Open Science Guidelines for Research Grants

Arnold Ventures (AV) is committed to funding research that meets the most rigorous standards of research quality and transparency. We believe in the value of reproducibility — that our investments should be accessible to scholars, researchers, and policymakers, and that transparency is essential for building public trust in science and supporting evidence-based policy decisions. 

While recognizing that specific requirements may be adapted to a project’s design and context, AV’s core open science expectations for research grantees include the following: (i) preregistration of research projects, including a detailed pre-analysis plan; (ii) open data and related materials; (iii) open code; and (iv) open access to research articles and results. 

I. Preregistration and pre-analysis plans for research projects 

Any empirical studies involving causal inference should be preregistered prior to the start of the intervention (RCTs) or prior to accessing or analyzing outcome data (QEDs). We ask that researchers preregister their studies on the Open Science Framework (OSF), which supports open-ended preregistrations and allows research teams to document and justify any adjustments made over the course of the study. If there is a compelling reason to use an alternative registry, please consult your program officer before proceeding. 

Preregistration should include a well-specified pre-analysis plan (PAP), which details how each research question will be analyzed before any outcome data is accessed. A separate guide describes the elements typically expected in PAPs for AV-funded studies, recognizing that requirements may vary by study design and context. At a minimum, PAPs should clearly specify the research questions, primary and secondary outcomes, sample definitions, model specifications, and planned subgroup or heterogeneity analyses. For certain projects — particularly quasi-experimental studies — it may not be feasible to pre-specify every modeling decision in advance. In such cases, researchers should articulate the overarching analytical framework and clearly identify any decisions that depend on future, currently unknown information. To further strengthen reliability, researchers are encouraged to propose and conduct robustness and sensitivity analyses. 

Upon completion of the research project, reports and publications should adhere closely to the preregistered design and PAP, with any deviations clearly documented, justified, and labeled as such. The results of each pre-specified confirmatory hypothesis test should be published publicly. 

II. Open data and related materials 

All data and related materials created in whole or in part with AV funding should be made publicly available, subject to legal and logistical constraints. If legal or privacy restrictions prevent sharing, researchers should clearly document the data request parameters for each source (e.g., a government agency) so others can replicate the process. 

This policy applies to all data and materials (such as survey instruments), not only a limited subset used for published findings. Researchers should share all raw and processed data, along with a codebook (when applicable), to enable other researchers to understand their structure and content. Data may be stored directly on OSF or in another recognized repository. If an external repository is used, the OSF project page should include a link to the data location. 

III. Open code 

Researchers should produce well-annotated code scripts to process, clean, and analyze data, and the final version of these scripts should be made publicly available in a permanent fashion. The code scripts should enable another researcher to take the original raw dataset(s), clean and merge them, and rerun the original analysis. As with data, code may be stored directly on OSF or in another repository. If an external repository is used, the OSF project page should include a link to the code location. 

IV. Open access to articles and results 

All results from AV-funded research (including articles, reports, and related outputs) should be freely and publicly accessible. For peer-reviewed articles, this requirement may be satisfied by publishing the article as open access or by posting the final published version, the accepted author manuscript, or a near-final working paper or preprint. Researchers are responsible for retaining sufficient rights to ensure compliance with this policy. If the final publication will appear in an outlet that imposes additional costs for open access, those fees should be included in the project’s original budget. If the results from a pre-specified confirmatory hypothesis test are not formally published, researchers should post written findings on OSF as a working paper or preprint to ensure public access.