Methodology

How we count Albanian Americans.

Every number on this site comes from a source you can verify. This page is the audit trail. If you're a journalist, a funder, or another community organization checking our work — start here.

Last reviewed: May 2026

How many Albanian Americans are there?

The honest answer climbs as you widen the question. Three definitions, three numbers — what the Census records, what NAR's analysis estimates, and what the community has long counted. We show all three, because choosing one and hiding the rest is how this kind of count goes wrong.

Recorded — U.S. Census

224,000

The floor. The people who reported "Albanian" as their first ancestry on the American Community Survey (ACS 2024, Table B04006). It is the federal government's official figure — and it misses people by design.

What it misses:

Ethnic Albanians from Kosovo, North Macedonia, Montenegro, and Çamëria — the form has no "ethnic Albanian" box, only "Albanian" (read as Albania-the-country), so they fall under "Yugoslav" or their country of birth.
Anyone who reported a different first ancestry — the form takes one per person, so a mixed-heritage grandchild picks one.
2nd- and 3rd-generation Americans whose families now identify primarily as American.
Households that skip the ancestry question (~10–15% of ACS responses).

Source: data.census.gov ACS Table B04006 (the precise 2024 combined-ancestry figure is 230,682; we use the rounded ~224,000).

The ACS is the official federal count and remains essential. Please complete it when it arrives — federal funding formulas, voting-rights determinations, and most program decisions read from this number. NAR is not affiliated with the Census Bureau and does not replace the ACS; we run the community-led parallel count alongside it, capturing what an optional ancestry write-in once a decade by design cannot.

NAR's estimate — by ancestry & close descent

~650,000

The figure NAR stands behind. Correct the documented undercounts one at a time — counting each person once — and the estimate lands near 650,000, in a range of 500,000–810,000. About twice the recorded figure: the same undercount ratio documented for Arab-American and Iranian-American communities. Five mechanisms, each a population the Census sees only sideways:

Yugoslav-ancestry misallocation — ethnic Albanians from the former Yugoslavia filed under the state they left, not the people they are. The largest single correction.
Pre-1912 classification drift — Albanians who arrived before independence, logged as Greek, Turkish, or Italian.
Self-identification reluctance and foreign-born / limited-English undercount — households that answer the ancestry question at lower rates.
Heritage-variant communities counted under host-country ancestry — a conservative, documented slice.

This is a mechanism-based estimate — prepared in-house, still preliminary, and open for outside review. Because the mechanisms partially overlap, the arithmetic sum is an upper bound, so read ~650,000 as upper-leaning. It uses the definition the federal frameworks this analysis serves actually count — voting-language access, interpreter requirements, and program funding reach people who would identify or be served as Albanian, not distant heritage. That is why it is the number we stand behind.

Read the full mechanism methodology (markdown) →

The community's estimate — broad heritage

~1,000,000

Widen the definition to everyone of Albanian origin in America — including the full Arbëreshë line who reached America as "Italians" centuries ago, and distant descent to great-grandparent — and you reach the figure the diaspora has named for years: close to a million. This is the community's estimate, not NAR's analytic claim — provisional, built on the widest definition, cited by four classes of sources, each with limits:

Albanian-state diaspora estimates — Albania's Ministry of Foreign Affairs and the Council of Albanians in the U.S. have cited 500K–800K for years. Diaspora-engagement figures, not census-grade counts.
ACS foreign-born stock — ~85K Albania-born + ~30K Kosovo-born; converting that to a total ethnic population needs cohort math not yet done.
Religious-institution rolls — Orthodox, Catholic, and Muslim Albanian congregations report active members in the low hundreds of thousands. Attendance, not population.
Ethnic-association membership — AANO, VATRA, and regional groups, at a cited 5–10% capture ratio, imply 750K–1.5M. The ratio's transferability is an assumption.

How we carry it: these sources bracket a range commonly cited as 800K–1.2M. We report ~1M as the community's estimate — because it is real under the broadest definition and because it is the number this community has always known — not as a precise or independently validated figure. We could not locate a single citable institutional publication that states it with a clear methodology, and no published U.S.-specific count confirms it. It is the honest ceiling of the range.

Which number is true? NAR's analysis says ~650,000. The community says close to a million. The Census says 224,000. None of them is a count — so we are building one, name by name, to settle it.

Who counts as "Albanian American"

We use the broadest defensible definition. A person is Albanian American for our purposes if any of the following is true:

They were born in Albania, Kosovo, North Macedonia, Montenegro, or Çamëria (Greece) and now reside in the U.S.
They have at least one parent, grandparent, or great-grandparent who meets the above.
They are Arbëreshë — descendants of the 15th-century Albanian migrations to southern Italy and Sicily — and now reside in the U.S.
They self-identify as ethnically Albanian (Shqiptar) regardless of country of birth.

