Who counts in 2030?
A new Census Bureau proposal would change who is included in the population used to divide House seats and would remove race and ethnicity questions from the core decennial enumeration form. The rule is consequential, contested—and, for now, only a proposal.
The proposal in plain English
A census rule that separates enumeration from representation
For more than two centuries, the decennial census has been built around a deceptively simple assignment: determine where people usually live, count them there, and use state population totals to apportion the 435 seats in the U.S. House. The rule published by the Commerce Department on September 10 would redraw that connection.
Under proposed 15 CFR Part 60, foreign citizens would be treated differently for congressional apportionment. A foreign citizen who is also a U.S. citizen or a lawful permanent resident on April 1 of the census year would be included according to the residence rules. Other foreign citizens would not be counted for the purpose of determining representation in Congress. The text is broader than a rule aimed only at undocumented immigrants: its dividing line is citizenship or lawful permanent residency on Census Day.
That distinction matters because a person can be lawfully present in the United States without holding a green card. Temporary workers, international students and others admitted in temporary categories are not lawful permanent residents. Some refugees and people granted asylum may also be living in the country before they obtain permanent residence. The proposed regulation does not create a separate apportionment exception for those categories; its operative test is whether the foreign citizen is also a U.S. citizen or lawful permanent resident on April 1.
This is a proposed rule, not the rule governing the 2030 Census today. Public comments are due October 13, 2026. The Commerce Department would still have to review the record and issue any final rule, and litigation is widely expected if the proposal survives in materially similar form.
The proposal also changes the residence concept itself. The Census Bureau’s 2020 rule generally located a person where they lived and slept most of the time as of Census Day. The new proposal says a person’s “usual residence” would be the place where the person lawfully spent the greatest number of days during the roughly three months before April 1, and says that determination should be consistent with and evidenced by tax records. That language could make administrative records much more important to the 2030 count.
The administration describes the package as a move toward constitutional fidelity, efficiency, lower response burden and less collection of sensitive information. Critics see a different risk: creating one population for statistical purposes and a narrower one for representation, while asking the Census Bureau to reconstruct important demographic detail through surveys and administrative sources instead of the universal decennial questionnaire.
Current practice vs. proposed practice
The line moves from “where you live” to “what status you hold”
Residence first
The Census Bureau said the resident population included citizens and noncitizens living in the United States. Foreign citizens living here were counted at the place where they usually lived and slept; short-term foreign visitors were generally not counted.
Status affects apportionment
For foreign citizens, only U.S. citizens and lawful permanent residents would enter the apportionment population. Other foreign citizens would be excluded from the count used to determine each state’s House seats.
The proposal’s wording is therefore easy to misstate. It does not simply say that every noncitizen disappears from every Census Bureau product. The operative exclusion in Part 60 is specifically tied to determining representation in Congress. The decennial operation could still need to identify people who are not in the apportionment population, both to administer the count and to support other statistical products. Exactly how those parallel populations would be processed and published is one of the implementation questions not fully answered by the short regulatory text.
Nor is the potential effect limited to states with the largest undocumented populations. Apportionment is a zero-sum allocation of 435 seats. A relatively small difference near the final seat cutoff can shift a seat from one state to another. What matters is not just how many people are excluded nationally, but where they live, how the final population totals compare across states, and where the priority-value formula used for apportionment places the last few seats.
The policy question is not merely how many people are counted. It is whether the same count should serve both as a portrait of the country and as the constitutional denominator for representation.
A useful historical illustration comes from a Pew Research Center analysis of the 2020 census. Pew modeled what might have happened if unauthorized immigrants alone had been excluded from apportionment. In that model, California, Florida and Texas each would have ended up with one fewer House seat than under the total-population count, while Alabama, Minnesota and Ohio would have gained or retained a seat. That exercise is not a forecast for 2030. The new proposal uses a broader status line than “unauthorized,” and the population distribution in 2030 will be different. But it shows why even a small percentage change can matter at the seat margins.
