Written by Marcel LaFlamme With the latest release of the PLOS Open Science Indicators (OSI) results, we are introducing a new indicator…
Open Science Indicators: How might we start thinking about the leading adopters of Open Science practices?
Author: Lauren Cadwallader, Open Research Manager, PLOS
The latest version of PLOS’ Open Science Indicators (OSI) dataset has been released, extending to six and a half years’ worth of data. The dataset now covers 141,864 articles published from January 1st 2018 to June 30th 2024, offering insights into Open Science practices over time for both PLOS publications and a set of comparable articles from other publishers.
Figure 1: Open Science Indicators for PLOS and comparator content by publication quarter 2018-Q2 2024. Rates are given as a proportion of all published articles.
This release includes the latest results for data and code sharing, and preprint posting. For preliminary rates of study registration, and protocol sharing, check out the June 2024 and September 2023 releases, respectively.
As well as tracking changes over time, the dataset allows us to focus on different Open Science practices and how these vary by discipline, geography, or between PLOS and other publishers. But it also allows us to delve into how different Open Science practices may be connected. In the rest of this post we demonstrate some ways in which the OSI data can be explored with a view to multiple Open Science practices, raising some questions and possible interpretations. We would love to hear your thoughts on this post or how you have used the OSI dataset to explore Open Science practices.
Are researchers sharing everything they potentially could?
One important aspect of the OSI dataset is the determination of whether an article generates an Open Science output. All articles could generate a preprint but not all (96%) of PLOS articles generate data, and 37% generate code – as determined by DataSeer’s algorithms. Understanding whether an article generated an output means that we can put the sharing of outputs into context. In this blog we start to explore sharing rates in the context of generation rates, as well as possible correlations between different Open Science practices.
If we only look at practices individually then we get one view of how Open Science is being carried out. For example, Figure 2 below looks at how code, preprint and data sharing rates change in 2 year increments for PLOS articles.
Figure 2: Proportion of articles generating and sharing code, data or preprints for PLOS articles in 2018, 2020, 2022 and 2024 (first 6 months only). The segment of code and data that fall into the “not shared” category includes articles that cannot share code/data for ethical or legal reasons but may make it available via access requests.
If we consider how many outputs an article is sharing compared to how many were generated, then a different view of Open Science practice emerges. In Figure 3 we can observe that for PLOS, the percentage of articles sharing all outputs that could be shared remained the same in 2022 to 2024, while the percentage of articles sharing at least one of these three outputs has grown. This finding is not evident for the PLOS results in Figure 2, which show very little change in sharing rates of code and data, for those generating it, and preprints.
Figure 3: Proportion of articles sharing all, some or none of the Open Science outputs generated by the article for PLOS and Comparators in 2018, 2020, 2022 and 2024 (first 6 months only).
Comparators on the other hand show growth over time in both the segment of articles sharing everything that could be shared, and the segment sharing at least one output.These rates are still lower than the rates at PLOS – and the PLOS rates also have room for growth – suggesting that many opportunities remain for publishers to support researchers in carrying out Open Science practices.
Who are the leading adopters of Open Science practices?
There is an interesting segment of articles in the OSI dataset, which correspond to what one could call “leading adopters” of Open Science practices: those sharing all the outputs that could be shared, as measured by OSIs (14% of articles published in the first half of 2024). Taking a deeper look at the articles that generate all three outputs (preprints, data and code), what proportion of these share, and how this changes over time also reveals some patterns. Figure 4 below shows this for PLOS articles in 2018, 2020, 2022 and 2024. Since 2018:
- The percentage of articles sharing only data and code, which the dataset can also reveal are highly correlated, has remained relatively stable
- The percentage of articles just sharing data and preprints increased between 2018 and 2020 but then stays relatively stable
- The percentage of articles sharing all three generated outputs increased from 2018 to 2020 and again to 2022
- The percentage of articles sharing only one output has decreased from 2018 to 2020 and again to 2022
- The percentage of articles not sharing any outputs has remained relatively stable
Figure 4: Breakdown of the types of outputs shared, for PLOS articles generating data, code and preprints, by year.
What might this be telling us? What questions does it raise? Researchers are sharing more outputs over time, yes, but we already know this. The more interesting questions are: how and why are researchers adopting more Open Science practices? Have there been lasting effects of global events since 2018 on sharing practices (preprinting, for example, accelerated during the pandemic and has since plateaued)? Or is it a case of incremental adoption? The way in which adoption of practices spreads, how different practices relate to one another, and differ across disciplines and geographies are all interesting topics for those of us who are championing the adoption of Open Science practices.
The observations above are preliminary and based only on quantitative data, but combining them with a deeper understanding of the causes of these differences as well as the effects of Open Science practices could reveal more opportunities to support authors in becoming leading Open Science adopters. These are questions we are continuing to explore at PLOS, and we welcome feedback from and collaboration with others who are doing the same.