Implementation

Authors
Affiliations

Max Planck Institute for Human Development

Tobias Bengfort

RDI

Josefine Blunk

RDI

Neele Engelmann

CHM

Thomas Feg

SCT

Stefan Herzog

ARC

Maike Kleemeyer

RDI

Sina Schwarze

LIP

Sebastian Nix

RDI

Aaron Peikert

LIP

Ilse Pit

ARC

Penelope Tilsley

CEN

Published

July 12, 2026

1 Preamble and Purpose

In addition to the general strategic provisions on OS, this part contains more detailed information on how to practically implement OS at the Institute. It describes both the responsibilities of the Institute as a whole and of its research centers and groups, as well as those researchers’ responsibilities who fall under the scope of the OS guideline.

2 Open Science Dimensions

2.1 Open Methods

2.1.1 The MPIB is responsible for

  • Providing training and resources to support researchers in documenting and sharing their methods openly.
  • Establishing and maintaining an infrastructure that facilitates the open sharing of research methods.

2.1.2 The centers and groups are responsible for

  • Ensuring that research conducted within their units adheres to Open Methods standards, making all relevant methodological information publicly accessible.
  • Supporting researchers in documenting and sharing their methods in a standardized and reproducible format.

2.1.3 Researchers are responsible for

  • Making all methods used in their research, including study designs, data collection procedures, and analysis techniques, openly available in sufficient detail to allow replication.
  • Ensuring that any tools, materials, or instruments critical to the research are adequately documented or accessible, and that proprietary tools are supplemented with open alternatives where possible.

2.2 Open Access to Publications

2.2.1 The MPIB is responsible for

  • Providing adequate information on Open Access publishing (legal issues, funding options) through the Institute’s Open Access Team.
  • Providing an Open Access publication fund to cover Open Access publishing fees in cases where no central MPS funding and/or external funding is available, subject to the fulfillment of explicit funding criteria and within defined limits.
  • Making most of the journal-based publication output of the MPIB available on the institutional publication server MPG.PuRe in Open Access by checking the conditions of a legally permissible secondary publication of works previously published in closed access (taking advantage of UrhG § 38) and coordinating this secondary publication with the authors. The Open Access Team of the Research Data & Information unit proactively reaches out to all MPIB authors with newly recorded closed access publications and supports them in sharing their work on MPG.PuRe. This includes the creation of postprint versions (provided the last accepted manuscript after peer review is no longer available), the monitoring of publisher embargos and conducting the legally compliant re-publication on MPG.PuRe. Authors retain the right to select the publication outlet they consider most appropriate in line with their funder requirements.

2.2.2 The centers and groups are responsible for

  • Checking the availability of financial resources for publication fees of pure Open Access journals when the limits of the Open Access publication fund are exhausted and providing financial support if possible.
  • Checking the availability of supplementary financial funding for publications in pure Open Access journals if central funding options that would have been possible based on the respective criteria are exhausted and providing such supplementary funding if possible.
  • Refusing funding specifically for making available only single articles in Open Access in journals whose content is usually not available in Open Access (so-called hybrid journals).

2.2.3 Researchers are responsible for

  • Publishing every original research article in a peer-reviewed journal Open Access as early as possible, ideally at the time of first publication. If the first publication is not in a dedicated Open Access publication outlet, authors are recommended to make use of the secondary Open Access publication workflow provided by the Institute in order to ensure a timely and copyright-compliant Open Access publication.
  • Informing themselves about Open Access publishing options and the formal requirements for central funding, at the latest when they consider publishing in a specific form (e.g., journal article, contribution to an edited volume, monograph).
  • Securing their own exploitation rights to all parts of their work as far as possible when concluding publishing contracts and aiming not to grant publishers exclusive rights of use to publications.
  • Using adequate publishing licenses in scientific publications (e.g., Creative Commons Licenses), to make usage conditions explicit. The licenses CC BY and CC BY-SA are the default standards, as these are the only two license types truly compliant with Open Access in accordance with the “Berlin Declaration on Open Access to Knowledge in the Sciences and Humanities”. Furthermore, the metadata referring to a scientific publication are recommended to be openly accessible under a free license that does not restrict their usage (e.g., CC0).
  • Assigning a persistent identifier (e.g., DOI) in order to make sure that a scientific publication in Open Access can be reliably accessed.
  • Informing the Research Data & Information Team about any of their publications to ensure appropriate integration into reports related to the Institute’s OS activities and, if the publication is not Open Access, to provide an opportunity for a secondary publication in Open Access. This can be done by adding it as a publication in the study registration tool or notifying the Research Data & Information unit (via mail or form).

