The First Question: Who Is Counting?
Whenever you encounter a statistic about ethnic distribution, the first question to ask is who produced it. National census bureaus, international organizations like the UN, academic researchers, advocacy groups, and journalists all produce ethnic statistics, and they use different methodologies with different biases.
National censuses are usually the most systematic source, but they reflect the categories that the national government considers politically relevant. The US census, for example, has changed its racial and ethnic categories multiple times over the past century, and the way it counts Hispanic populations differs from how European countries handle similar diversity. Some countries (France, for example) explicitly do not collect ethnic data in their censuses on principle.
International compilations like those from the CIA World Factbook or the Minority Rights Group aggregate data from many national sources, which means they inherit the inconsistencies of those sources. The numbers they report are often best-effort estimates rather than measured values, and they can be significantly out of date.
Self-Identification vs Ascription
Ethnic data can be collected in two very different ways: by asking people how they identify themselves, or by having outside observers assign them to categories. These two methods often produce very different numbers.
Self-identification tends to produce fluid, sometimes surprising results. In the US, for example, the number of people identifying as "Native American" or "American Indian" jumped substantially between the 2010 and 2020 censuses, not because populations actually grew but because more people chose to identify that way. Similarly, self-identified "European" populations in Latin American countries have shifted over time as social attitudes about identity have changed.
Ascriptive classification, where outside observers assign categories, is often used in historical research but is controversial because it substitutes the researcher's judgment for the subject's own understanding. It can also perpetuate outdated categories or overlook the ways in which identity has changed. Modern academic practice generally prefers self-identification when possible, while acknowledging its limitations.
The Trouble with Categories
Every ethnic category is a construction, and different construction choices produce very different maps. Is "Latino" or "Hispanic" an ethnic category, a linguistic category, or a geographic category? Different sources treat it differently. Are "Jewish people" defined by religion, ancestry, or both? The answer affects population counts significantly.
Some categories are highly composite. "Sub-Saharan African" groups together hundreds of distinct peoples with little in common beyond geography. "White" or "European" in North American censuses lumps together communities that in Europe consider themselves quite distinct. "Asian" in Western contexts often means something entirely different from what it means in other parts of the world.
Other categories are contested. Is Kurdish a distinct ethnicity or a subset of a larger Iranian identity? Are Sikhs an ethnic group, a religious community, or both? Are Roma a single people or a collection of related but distinct communities? Reasonable people disagree, and different sources will give you different answers.
Why Two Sources Disagree
When you find two statistics that appear to disagree, the disagreement is usually explained by one of a few factors. First, different definitions: source A might use "Kurdish" to mean only Kurdish speakers, while source B includes anyone of Kurdish ancestry. Second, different time periods: a statistic from 1990 will not match a statistic from 2020 for any population undergoing significant demographic change. Third, different geographic scope: "Turkish population" might refer to Turkey only, or include the diaspora, or include ethnic Turks in Balkan countries.
A fourth common source of disagreement is methodology. A census that asks about ancestry produces different numbers from a census that asks about primary language, which produces different numbers from a survey that asks about identity. All can be internally consistent while disagreeing with each other.
When you see a big difference between sources, do not automatically assume one is wrong. Try to understand what each source is actually measuring. Often both are correct within their own frame, and the difference tells you something interesting about how the underlying category is being conceptualized.
Common Pitfalls to Avoid
The first common pitfall is treating ethnic categories as biologically fixed. Modern genetics has thoroughly demonstrated that human populations grade smoothly into each other, that within-group variation is usually greater than between-group variation, and that the sharp boundaries in ethnic categories are social constructs rather than natural kinds. Ethnic maps should be understood as maps of culture, identity, and community, not maps of biology.
A second pitfall is projecting present categories onto the past. Modern ethnic identities often did not exist in their current form even a century ago. Talking about "German" or "Italian" or "Chinese" populations in the year 1500 is anachronistic; the people who lived in those regions had different self-conceptions and different loyalties. Historical demographic estimates should always be treated with skepticism.
A third pitfall is over-precision. When you see a claim that a country is "17.3 percent" some ethnic group, you should be suspicious of the decimal. Real ethnic statistics almost always come with substantial uncertainty, and precise-looking numbers can create a false sense of accuracy. Ranges and rough estimates are usually more honest than exact figures.
The final pitfall is treating any single map or statistic as authoritative. Ethnic distribution is best understood through triangulation: looking at multiple sources, understanding their methodologies, and being comfortable with a picture that is fuzzy rather than sharp. A user of ethnic data who becomes attached to any single source is probably missing important nuance from the others.