The Transparency and Openness Promotion (TOP) Guidelines were developed in 2015 by a coalition of researchers, journal editors, and funders from the social sciences. The TOP Guidelines were designed to provide journals and publishers with specific policies for increasing transparency and reproducibility in research. TOP focuses specifically on practices that can increase the verifiability of empirical research findings. While TOP is applicable to many domains, it is particularly relevant to methods that involve statistical inferences from studied populations to broader populations.

Over the past decade, the TOP Guidelines’ impact has expanded to inform policies adopted by funders, universities, and research institutions. The framework provides clear standards for policymaking and has been supported by over 5,000 research journals. Adoption of policies that comply with TOP requirements has been mixed, with many journals using some policies that correspond to TOP’s data policy recommendations, while many of its more stringent levels or practices, including replications or preregistration, have not been widely adopted.

TOP 2025 marked a major update of the framework, and incorporated lessons from the past 10 years of use. New practices for sharing research protocols were added. Some sections were reorganized to combine specific benefits from multiple individual practices, e.g., computational reproducibility requires review of both shared data and analytic code. The core focus remained, which was to provide policies that increased the verifiability of empirical research.  

The framework provides clear standards for policymaking and has been supported by over 5,000 research journals

VIEW TOP SIGNATORIES

The TOP 2025 Framework

TOP 2025 is organized around seven research practices, two verification practices, and four verification study types. These components can be implemented independently or in combination, depending on policy goals and context. As with TOP 2015, the research practices can be implemented in three levels of increasing rigor, depending on the capacity and intent of the researchers.

Research Practices

The seven research practices map onto the research lifecycle, from study planning through reporting. The first three can be understood as planning-related practices, typically completed before data collection begins:

• Study registration involves creating a structured, time-stamped record of the existence of a study and a summary of the study plan in a persistent registry.

• Study protocols provide detailed, step-by-step descriptions for conducting the study, exceeding the level of detail found in typical methods sections.

• Analysis plans specify how data will be statistically analyzed in advance, helping distinguish planned from unplanned analyses.

The next three practices concern research outputs:

• Materials transparency requires clear identification of all physical and digital resources used in a study, including names, sources, and persistent, unique identifiers when available.

• Data transparency involves sharing the data underlying reported findings, accompanied by sufficient documentation (such as processing steps, metadata, and README files) to enable reuse.

• Analytic code transparency requires that the code used to generate results be shared in a form that is understandable and executable by others.

The seventh practice is reporting transparency and focuses on ensuring that the final manuscript includes sufficient detail to evaluate and reproduce the study. Structured reporting guidelines, such as those cataloged by the EQUATOR Network, help standardize reporting of key methodological details, enabling readers to more fully understand and potentially replicate the work.

Guidelines for Policy Implementation

Each research practice can be implemented in one of three types, which represent distinct policy approaches rather than a sequential progression:

●       Disclosure: Requires authors to state whether a practice was used. This is most often implemented as a Data Availability Statement, where authors state whether or not datasets are available and, if so, how and where they can be accessed.

●       Share & Cite: Requires that outputs be shared in a citable form, with a persistent identifier. This is most often implemented by posting outputs in a registry, data repository, or institutional repository.

●       Certification: Requires independent assertion that shared outputs meet defined best practices. This exceptional practice requires reviewers or paid staff to evaluate outputs against a rubric, an uncommon practice that can add substantial credibility to research findings.

While maintaining a consistent framework across practices, this structure enables policymakers to implement requirements based on feasibility, disciplinary norms, and available resources. 

Verification Practices

The two verification practices require a combination of research practices implemented at the Certification level.

• Computational reproducibility requires that data and analytic code be shared and verified such that a third party can reproduce reported results.

• Results transparency focuses on alignment between prespecified plans and reported findings, which helps address and minimize selective reporting or p-hacking, which undermines many statistical models.

In practice, computational reproducibility requires careful documentation and management for code and data, while results transparency includes reporting all prespecified analyses, clearly noting deviations, and labeling exploratory analyses. 

Verification Study Types

TOP 2025 identifies four study types that directly contribute to the verifiability of the research literature. These study types provide specific benefits to the scientific literature by helping to confirm early findings, but may be more difficult to fund or publish without concerted effort.

• Replications test whether independent teams can generate similar findings using the same methods.

• Registered Reports shift peer review to the study design stage, with in-principle acceptance granted before results are known.

• Many-analyst studies examine how analytic decisions influence outcomes by having multiple teams analyze the same dataset.

• Multiverse analyses systematically explore all plausible analytic pathways within a single study, by a single author or team.

A Framework for Policy Design

TOP 2025 provides shared language and policy options that can be tailored to specific research environments. For policymakers, TOP supports incremental adoption and accommodates disciplinary variation. The guidelines are an evolving effort, and the TOP Advisory Board aims to incorporate feedback for future improvements. Those interested in supporting TOP’s continual improvement can apply to join the board or reach out to top@cos.io. As the research ecosystem continues to evolve, TOP offers a framework for strengthening transparency while remaining responsive to distinct research community needs and contexts.

To learn more about what the TOP guidelines look like in practice, see our downloadable guide.

Disclaimer: David Mellor is the Senior Policy Analyst of the Center for Open Science, and is a member of the Science Integrity Alliance Advisory Board.