This is the same boundary the 2020 Albanian Citizenship Law (No. 113/2020) draws for citizenship-by-descent eligibility. We chose to mirror it because it's already the legal standard the Albanian state uses for "who is Albanian," and because a registry that conflicts with the citizenship pathway helps no one.

What kind of sample is this

NAR's registry is a voluntary, self-selecting sample. It is not a probability sample. We do not randomly draw registrants; people choose to register. We make no claim of population-representativeness.

The sample's composition is shaped by which segments of the Albanian-American community respond to our outreach. Today that means primarily: paid social-media campaigns, regional ambassadors, organic search arrivals, and partner-organization referrals. Segments under-represented in those channels are also under-represented in the registry. Examples we know of and don't yet correct for: less-online elders, recent foreign-born immigrants without strong English-language exposure, communities served primarily by religious institutions we haven't partnered with, and Albanian-Americans in states where we have no ambassador.

What this means for using NAR's data: the registry is defensible for "we have N self-attested Albanian-Americans who agreed to be counted, with the following characteristics." It is not, today, defensible for "the Albanian-American population in state X is N." Converting registry data into population estimates requires a written bias / coverage / weighting plan we have not yet built. That work is on our roadmap; it is not done.

The live registry number

1443

On the registry

What this counts →

People who completed the registration form and passed our automated quality gates: Cloudflare Turnstile (bot protection), cross-email duplicate detection, and DNS deliverability check. This is the figure on every public Counter on the site and on the homepage. Refreshes live in your browser.

SQL: count(*) where is_test is not true

The Counter on the homepage and the "registered" cards on the pitch page query our Supabase database at request time, fetched fresh via the count_people() RPC. There is no cache layer between the database and the rendered number beyond a 10-second edge cache.

Each registrant row is one self-attested individual who completed the registration form. Household members are first-class rows so a household submission for four contributes four to the count, not one. Test and staff registrants are flagged is_test = true and excluded from the count.

The number is what it is: people on the registry. We don't filter further. We don't claim it represents the full Albanian-American population — that's the ~1M figure on the cards above, an external estimate triangulated from independent sources. The architecture and methodology decisions behind the public count are documented in the application audit at /admin/system/audit-2026-05.

How to audit our numbers

If you want to verify any specific claim:

1.For Census numbers — start at data.census.gov, search "B04006 Albanian," select 5-year estimates, vintage 2024. The number on the table is the number we cite.
2.For diaspora estimates — Albania's Ministry for Diaspora publishes annual reports; the Council of Albanians in the United States releases periodic figures. We can email you the specific reports we triangulated against.
3.For our live registry counts — every count card has a "Live" indicator confirming it's freshly fetched. Refresh the page; if the number changed, you've just seen the registry update in real time.
4.For anything else — email support@albanianregistry.org. We will share primary sources for any cited number on the site within seven days.

Empirical foundations — what the registry can and can't measure

The ~1M figure is a provisional working estimate compiled from external sources (above). The registry itself is a separate empirical instrument — it can directly observe certain things, and only indirectly observe others. Honest documentation of which is which is part of keeping the methodology defensible.

The 224K Census number is acknowledged by Albanian-American community organizations as an undercount, for at least five reasons we have identified. These are NAR's framing, not a closed canonical list from the demographic literature. Each row below shows whether NAR's registration form collects data that lets us identify registrants who fall into that mechanism — not whether the mechanism size is rigorously quantified at the population level.

