How to Submit Diaspora Data So Others Can Verify It

To submit diaspora data that anyone can use, send three things together: the data in an open, readable format, documentation that explains where it came from and what it leaves out, and proof that the people in it are protected. A table with no context is a claim, not evidence. This guide walks through what counts as diaspora data, how to document it, which formats work, how to handle consent and privacy, and how to report gaps honestly. It also covers how to build a diaspora database of your own and how to contribute to Hellenism Watch.

Hellenismwatch

Support Verifiable Diaspora Research

Explore how Hellenism Watch monitors diaspora developments through independent analysis and contributions from scholars across its global network.

Visit Hellenism Watch

TL;DR

Diaspora data covers both individuals (location, age, qualifications, occupation) and collective bodies (associations, clubs, networks, community organizations).
Ship the dataset with its metadata: definitions, source, collection date, method, limitations and version history.
Use open formats such as CSV or TSV for tables and a plain Markdown README. Remove personally identifiable information before sharing.
State what is known, what is unknown and what was not measured. Never let a precise-looking number imply certainty.
Researchers who want to contribute to Hellenism Watch can start from our fellows page.

What counts as diaspora data
Start with the question, not the spreadsheet
Find and assess your sources
Document provenance so the data can be checked
Choose formats other people can open
Protect the people behind the numbers
Report uncertainty without overstating
How to build a diaspora database
Our perspective: why verifiability comes first
Contribute to Hellenism Watch
FAQ
Sources

What counts as diaspora data

Most people picture a headcount. In practice, the field is wider. An IOM publication on engaging diasporas as development partners describes two layers. The first is individual information: demography, location, gender, age, qualifications and occupation. The second is collective data on associations, networks, community organizations, clubs and societies.

The second layer matters more than it seems. A community that is small in numbers can be large in institutional reach, and the reverse is also true. If you only count people, you miss the schools, churches, cultural centers and professional networks through which a community actually acts.

The IOM Diaspora Mapping Toolkit, published in 2022, organizes the field into four kinds of capital:

Type	What it covers
Economic	Business, remittances and investment
Human	Skills and knowledge
Cultural	Identity and participation
Social	Civic and political engagement

Before you submit anything, decide which of these your data speaks to. A dataset on the membership of cultural associations is cultural and social capital. A survey of professionals’ occupations is human capital. Saying so in your documentation tells a reader at once what the data can and cannot support.

Start with the question, not the spreadsheet

Good submissions begin with a sentence that defines the population. “Greeks in the Balkans” is not a definition. Who is included: citizens, descendants, self-identified members of the community, members of registered organizations? Which period? Which territory? Different definitions produce different numbers, and two datasets that look contradictory often just count different people.

Write that definition down before you touch the data. Then write down the unit of observation: a person, a household, an organization, an event. Mixed units in one table are the most common reason a dataset cannot be reused.

Find and assess your sources

The IOM publication mentioned above suggests combining domestic and external sources: consulates, ministries, statistics offices, population registers, censuses, employment agencies and surveys of diaspora groups. Each has a blind spot. Registers count only those who register. Censuses may not ask about ancestry or identity. Surveys reach only people who answer.

The same publication lists examples of how countries estimate diaspora size, including consular registers in Colombia, identification cards such as India’s PIO and NRI schemes and Mexico’s Matricula Consular, and national registries of overseas voters. The lesson for any contributor: name the type of source and say who it leaves out. If your figure comes from a consular register, the honest description is “people registered with the consulate”, not “the community”.

Whenever you can, cross-check. If one source says a community has 30,000 members and another says 80,000, do not average them. Report both, explain the definitions behind each, and say which one you consider more reliable and why.

Document provenance so the data can be checked

Readers cannot verify what they cannot trace. UNESCO’s Data Governance Toolkit (2025) says documentation should include metadata and contextual information, limitations and biases, and a record of provenance: where the data originated, how it was transformed, and which version you are looking at.

In practice, that translates into a short checklist to attach to every submission:

Source. The institution, document or survey the data came from, with a link or full citation.
Date. When the data was collected and when it was last updated. These are different dates.
Method. Census, register, survey, scraping, interviews, manual compilation. For surveys: sample size, sampling approach and response rate.
Definitions. Who counts, what each column means, what each code stands for.
Transformations. Anything you cleaned, merged, recoded or estimated, and how.
Limitations. Known gaps, biases and anything you were unable to verify.
Version. A version number or date, so later corrections do not create confusion.

