Ä¢¹½ÊÓƵ

Meet the Senior Data Editor team

is an Assistant Professor at Wageningen University & Research in Wageningen, the Netherlands. He studies fungal biology, particularly the factors that give rise to individuality in these organisms, and how this affects social interactions.

is a Quantitative Marine Ecologist at the Oregon Department of Fish and Wildlife. He uses statistical and simulation-based mathematical models with data at multiple scales ranging from animal behaviour to population ecology and ecosystem dynamics to study conservation-related questions.

is a Senior Scientist in Data and Modelling within the Applied Aquatic Ecology team at the Arthur Rylah Institute of Environmental Research in Melbourne, Australia. His research uses acoustic telemetry and quantitative modelling to understand how aquatic organisms move and respond to environmental change, from pharmaceutical pollution to altered river flows.

What we check

Proceedings B , so that reviewers and Data Editors can see the evidence behind a paper while it is being assessed. Our job as Data Editors is to check that this archive is complete and usable. Authors submit their data, code, and a README file with the manuscript, usually as a repository link. Papers are sent to a Data Editor when an Editor or reviewer flags them, typically at revision. The journal currently checks at least one in five accepted papers, but aims to eventually check all accepted submissions.

Our checks generally cover the following things:

• Data re-use: the data are in a repository, in a machine-readable format such as .csv, and they match the manuscript (ie the same variables and the same sample)
• Transparency: the code is archived in an open format, is annotated, and covers the whole workflow, from filtering the raw data to producing the figures
• README: the README file describes the data and code files and includes important details such as describing each variable (including units of measurement) in each dataset and providing important code version information
• Basic reproducibility: could someone outside the project work out how to go from the raw data to the figures and tables?

These stages are adapted from the Society for Open, Reliable, and Transparent Ecology and Evolutionary biology (SORTEE) guidelines for data and code quality control (), which several of us helped write. They are shared across several ecology and evolution journals, so authors can ideally expect similar standards wherever they submit.

What we don't do

This is probably the most misunderstood part of the role. Data Editors do not perform a second round of peer review. We don't judge whether your statistical model is appropriate, whether your conclusions follow from your results, or whether your code is elegant. We also can't verify that the data themselves are genuine. Our job is not focused on detecting errors or fabrication.

What we do is to make sure the evidence is available in a form that others can use and scrutinise far into the future. Open data are what allow mistakes to be found and corrected later (whether by readers or by the authors themselves), but that scrutiny comes from expert members of the community over time, not from us at the editorial stage.

The aim is to help authors meet this minimum standard, not to find reasons to reject work. Most papers only need a few small changes, and the exchange is usually quick and friendly. In our experience, mostly the changes requested are in the form of making the metadata (ie README file) clearer, and highlighting that some data or code files have not been uploaded to the repository.

The tricky cases

Not all data can be fully open. Precise locations of threatened species can expose them to poaching or disturbance, some data carry dual-use or biosecurity risks, and some are governed by Indigenous Data Sovereignty agreements (see , Proceedings B, for more). Proceedings B allows exceptions in these cases, but "available on request" on its own is not enough.

When raw data can't be shared, we still ask authors to share everything they can. This may include anonymised processed data, the numbers behind each figure and table, the code, a header file showing the data structure, and a README explaining how to request access. It helps most when authors flag these issues early in the submission process.

What we wish every author knew

The most common problems we see are not necessarily missing data, but data that nobody else could use. This includes things like unexplained column names, missing units, Excel files with several sheets, large gaps in Excel sheets (ie not tidy long- or wide-form data), and scripts that point to folders on the author's own computer that aren’t in the repository, or to files that were renamed just before submission.

A few habits can fix most of these issues:

• Start the README on day one and update it as you go. Describe every column in every data file, with units. Readers should not have to guess what a variable means. This sounds like a chore, but in most cases it only takes 15-30 minutes
• Save data as .csv or .txt files, one table per file, in the rawest form available, without blank rows or columns separating batches of data
• Archive the final version in a repository that mints a DOI, and keep data and code together where possible
• Run everything again. Running everything (within reason) before submission ensures that after cleaning your repository and renaming files, everything still works. Better yet, run it on a different machine and make sure it works

Why do it?

Most research is publicly funded, yet data are routinely lost when a researcher changes jobs or a hard drive fails. An informative, well-archived dataset doesn't have this problem. The data can feed into meta-analyses, which often need more detail than a published paper provides. It can also be re-analysed as methods improve, or used to answer questions that the authors never thought to ask themselves. It allows new groups of scientists to re-run the same analysis on a new dataset or new analyses on the archived one. Given how expensive most data are to collect, a complete and useable data repository can add a lot of extra value for relatively little extra effort.

Open data and code also make the research process visible. As analyses become more complex, a methods section alone often isn't enough to show how a result was reached. Archived code can fill that gap, and it means that mistakes (which everyone makes) can be found and corrected later. It can also improve trust in scientific results.

There are also documented benefits for authors. Sharing data and code has been linked to higher citation rates (; ) and a track record of reproducible code is increasingly valued, both in academia and beyond. For us, the role has also changed how we manage our own projects.

Image credit: Suebsiri/iStock.

Interested in becoming a Data Editor for Proceedings B? . The deadline to apply is 30 November 2026.