Mechanism	Direct?	How the registry handles it
1. Pre-1912 descent	Partial	Albania declared independence in 1912. Albanians who immigrated to the U.S. before that date were recorded by the Census as Greek, Italian, Yugoslav, Ottoman, or Turkish subjects, depending on origin and Census year — the "Albanian" ancestry option didn't exist as a category their ancestors could pick. Their U.S.-born descendants now carry that label downstream. Registry handling: we ask anyone identifying as 2nd-generation or later "Did your earliest Albanian ancestor in the U.S. arrive before 1912?" with options Yes / No / Not sure. Yes/no/not-sure rather than asking for a year, because most respondents three or four generations down don't know the year. The yes-bucket count gives a direct empirical floor on the pre-1912 mechanism. We'll publish the bucket distribution on this page once the registry has enough rows for the figure to be stable (target: post-1k registrations).
2. Yugoslav-ancestry misallocation	Yes	Ethnic Albanians from Kosovo, North Macedonia, Montenegro, the Preševo Valley, or pre-1991 Yugoslavia broadly were often recorded as "Yugoslav" or by the new state's nationality on the Census, even when they identified as Albanian. Registry handling: `albanian_origin` is multi-select including AL, XK, MK, ME, PV, RS, CM. A Kosovar-Albanian can register without checking Albania-the-country and the row still carries the ethnic claim. Direct count: rows where `albanian_origin` contains a non-AL code is the empirical floor on this mechanism.
3. Heritage-variant communities	Yes	Arbëreshë (15th-century Albanian migration to southern Italy and Sicily), Çam (Albanian community of Greek Epirus), and smaller diaspora variants typically don't appear in standard Albanian-ancestry counts. Registry handling: IT (Arbëreshë), GR/CM (Çamëria) are explicit options in `albanian_origin`. Croatian Arbanasi, Mandritsa Bulgarian Albanians, and Ukrainian Albanian descendants fall under "other" with a free-text capture. Open UX question: do registrants unfamiliar with the codes "Arbëreshë" or "Çam" recognize themselves in the form? Worth periodic UX review.
4. Self-identification reluctance	Indirect	Some Albanian-Americans don't pick "Albanian" on the Census ancestry question for reasons ranging from assimilation pressure to ancestry-prompt design to mixed-heritage tie-breaking. The mechanism is inherently a survey-of-non-respondents problem — anyone reluctant enough not to self-ID won't appear in NAR's data either. Registry handling: indirect only. We capture `heritage_connection` (very-strongly / somewhat / not-very-much / not-at-all), `parents_albanian_identity` (one-parent vs. both vs. neither), and language proficiency. These let us identify recovered reluctance — registrants who self-attest on NAR but report weak connection — but not the population-level non-respondent rate. The 1M estimate's reliance on this mechanism remains tied to external sources (Albanian state diaspora figures, religious-institution rolls) rather than registry data.
5. Foreign-born / LEP undercount	Yes	Limited-English-proficiency populations are systematically underrepresented in Census/ACS due to language access, response burden, and trust gaps with federal surveys. Registry handling: `country_of_birth`, `year_moved_us`, `albanian_language_proficiency`, and `language_preference` directly identify foreign-born and LEP populations within the registrant base. Limitation: the registration form itself is bilingual EN/SQ only — Greek-only Çam migrants or third-language registrants don't have a fluent intake path. We treat that as a known floor on our LEP capture, not a measurement of the LEP population.

How NAR's registry compares to ACS

ACS Table B04006 is published as aggregate counts by state, not as a record-level dataset, so a one-to-one record-overlap comparison with NAR registrations is not possible. What we can do is a benchmark comparison: filter NAR registrants to albanian_origin = ['al'] (the closest analog to ACS's "Albanian" ancestry category) and compare per-state counts side-by-side, not claim that any specific NAR registrant is or isn't in the ACS dataset.

Useful comparisons we could publish from this benchmark:

Per-state ratios: NAR Albania-only count vs ACS B04006 Albanian count. Where the ratios diverge from each other, that's a signal worth investigating (recruitment skew, undercount geography, etc.).
State-level distribution: does NAR's geographic distribution match ACS's, or diverge? A divergence is a recruitment-bias signal, not a Census-undercount signal — both directions are informative.

What this comparison cannot do: identify which specific people are missed by ACS, or quantify the size of any undercount. That requires a coverage / capture-recapture / IPW-weighting plan with explicit assumptions — work that is on our roadmap but not yet built.

A meaningful per-state comparison also requires meaningful sample size in NAR. With a few hundred registrants spread across 50 states, per-state benchmarks are not yet statistically interpretable. The threshold is a question for the consulting demographer who reviews this methodology.

What this section is for

This page exists because every methodology claim should answer two questions: what does the data directly support? and what does it support only indirectly or not at all? An organization willing to publish the second list is more credible than one that publishes only the first. If you're a demographer, journalist, or funder reviewing NAR, this section is the audit trail for our empirical foundations. Mismatches and gaps belong here, in writing, before they're discovered elsewhere.

What we don't claim

The 1,000,000 estimate is not a precise number. It's a range. We round to a single figure for communication, but treat it as 800K–1.2M.
NAR's registrant count is not a population estimate. It's a count of self-attested registrations. We make no claim that the registry is statistically representative of the broader community.
Registrants are not legally certified Albanians. The Certificate of Albanian Identity is a community-recognition document, not a passport, identification, or legal credential.

Found an error or have a better source? Email support@albanianregistry.org. We update this page when challenged with better data.