The Journal of Social Psychology | 2021
Making data meaningful: guidelines for good quality open data
Abstract
In the most recent editorial for the The Journal of Social Psychology (JSP), J. Grahe (2021) set out and justified a new journal policy: publishing papers now requires authors to make available all data on which claims are based. This places the journal amongst a growing group of forward-thinking psychology journals that mandate open data for research outputs. It is clear that the editorial team hopes to raise the credibility and usefulness of research in the journal, as well as the discipline, through increased research transparency. As this editorial appeared, we had a paper accepted for publication in Behavior Research Methods (Towse et al., 2020) that reported empirical data on open data practices across psychology. Between 2014 and 2017, we found that public data sharing was uncommon (less than 4% of empirical papers; see Hardwicke et al., 2020, for similar data across the social sciences). We also observed that when data was publicly shared, the majority of datasets were incomplete and had limited reusability. Nearly half were at risk of being orphaned due to the lack of a permanent link between data and research paper (for similar dataset quality issues, see also D. G. Roche et al., 2015; Hardwicke et al., 2018). Although the time period for our study already might appear distant there is evidence that, despite researchers being encouraged to include open data, the inclusion and quality of datasets remains disappointingly low. For example, of 5,905 published articles on COVID19, only 13.6% shared their data and only 1.2% shared their data in nonproprietary format such as .csv (Lucas-Dominguez et al., 2021). This commentary represents a natural and complementary alliance between the ambition of JSP’s open data policy and the reality of how data sharing often takes place. We share with JSP the belief that usable and open data is good for social psychology and supports effective knowledge exchange within and beyond academia. For this to happen, we must have not just more open data, but open data that is of a sufficient quality to support repeated use and replication (Towse et al., 2020). Moreover, it is becoming clear that researchers across science are seeking guidance, training and standards for open data provision (D. Roche et al., 2021; Soeharjono & Roche, 2021). With this in mind, we outline several simple steps and point toward a set of freely available resources that can help make datasets more valuable and impactful. Specifically, we explain how to make data meaningful; easily findable, accessible, complete and understandable. We have provided a simple checklist (Table 1) and useful resources (Appendix A) based on our recommendations, these can also be found on the project page for this article (https:doi.org/10.17605/OSF.IO/NZ5WS). While we have focused mostly on sharing quantitative data, much of what has been discussed remains relevant to qualitative research (for an indepth discussion of qualitative data sharing, see DuBois et al., 2018).