2.3 Open Data

The Institute adopts publication of research data as the default.

2.3.1 The MPIB is responsible for

  • Providing adequate information on data publishing (outlets, licenses) through the Institute’s RDM Team.
  • Providing adequate support for handling legal issues through the Institute’s Data Protection Coordinator.
  • Providing training and resources to support researchers in organizing, documenting, and sharing their data openly in accordance with the FAIR criteria.
  • Establishing and maintaining an infrastructure that facilitates the open sharing of research data.

2.3.2 The centers and groups are responsible for

  • Ensuring that research conducted within their units adheres to Open Data standards, making all relevant documentation publicly accessible to the extent that is legally admissible.
  • Supporting researchers in organizing, documenting and sharing their data in a standardized format.

2.3.3 Researchers are responsible for

  • Making their data publicly accessible in an appropriate format and in compliance with legal provisions (please also see the Data type specific regulations in the RDM implementation).
  • Notifying the head of the unit in good time before publication.
  • Ensuring data are publicly accessible with minimal access hurdles. As a general rule, the Institute prefers data sharing options where commercial usage is excluded through simple declaration mechanisms (such as platforms that require users to confirm non-commercial use) rather than complex approval processes. If legal or ethical reasons prohibit unrestricted accessibility, restricted data access is recommended to be considered including concrete implementation, i.e., how is access to the data managed, who to contact to request access, any restrictions on who the data can be made available to or for which purpose.

The following sections contain important considerations when making data accessible.

2.3.3.2 Outlet

  • Finding the right outlet for their data publication. Usually, data repositories are well suited for data publication. They can be discipline-specific (e.g., EBRAINS), general (e.g., Zenodo, OSF), or institutional (e.g., Edmond (MPG), Open Data (Ludwig-Maximilians-Universität München)). If available, discipline-specific repositories are preferable because they tend to be well-known in their specific research field, and researchers are more likely to search for data in those repositories. The search function of the re3data portal can be helpful for finding a trustworthy, discipline-specific repository. Important criteria when looking for a repository include provided licenses and persistent identifiers, associated costs and user rights, the repository’s hosting country (GDPR) and certification as well as potential upload limits. Research data centres(RDCs) are another suitable option for data publication (e.g., RDC at ZPID). Besides quality-assured curation, RDCs offer various access routes for the scientific use of data corpora.

    Outlets for identifiable data need to fulfill additional requirements based on data protection legislation: First, the outlet must ensure that servers are located in areas that fall under German and European data protection regulations. Furthermore, they need to allow restricted access only for researchers at academic institutions who can provide a detailed data protection plan assuring that data are handled according to German and European data protection regulations. Finally, they must provide an option for getting all data users’ consent to not even attempt to identify people.

    Outlets for data specified as “special categories of personal data” by GDPR §9 must additionally ensure that the data access application requires the applicants to provide approval from their local IRB.

    Non-exclusive examples for such sensitive data that are frequently produced at our Institute include raw brain data cf. MR data publication or genome-wide data. While Ebrains is a feasible outlet for the former, the latter can be made available via the European Genome-phenome Archive.

2.3.3.3 Documentation

  • Making metadata as completely as possible freely and openly accessible, including links to the corresponding article and/or code repositories if applicable. As they contain essential information about the data’s origin and meaning, this ensures that all information relevant to the data is immediately accessible, ideally together with the data.
  • Including a meaningful README file at the uppermost level. A README file is a plain text file often written in Markdown and named “README.md”. This naming convention helps users easily identify the README file, and it is automatically displayed on the main page of relevant repositories such as GitHub.
  • Documenting data files adequately. Understanding the contents of tabular data is best supported by a sidecar file describing the variables, their units, value labels, etc. Sidecars should be stored as separate files, ideally in .yml or .json format, using the name of the corresponding data file, e.g., if survey_data.csv is the data file, the corresponding sidecar is survey_data.json.

2.3.3.4 Persistent Identifier

  • Assigning the published data a persistent identifier (PI), most commonly a Digital Object Identifier (DOI). Many repositories (e.g., Zenodo, OSF, Edmond) can assign PIs to data once they are uploaded, so this is something to watch out for. Data (and other digital objects) can be uniquely identified and accessed reliably by using the DOI proxy server https://doi.org/ and appending the DOI, e.g., https://doi.org/10.5281/zenodo.4322849.