A second change
Race and ethnicity would leave the core enumeration form
The other half of the proposal is less visible in the headline but just as important for the data ecosystem. Proposed Part 70 says the decennial short form and other enumeration questionnaires would not ask a respondent’s race or ethnicity. It would also prohibit a sexual-orientation question on those enumeration forms. Biological sex, date of birth and household relationship could still be asked, while other demographic questions would not be mandatory.
That requires an important clarification. The 2020 decennial census did ask detailed race and Hispanic-origin questions. It did not ask a sexual-orientation question. So the proposal would remove race and ethnicity from the core decennial form, while the sexual-orientation provision is a forward-looking prohibition rather than the deletion of a question Americans answered in 2020.
Race and Hispanic-origin data from the decennial census are not just descriptive statistics. The Census Bureau’s 2020 redistricting files included block-level tables for race, Hispanic or Latino origin, voting age, housing occupancy and group-quarters population. States used those population counts to redraw congressional and state legislative districts. Courts, civil-rights lawyers, researchers and local officials also rely on detailed race and ethnicity data when they evaluate whether electoral maps or public policies have unequal effects.
The proposed rule does not ban the Census Bureau from collecting demographic information everywhere. It explicitly leaves room for the American Community Survey and other non-enumeration surveys to ask demographic questions. It also allows the bureau to receive demographic information from administrative sources. That means “the census will stop collecting race data” is too broad. The more precise question is where the data would come from, at what geographic precision, with what statistical uncertainty and on what timetable.
The proposal would bar race and ethnicity questions from the core enumeration questionnaire that reaches households for the once-a-decade count.
The ACS could continue to ask demographic questions. But it is a sample survey, not a full-population enumeration, and every estimate carries sampling uncertainty.
The Bureau could use information from government records and other sources. The proposal emphasizes those records but does not resolve every question about completeness, consistency or local detail.
The distinction between a census and a sample survey is central. The Census Bureau says the ACS currently samples about 3.5 million housing-unit addresses each year. Those responses are weighted to represent the broader population, and the resulting estimates come with margins of error. That is a sound and well-established statistical system, but it is a different instrument from a universal decennial questionnaire designed to enumerate the population on a common reference date.
For 2030, the technical challenge would be to preserve the geographic detail and legal usefulness of race and ethnicity statistics if the enumeration form no longer directly supplies those answers for every responding household. The proposed regulation does not provide a complete production blueprint for that task. That gap is likely to be a major focus of comments from states, civil-rights organizations, researchers and data users.
What changes—and what does not
Representation, funding and local planning are related, but they are not the same pipeline
Coverage of census fights often collapses several consequences into one sentence: House seats, federal money, school planning and business decisions. In practice, those systems use different datasets and formulas. The proposed Part 60 change directly addresses the population used for congressional apportionment. It does not by itself rewrite every federal grant formula or order agencies to ignore every noncitizen in every statistical product.
Still, the decennial census is the base layer for much of the federal statistical system. The Census Bureau uses decennial results to anchor population estimates; the ACS uses population controls derived from official estimates; agencies and researchers combine census geography with survey and administrative data. If the government maintains more than one population concept—one for apportionment, another for resident-population statistics—users will need clear documentation to know which denominator they are seeing.
That is why claims that a state would automatically “lose billions” under the proposal are premature unless they identify a particular program and formula. Some federal funding streams use population, poverty, age or other characteristics from ACS estimates; others use program enrollment or administrative caseloads. The proposal could have downstream effects through the statistical system, but those effects cannot be responsibly reduced to one national dollar figure today.
The same caution applies to redistricting. House seats are apportioned among states using state totals. District lines within states are then redrawn under separate constitutional and statutory rules using detailed population data. Removing people from the federal apportionment population does not automatically answer the separate legal question of what population a state may or must use when drawing districts. The proposed regulation itself does not resolve the population base for intrastate redistricting, and future disputes over 2030 data products could raise separate questions depending on how the final system is designed.
