Deposit requirements differ by funder, and transferring data out of Vietnam may be subject to national rules — confirm with your research office and ask your partner to confirm with theirs.
Your grant requires the underlying data to be deposited on publication. You mention this to your partner while drafting the paper, and discover that the consent forms say the data will be used for this study only.
At that point the options are poor. Raised at proposal stage, the same issue is routine.
Where the collision comes from
Open-data mandates were largely written in systems where deposit infrastructure, standard consent language and institutional support already existed. Applied to a partner operating under different rules, three frictions appear:
Consent wording. Standard templates at many institutions do not cover secondary use or repository deposit. Without that wording, no agreement between funders and universities can authorise sharing.
Institutional and national rules on data leaving the country — particularly for health data, human subject data, and data on genetic resources or protected species. Your partner's institution may have a review process for this, and it takes time.
Who bears the work. Preparing a dataset for deposit — documentation, cleaning, a data dictionary — takes real effort. If it is not budgeted and assigned, it lands on whoever collected the data, which is usually the partner side.
Settle these at proposal stage
Read the mandate properly and tell your partner what it actually requires. Most funders accept controlled access or managed sharing for human-participant data. "Everything must be public" is usually an overstatement of the policy, and it causes partners to agree to things they cannot deliver.
Get sharing into the consent form and the ethics application. This is the single decisive step. Offer your partner the wording you need; drafting it is easier for you than for them, and their ethics committee will review it in any case.
Ask about transfer restrictions in writing — not whether data can be shared in principle, but what approval is needed for it to leave the country and how long that takes.
Budget and assign dataset preparation, naming who does it and with what funding. This is ordinary project work and should be costed as such.
What may not be shareable
Four categories to expect, and none reflects reluctance on your partner's part:
Data that cannot be adequately de-identified. In a study conducted in one commune, one hospital, or one narrow occupational group, removing names is not sufficient — a combination of a few variables identifies individuals.
Data on vulnerable groups, which generally warrants controlled access at most.
Precise locations of protected species, heritage sites or resources, where publication can cause direct harm. Coordinate fuzzing is standard practice.
Third-party data — held under licence from an agency, a hospital system or a company. Neither party can share what neither owns.
Where one of these applies, document the reason in the data availability statement. Funders accept ethical and legal grounds; what they do not accept is silence.
Credit for the group that collected the data
This concern is legitimate and is often left unspoken. Years of fieldwork become a public dataset from which better-resourced groups publish quickly.
Three mechanisms, agreed in advance:
An embargo period before release, allowing the collecting group to publish their primary papers first. Most funders permit a defined embargo when it is requested with a reason.
Dataset citation. Deposit with a persistent identifier so the dataset is itself a citable output, and state the required citation in the licence terms.
A stated preference for collaboration in the licence — reusers are asked to contact the original team. This is a norm rather than an enforceable term, but it is observed more often than not.
What is not reasonable to promise your partner is automatic co-authorship on every downstream paper. Say so plainly rather than leaving the expectation to form.
Preparing a dataset people can actually use
A deposited dataset with no documentation satisfies the letter of the mandate and helps nobody.
Four things it needs: a data dictionary defining every variable, units and codes; a description of how the data were collected; a record of processing decisions including exclusions and missing-data coding; and open formats readable without commercial software.
Prepare this during the project. Reconstructing collection details two years later produces documentation that is incomplete and, in places, wrong.
One question to ask at the first budget meeting
Ask your partner: does your institution have a process for approving data leaving the country, and has your group been through it before?
If the answer is no to the second part, add time to the schedule and offer to help prepare the submission. A partner navigating an unfamiliar internal process alone, under your deadline, is a predictable source of delay — and it is one of the few frictions in this area you can remove simply by planning for it.
Does an open-data mandate mean everything must be public?
Usually not — most funders accept controlled access or managed sharing for human-participant data, and overstating the requirement leads partners to agree to what they cannot deliver.
What is the single decisive step?
Getting repository deposit and secondary use into the consent form and the ethics application at the start — no later agreement can authorise sharing without it.
What data may not be shareable?
Data that cannot be adequately de-identified in small populations, data on vulnerable groups, precise sensitive locations, and third-party licensed data.
How is the data-collecting group credited?
Through an agreed embargo before release, dataset citation via a persistent identifier, and a stated collaboration preference in the licence — not automatic co-authorship downstream.
