities intersecting and overlapping. Thus, the concept of diversity must be expanded beyond the Western perspective (e.g., in India, caste might be a category of diversity more salient than race). “We need to urgently ask scholars from different diverse groups to weigh in and help us develop a more nuanced understanding of diversity,” Ghai said.
“Diversifying any scientific field is an iterative process involving many stakeholders—publishers, donors, institutions, researchers, and the next generation of scholars.” Sakshi Ghai
Alexandra Sarafoglou (University of Amsterdam) highlighted the important role of Open Science Communities (OSCs) within academic institutions. To change the way researchers do their work, Sarafoglou argued that the field needs a “strategy of culture change” that encompasses (1) reaching a critical mass of researchers who embrace open science; (2) establishing dialogue across disciplines; (3) engaging researchers who are not yet doing open science; and (4) bridging the gap between researchers and policymakers. OSCs enable the research community to be involved and engaged in developing and implementing such a strategy. OSCs’ purpose is to accelerate the normalization of open science practices among researchers by making OSC members’ practices visible and accessible and supporting researchers transitioning to open science. OSCs also work to shape local and national policies surrounding open science. Since the first OSC was established in 2018 at Utrecht University, every university in the Netherlands has established its own OSC, and the international network of OSCs includes countries such as Sweden, Portugal, Serbia, and Nigeria (see more information here). (Readers interested in creating an OSC may want to check out the OSC Starter Kit.)
Education is also needed to improve open science practices. Flavio Azevedo (University of Groningen) focused on the Framework for Open Science and Reproducible Research Training (FORRT), which provides a pedagogical infrastructure and open educational resources to support the teaching and mentoring of open science. Started by graduate students at the Society for the Improvement of Psychological Science in 2018, FORRT currently counts more than 850 scholars in different academic fields and disciplines. It partners with the major open science organizations to foster the adoption of open science practices in higher education, reframe open science as inclusive scholarship, advocate for greater recognition of educational resources (in addition to research), and educate on epistemic uncertainty, plurality, humility, and research integrity and credibility.
Azevedo highlighted that FORRT leverages big-team science and large-scale collaboration to create and curate educational materials (e.g., ready-to-use Open Science lesson plans, a consensus-based glossary of over 250 Open Scholarship terms, a review of the impact of Open Scholarship on students’ outcomes, a review on how research structures, procedures, and communities have changed in response to the replication crisis in psychological science, and much more), promote neurodiverse voices and scholarship in Open Science, and advocate for more space for Latin America research in psychological science. Several current initiatives aim to democratize access to open scholarship discussions, empower underrepresented researchers, and make science more equitable.
Although open science reforms have focused primarily on empirical research, there is also a need to improve how we develop and evaluate psychological theories. Donald Robinaugh (Northeastern University) explained how formalizing theories as computational or mathematical models might improve their quality and strengthen their relationship with empirical research. Most theories in psychology are expressed only in words and therefore are often too imprecise to make clear and precise predictions that can be rigorously tested, making room for questionable research practices and limiting what can be learned from empirical research. The use of models allows researchers to simulate precisely what a theory predicts and compare those predictions to the experiment’s actual results. If the simulated and empirical results match, researchers can be much more confident in their theories; if they do not match, researchers can discard the theory or learn how to improve the theory.
Robinaugh also explored how the use of computational models supports the norms Fried mentioned in his introduction: communalism (scientists can have access to, interrogate, use, and improve models) and universalism (because it is models, not experts, that determine what the theory predicts, anyone with the model can determine those predictions for themselves).