The envelope arrives in the second week of March, sometimes April. It is thin, white, and official, with a return address in Jeffersonville, Indiana, and a barcode printed above my name. Inside are the instructions, the login code for the online version, and — if I have let the deadline pass — a paper questionnaire. The American Community Survey is not the decennial census. It comes every year to a sample of addresses, and it asks more. But the object is the same: a form that wants to know who lives here, and what they are.
I have kept copies. Not out of diligence. Out of a habit I cannot fully explain, the way my mother kept esfand smoke in the house on the last Wednesday of the year and never threw away the little clay dish. The forms go in a folder with the lease and the vaccination records. Three of them now. Three different answers to the same question.
What the form actually asks
The ancestry question on the ACS reads: What is your ancestry or ethnic origin? Below it, examples — Italian, Jamaican, African American, Cambodian, Cape Verdean, Norwegian, Dominican, French Canadian, Haitian, Korean, Lebanese, Polish, Nigerian, Mexican, Taiwanese, Ukrainian. Then two write-in lines. The Census Bureau codes up to two ancestries per person; if you write three, the first two are generally taken (Census Bureau, Ancestry FAQ). The question was added to the census form in 1980, so the earliest data from it begin there (About Ancestry).
The race question is separate. It offers checkboxes — White, Black or African American, American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Some Other Race — and a write-in area. The Bureau’s own page states the rule plainly: an individual’s response to the race question is based upon self-identification, and the Census Bureau does not tell individuals which boxes to mark or what heritage to write in (About the Topic of Race).
That is the whole architecture. A box, a line, a code. The rest is what a household brings to it.
Three answers, one household
I am not going to pretend I have the exact forms in front of me, dated and notarized. What I have is a memory of three mailings and three different ways of filling the same two lines. The first time, I wrote Iranian on the first line and left the second blank. The second time, I wrote Iranian and then, after a pause, Persian — a word my grandmother used and my father did not. The third time, I wrote Iranian and American, because by then I had been here long enough that the second word was not a lie, only an incomplete one.
None of these was false. That is the point. The Bureau’s rule is self-identification, not verification. It does not ask for a passport, a birth certificate, a grandfather’s shenasnameh. It asks what you say you are, and then it codes what you said.
The gap between the writing and the coding is where the essay lives. I wrote Persian once. The code list may or may not have a separate entry for it; the tabulation may fold it into Iranian, or into a broader category, or into Other ancestries. I do not know, and the published tables will not tell me which of my three answers ended up where. What I know is that the household record and the published record are not the same document.
The MENA question, and the clock it runs on
Here is a concrete example of how the form changes on a slower clock than the household does. In the 2020 Census, a distinct Middle Eastern and North African category was not included, based on the 1997 Office of Management and Budget standards. But two MENA examples — Lebanese and Egyptian — were added to the White racial category, along with a new dedicated write-in response area, for the first time in the history of the decennial race question (Coritz, Marks, and Jacobs, POP-WP108, December 2024).
The 2020 Census was also the first time households were invited to respond online (Census Bureau). So the same decade in which the form moved to a screen was the decade in which the write-in line became, for MENA respondents, the only place to say what the checkboxes could not.
The results were published in September 2023. The Lebanese (685,672), Iranian (568,564), and Egyptian (396,854) alone-or-in-any-combination populations represented nearly half of the 3.5 million respondents who reported a MENA response in the 2020 Census (Census Bureau press release CB23-CN.156, September 21, 2023). That figure — 568,564 — is the number my third answer helped to make, or did not, depending on how the write-in was coded. I cannot trace my own line to that total. No household can. The published count is an aggregate, and the aggregate is what the state sees.
In 2024, OMB updated the standards. The new rule requires a Middle Eastern or North African minimum reporting category and a combined race/ethnicity question (POP-WP108). The Census Bureau has said it will implement the combined question and the MENA category in collections such as the 2027 American Community Survey and the 2030 Census. That is a plan, not a completed collection. The form my household will fill in 2027 will not be the form I filled in 2020. The category will have moved. My answer may not have.
What the ancestry question is not
The Bureau is explicit about this: the intent of the ancestry question is not to measure the degree of attachment the respondent had to a particular ethnicity (About Ancestry). A response of Irish might reflect total involvement in an Irish community or only a memory of ancestors several generations removed. The form records a report, not a belonging.
That distinction matters more for a diaspora household than for almost anyone else, because the report is often the only official trace. There is no consulate queue for ancestry. There is no bank transfer that returns unexplained. There is a line on a form, two of them, and a code list that decides what the line becomes.
The two-line limit is its own kind of constraint. I have a friend — this is a real person, not a hypothetical — who wrote Iranian and Jewish on her form and then spent a week worrying that the second word would be coded as a religion and folded into Other ancestries. The Bureau does not tabulate data on religious groups; if people write in a religious group as an ancestry response, it is included under Other ancestries (Ancestry FAQ). She knew this. She wrote it anyway. The form asked for two lines, and she had more than two things to say.
The folder
I keep the copies because the published tables will never show what I wrote. They will show a count, a percentage, a category that may or may not match the word I put on the line. The folder is my own record — not a legal document, not a proof of anything, just a household’s account of what it said.
The form is a domestic object. It arrives in an envelope. It sits on the kitchen table next to the bills. It asks you to compress a life into two lines and a checkbox, and then it takes that compression and turns it into a number that a state agency will use to decide where a clinic goes, or a school, or a language assistance program. The number is real. The life behind it is not in the number.
I will fill out the 2027 form when it comes. I do not know yet what I will write. I know I will keep a copy.
Questions readers ask
Does the census ask for proof of ancestry? No. The Census Bureau states that responses to the race question are based on self-identification and that it does not tell individuals which boxes to mark or what heritage to write in (About the Topic of Race).
Can I write more than two ancestries? The ACS response area has two write-in lines, and the Bureau codes up to two ancestries per person. If you report more than two, the first two are generally taken (Ancestry FAQ).
Was there a MENA box on the 2020 Census? No. A distinct MENA category was not included in the 2020 Census under the 1997 OMB standards, though Lebanese and Egyptian examples were added to the White category with a new write-in area (POP-WP108).
When will the MENA category appear? OMB updated the standards in 2024 to require a MENA minimum category and a combined race/ethnicity question. The Census Bureau has described implementation in collections such as the 2027 ACS and the 2030 Census; those collections have not yet occurred (POP-WP108).
Where can I see the actual ACS questions? The Census Bureau publishes sample ACS and PRCS forms and instruction guides from 1996 to the present (Sample ACS & PRCS Forms and Instructions).