2.3.3.5 License

  • Making usage conditions explicit, usually by assigning adequate licenses. By default, research data are to be published under the CC0 license, given that most research data at the Institute do not imply copyright. Note 1: As a general rule, the Institute prefers data sharing options where commercial usage is excluded through simple declaration mechanisms but usually not with a license that excludes commercial use. Note 2: The obligation to attribute reused scientific achievements arises from good scientific practice, regardless of license requirements. The preferred way to promote data citation is not through a restrictive license but by providing a citation recommendation. If in doubt whether research data might imply copyright, reach out to the RDM Team. If open licenses are not feasible, alternative access and usage rights are recommended to be granted to allow (i) scientific discourse based on empirical work with research data and (ii) achievement of the overarching objectives of this guideline.

2.3.3.6 Data Availability Statement

  • Including a data availability statement (DAS) in the corresponding text publication to transparently indicate that the data underlying the publication’s findings are available. A DAS is a section in a scientific publication, e.g. research article, stating that the authors have made the evidence supporting their findings available, along with details on how to access it. Note: When submitting to a journal that uses a double-blind peer review process, ensure that information in the data availability statement does not compromise your/or your co-authors’ anonymity. If your data availability statement includes information that could identify the manuscript authors (e.g., linking to a repository that reveals author information), ask the journal for guidance on how to proceed (cf. Cambridge Data Availability Statements Guide).

2.3.3.7 Notification of Publication

  • Notifying the head of the unit in good time before publication.
  • Informing the Research Data & Information Team about data publications to ensure appropriate integration into reports related to the Institute’s OS activities. This can be done by adding it as a publication in the study registration tool or notifying the Research Data & Information unit (via mail or form), which will create an entry on https://pure.mpg.de for you.

2.4 Open Software/Code

2.4.1 The MPIB is responsible for

  • Providing training and resources to support researchers in documenting and sharing their software/code openly.
  • Establishing and maintaining an infrastructure that facilitates the open sharing of research software/code.

2.4.2 The centers and groups are responsible for

  • Ensuring that research conducted within their units adheres to Open Software/Code standards, making all relevant information publicly accessible.
  • Supporting researchers in documenting and sharing their research software/code in a standardized and reproducible format.

2.4.3 Researchers are responsible for

2.4.3.1 Versioning

Versioning their code to ensure clarity regarding which code produced specific results; this requirement is typically fulfilled when using version control software like Git.

2.4.3.2 Publication

  • Making their code publicly available together with the corresponding publication (if any). As a publication platform for the code, researchers typically choose between <github.com>, <git.mpib-berlin.mpg.de>, <arc-git.mpib-berlin.mpg.de>, or <gitlab.gwdg.de> depending on technical requirements and the primary audience. Real-time collaboration platforms like Google Colab, JupyterLab, or Posit Cloud may also be considered. Generic file-sharing platforms (e.g., Dropbox, Google Drive, OneDrive, or Seafile) are discouraged.
  • Providing required files for publication, i.e., README.md (see Reproducibility), LICENSE.md (see under License), and CONTRIBUTING.md (see under License).
  • Adding a link to the code in the Institute’s study registration tool, if the published code was used for generating, preprocessing, or analyzing data in a registered study.
  • Considering the creation of a CITATION.cff file.
  • Assigning a DOI (via Zenodo or through the Research Data & Information unit on MPG.PuRe) is highly encouraged.
  • Archiving software for at least ten years, which publication on Zenodo or MPG.PuRe can fulfill.

2.4.3.3 Notification of Publication

Informing the Research Data & Information Team about software publications when a DOI has been assigned to ensure appropriate integration into reports related to the Institute’s OS activities. This can be done by adding it as a publication in the study registration tool or notifying the Research Data & Information unit (via mail or form), which will create an entry on https://pure.mpg.de for you. However, if the code is included in or explicitly referenced by a research output that you have already reported to the Research Data & Information unit, no further action is required.

