Research · Technology · Education · Advisory
Institutional innovation for practical impactPartner with Praxis
Research & Data

Making public-health survey data ready for new questions

How Praxis Data Lab approaches harmonising long-running public-health survey data so that researchers can trust, reuse and reproduce their results.

For
Public-health researchers and educators
Where
United States and international collaborators
Year
2026
What we did
Data harmonisation · Reproducible workflows · Provenance and documentation · Research collaboration
Illustration: a thread linking five survey releases into an open data dictionary, with a code badge and a magnifying glass
The challenge

Large public-health surveys such as NHANES are released in cycles over many years. Variable names, definitions and collection methods change between releases, so every new study risks repeating the same preparation work and quietly making different choices about comparability.

Our approach

Through Praxis Data Lab, Praxis is building a reproducible approach to harmonising variables across survey releases. Each prepared dataset is kept connected to its source releases, preparation code, variable dictionary and validation checks, so anyone can see how a result was produced and rerun it. Planned research will evaluate AI-assisted methods for semantic harmonisation and metadata generation before they are treated as validated capabilities.

What it delivers
  • A documented approach that connects every prepared dataset to its sources, code and validation records
  • Variable dictionaries that record definitions, coding, units and differences between releases
  • A research agenda for AI-ready data that is evaluated before being relied upon

Start with the question

Good data preparation begins with the study design: what is being measured, for whom, and whether the measure is comparable across the releases being combined. Documenting these decisions is part of the work, not an afterthought.

A trail you can follow

A result is only as reusable as the record behind it. Keeping source releases, transformations and checks together means that a reviewer, a student or a future project can understand and repeat the analysis.

Open to collaboration

Praxis Data Lab welcomes collaborations on dataset governance, transparent contribution processes and practical training for research and public-health education teams.

Visit Praxis Data Lab

More of our work

Other case studies

All case studies
Work with Praxis

Facing a similar challenge?

Tell us what needs to change. We will help shape a practical, realistic way forward.

Start a conversation