What quality control looks like at scale

You do not need a statistical agency’s budget to borrow its habits. The World Bank’s 2024 documentation for harmonized data describes automated quality checks, including audits of consistency against external sources and trend analysis to catch unexpected jumps in a series. The World Bank’s Pacific Labor Mobility Survey used a field quality control system with built-in computerized validation, daily dashboard monitoring, interviewer risk indexing and double-blind back-calls.

Scaled down, this means: compare your numbers to a second source, plot the series and look for jumps you cannot explain, and have someone else re-check a sample of your entries. A jump from 4,000 to 40,000 between two years is usually a typo or a change in definition, not history.

Choose formats other people can open

A dataset locked in a proprietary file or a scanned PDF is hard to reuse and impossible to check automatically. Dryad’s file requirements recommend open formats such as CSV, TSV or ODF for tables and Markdown for README files. If you do use Excel, Dryad says the essential Excel features needed to understand the data, such as formulas or color coding that carries meaning, should be explained in the README.

A workable package for a typical diaspora dataset looks like this:

data.csv: one row per unit of observation, one column per variable, consistent codes.
README.md: the definitions, source, date, method and limitations listed above.
codebook.csv: each variable name, its meaning, its allowed values and its unit.
Supporting files: the questionnaire, the scripts used to clean the data, and any reports.

The last item follows UNHCR’s guidance in its Raw Internal Data Library, which recommends including the questionnaire, preferably with variable names matching those in the data, plus supporting materials such as concept notes, fieldwork reports and processing or analysis scripts. ICPSR, a long-standing social science archive, puts it simply: deposits should include data files, documentation and study-level information needed for others to understand, evaluate and reuse the data.

Some repositories also let you decide visibility. The University of Toronto Scarborough’s Haven project, for example, accepts data through a sharing form or a contact route, and contributors can ask that their data be kept public or private. Whatever channel you use, say in your submission whether the data may be published openly.

Protect the people behind the numbers

Diaspora data is often sensitive. It can reveal where people live, which organizations they belong to, and in some cases their political or religious affiliation. A careless release can harm the very community you are studying.

Remove what identifies individuals

Dryad does not accept personally identifiable information, and it asks that human-subject data meet consent, IRB and anonymization requirements. It may also reject interview transcripts or recordings even when quantitative or summarized qualitative data can be properly de-identified. So submit coded or aggregated results, and keep raw recordings out of the package unless you have explicit permission and a secure channel.

Removing names is not enough. A row that says “priest, age 61, village of X” identifies someone. The World Bank’s 2024 microdata documentation describes the k-anonymity method, which aims to minimize the risk of re-identification by managing privacy thresholds while the data is edited. The practical rule: if fewer than a handful of people share a combination of attributes, merge categories, widen age bands or suppress the cell.

UNHCR’s data protection policy requires a data protection impact assessment for new systems, projects or data-transfer arrangements that may negatively affect personal data protection, especially where processing is large, repeated or structural. It also allows transfers to third parties only on a legitimate basis and for a specific purpose, limited to what is necessary, with written assurances of confidentiality and appropriate protection. Those are refugee-protection rules, not a legal requirement for an academic submission, but the logic transfers well: share the minimum, for a stated purpose, and with written terms.

UNHCR’s resettlement handbook adds a point worth remembering: where there is a large imbalance of power, consent may not be freely given. Even in ordinary research, ask yourself whether people could really have said no.

IOM’s Migration Data Strategy puts respect for rights, dignity, privacy and protection from harm at the center of its work across the whole data life cycle. That is a good test for any submission: would the people in this dataset be comfortable with how it is being used?

Be stricter when the topic is political

Researcher Joseph Mensah, writing in 2018 on estimating a diasporic population, warns that when data is collected for political purposes, such as diaspora voting, ethics, security and confidentiality protocols need tightening to guard against perceptions of political bias or manipulation. Data on influence, representation or institutional ties falls in that category. Document who funded the collection and who had access.

Report uncertainty without overstating

Every diaspora dataset has gaps. The credible ones say so. The UN Expert Group on Migration Statistics, in its 2025 task force report, advises defining the population and measures, disclosing coverage and missingness, identifying methodological and ethical constraints, comparing estimates with historical trends and other sources, and publishing uncertainty ranges where possible. It also asks authors to explain plainly what is known, unknown and not measured, and to flag changes in definitions or series so readers do not mistake a precise-looking number for a certain one.