2.4.3.5 Reproducibility

Employing appropriate measures to guarantee reproducibility, hence ensuring (1) availability of exact versions of code/data; (2) availability of compatible versions of external software; (3) that no copy&paste mistakes have been made; (4) that code execution is unambiguously specified. Therefore the following is recommended to researchers:

  1. Version their code (ideally using Git) and be diligent in their RDM.
  2. At least provide a machine readable list of all software dependencies including their versions. For example, in R you may use renv, in Python you may use venv, in Julia you may use the inbuilt Manifest.toml, but other solutions should also be considered. Ideally provide a containerized version of your application that contains all run-time and development dependencies. Rely on open-source tools where possible. Using closed-source software complicates collaboration, code review, computational reproducibility, and reuse.
  3. Consider using a dynamic document approach using tools like Quarto, Jupyter, RMarkdown.
  4. Ensure that it is abundantly clear what needs to be executed in what order to obtain the same results. Ideally, the whole process can be triggered with a single command. Consider using CI services, workflow orchestration tools, or suchlike to make this process seamless.

2.4.3.6 Pseudo Random Number Generation (RNG)

Diligently managing and recording the state of Pseudo Random Number Generation (RNG) upon use. RNGs are commonly employed in data simulation, MCMC analyses, random subsampling, multiple imputation, and many other procedures. As a minimum it is recommended to set a seed (e.g. python: import random; random.seed(42), R: set.seed(42), Julia: import Random; Random.seed!(42)). If an RNG state is set using a seed, it should be done as close to the point of randomness usage as possible to prevent unintended changes to the RNG state. Parallel use of RNGs (e.g., on Tardis) requires particular attention — ensure that each process uses an independent random number stream. Note that different seeds from the same RNG usually result in correlated number streams. For analyses involving multiple software environments, RNG management is recommended to be handled separately for each environment (e.g., a seed set in Python will not affect the seed in TensorFlow).

2.4.3.7 Code Review

Requesting code review to ensure reproducibility and correctness. Code is likely to contain errors that can have significant scientific and reputational consequences. It is advisable to request a formal review from a colleague, especially for code that underpins publications. You can join the MPIB Mattermost channel Code Review to find someone willing to review your code. A good approach is to start by reviewing others’ code and then request a review in return. Before requesting a review, follow the guidelines under “Publication” and “Reproducibility”.

2.4.3.8 Engagement and Reuse

  • Including information on how others can contact the authors to provide feedback, report bugs, and suggest extensions. This requirement is automatically fulfilled with an issue tracker that is usable by the public (e.g., Github, but usually not Gitlab, where the public can not create an account), which is therefore strongly recommended. The primary goal of Open Code/Software is to foster collaboration.
  • Continuously evaluate and facilitate your software’s potential for reuse by modularly structuring your code, publicly providing internal documentation, automating unit testing via continuous integration, considering and maximizing compatibility with up- and downstream dependencies, etc.
  • Encourage feedback, engagement, collaboration and reuse in the README.md, CONTRIBUTING.md, and other appropriate places.

2.5 Open Educational Resources

Open Educational Resources (OER) are freely accessible, openly licensed instructional materials that can be used for teaching, learning, and research. The Institute encourages the development, adaptation, and sharing of OER to support knowledge dissemination and educational equity. OER may include course materials, lectures, textbooks, assignments, and other educational content that researchers or educators create.

3 Research Assessment & Evaluation

As part of the recruitment and/or promotion process for academic positions, starting at the predoc level, applicants are explicitly asked to provide a statement on their commitment to OS practices, and encouraged to provide examples of their engagement with such practices.. Such commitment is positively taken into account when selecting or promoting applicants.

In decisions regarding contract extensions for scientific positions, a commitment to OS practices is also positively rated.

In the context of the Scientific Advisory Board’s evaluations and in line with the general MPG SAB guidelines, the Institute seeks feedback on activities related to the integration of OS practices into the research process.

When assessing scientific publications, researchers are encouraged to consider whether the authors of the reviewed manuscript provide Open Data and analysis code, and to point this out in their peer reviews for scientific journals.

To incentivize and celebrate MPIB researchers at all levels for making their research more accessible, transparent, and reproducible, the Institute implements an annual, internal Open Science Innovation Award. The award is presented at a designated event to foster exchange and recognition of OS practices at the Institute.

Reuse

Citation

BibTeX citation:
@misc{max_planck_institute_for_human_development2025,
  author = {{Max Planck Institute for Human Development} and Bengfort,
    Tobias and Blunk, Josefine and Engelmann, Neele and Feg, Thomas and
    Herzog, Stefan and Kleemeyer, Maike and Schwarze, Sina and Nix,
    Sebastian and Peikert, Aaron and Pit, Ilse and Tilsley, Penelope},
  title = {Implementation},
  version = {1},
  date = {2025-09-26},
  url = {https://os-rdm.mpib.berlin/guidelines/},
  doi = {10.17617/2.3682163},
  langid = {en}
}