The same report points to what makes results verifiable: definitions, sources, methods and limitations published with the dataset, not only a table or chart. Here is how that sounds in a README:

Estimate: 12,000 to 18,000 people. Based on the 2021 register, which excludes unregistered residents, and two association membership lists that overlap by an unknown amount. Not measured: seasonal residents. Definition changed from the 2015 series, so the two are not directly comparable.

That paragraph is more useful than “15,000”. It tells the reader what to trust, what to question and what to look for next.

How to build a diaspora database

If you are assembling many submissions into one database, the hard work is consistency. Three decisions come first.

Fix the definitions and fields. UNESCO’s toolkit says standardizing documentation through common protocols and definitions, for example DCAT for data catalogs and ISO 27001 for security, is necessary for interoperability, consistency and usability. For a small project, that means one shared codebook: the same spelling of place names, the same categories for organization types, the same date format, and a field that records which source each row came from.

Map who needs what. The same toolkit highlights stakeholder mapping as a foundational step: identify who is affected and what they need regarding quality, access and security. Contributors, readers, and the communities described will want different things. Write them down before you design the fields.

Agree on sharing terms. UN draft recommendations on international migration statistics, dated January 2025, call for legal frameworks and data-sharing agreements between government departments so data can be integrated and used in line with governance and protection standards. For an independent project, the equivalent is a short written agreement with each contributor: what they provide, how it may be used, and how corrections are handled.

For update methods, keep it simple: a version number, a change log, and a rule for what happens when a new source contradicts an old one. Never overwrite silently. Keep the old value, mark it superseded, and record why.

Our perspective: why verifiability comes first

We built Hellenism Watch because information about Hellenism across the Balkans and the Eastern Mediterranean is scattered, uneven and often shaped by whoever is telling the story. Governments, parties and advocacy groups all produce numbers, and many of them cannot be traced. That makes crises harder to answer, because analysts argue about whose figure is right instead of what to do.

Our answer is a simple standard. We stay independent of any government or party, and every claim must be supported by a credible source a reader can open. We apply the same standard to contributions. A modest dataset with clear definitions, stated limits and a traceable source is worth more to us than an impressive one nobody can check.

We also believe no single researcher sees the whole picture. A scholar working on one community, one institution or one country holds knowledge that an overall view needs. That is why we invite contributions from a global network of scholars: the aim is a shared, documented picture of Greek communities that serves academics, analysts and the communities themselves.

— Alexios

Contribute to Hellenism Watch

Hellenism Watch publishes free research, articles and evidence-backed assessments on the Greek diaspora, diaspora institutions, foreign influence networks and Greek representation. We also run an open fellowship for scholars worldwide who want to contribute data and analysis. If your work fits, begin on our fellows page, and learn more about the project on the Hellenism Watch homepage.

Hellenismwatch

When you reach out, bring the package described above: your definitions, sources, dates, method and limitations, plus a note on how personal data was handled. It makes review faster and your contribution stronger.

FAQ

How do I submit evidence for a claim about a diaspora community?

Send the evidence with enough context for a stranger to check it: a citation or link to the original source, the date it was published or collected, and a sentence on what exactly it supports. If you changed or summarized the source, say how. Claims that rest on documents anyone can open are the ones that hold up.

What file format is best for diaspora data?

Use open formats. CSV or TSV works for tables, and a Markdown README works for documentation. If your spreadsheet relies on formulas or color coding to make sense, explain them in the README, as repositories such as Dryad advise.

Generally not. Repositories such as Dryad do not accept personally identifiable information, and human-subject data needs documented consent, ethics review and anonymization. Share aggregated or coded results instead, and talk to your institution’s ethics board if you are unsure.

What if my data is incomplete or conflicts with another source?

Submit it anyway, with the gaps described. Report what is missing, compare your figures with other sources and historical trends, and give a range where you can. Conflicting numbers are often a sign of different definitions, so explain which population each one counts.

How do I start building a diaspora database?

Begin with the population definition and a shared codebook, then add a source field to every record. Decide who the users are, set written terms with contributors, and keep a change log so updates never overwrite earlier values without a trace.

How to Submit Diaspora Data So Others Can Verify It