The trust problem
The census is a counting operation, but participation is also a social contract
Any major questionnaire or eligibility change has to be evaluated not only for its legal theory but for its effect on response behavior. Census operations work best when households understand why the government is asking a question, believe their answers are protected, and can complete the form without uncertainty about who belongs on it.
The citizenship-question fight before the 2020 Census provides a useful warning against simplistic assumptions. A 2019 Census Bureau test found no statistically significant overall difference in self-response between forms with and without a citizenship question, though the mail response rate was modestly lower on the citizenship version and some subgroups showed small differences. The test also had limits: it did not include the full nonresponse follow-up operation needed to measure final coverage. In other words, neither “a status question obviously destroys response” nor “it has no effect” is supported by that test alone.
The 2026 proposal approaches the issue from another direction. Instead of requiring a citizenship item on the short form, it points toward administrative records and status information that can be used to build the apportionment count. That may reduce what a household must answer directly. It also makes record linkage, coverage and data-governance questions more important. A person can be present in one administrative system and missing from another; names and addresses can be outdated; tax records are designed for taxation, not necessarily for locating every person at a census residence.
The proposed residence definition makes that tension explicit by saying usual residence should be consistent with and evidenced by tax records. For many households, those records may align neatly with where people live. For others—students, people who moved recently, families with complex living arrangements, seasonal workers or people without a filing requirement—the relationship can be less tidy. The final methodology will need transparent rules for conflicts between questionnaires, administrative records and field evidence.
What happens next
There is a long road between a Federal Register notice and the 2030 count
The immediate next step is public comment. The proposal was published in the Federal Register on September 10 with comments due October 13. Individuals, states, local governments, advocacy groups, researchers and other organizations can submit arguments and evidence about the legal authority, operational design, data quality, privacy implications and costs. Those comments become part of the rulemaking record the department must consider.
The proposed rule is published, opening a formal notice-and-comment process for changes to apportionment residence rules and decennial demographic questions.
The current public-comment period closes. The proposal is not automatically final on that date.
Commerce can revise, withdraw or finalize the rule. A final version would need its own explanation and effective date, and major legal challenges would likely follow.
The Census Bureau must turn policy into tested questionnaires, administrative-record systems, residence procedures, field operations and data products capable of supporting a national count.
Readers should also distinguish this rulemaking from the Census Bureau’s recently completed 2026 Census Test. That field test in selected parts of Huntsville, Alabama, and Spartanburg, South Carolina, concluded August 31 and used a broader questionnaire that included citizenship along with race, sex, age and education. A test questionnaire is not the final 2030 questionnaire. The new regulatory proposal may influence future design choices, but it does not convert every experimental item in the 2026 test into settled policy.
Legal arguments will center on constitutional text, congressional statutes, executive authority and historical practice. The Fourteenth Amendment says representatives are apportioned according to the “whole number of persons in each State,” while the administration argues that the Constitution and federal law allow the proposed distinction. Because the proposal directly affects representation, any final rule is likely to receive fast and intense judicial scrutiny. The outcome should not be predicted before courts see a final agency action and a concrete implementation record.
The practical questions are just as important. Can the Bureau accurately classify status at national scale? How will it treat conflicting records? How will it construct block-level race and ethnicity data if those questions are no longer on the enumeration form? How will it explain multiple population totals to states and the public? And how will it maintain public confidence while using more linked administrative data?
Those are not side issues. In 2030, a technically successful census must do two things at once: produce numbers that withstand legal and statistical scrutiny, and persuade hundreds of millions of people that the count is understandable, legitimate and worth participating in.
Reader FAQ
What the proposal does—and does not—settle
Would undocumented immigrants be excluded from House apportionment?
Yes, if the rule is finalized as proposed. But the rule is broader than that category: it says foreign citizens who are not U.S. citizens or lawful permanent residents would not count for apportionment.
Would every noncitizen be excluded?
No. Lawful permanent residents would remain in the apportionment population under the proposal. The regulation also refers to foreign citizens who are U.S. citizens, such as dual citizens, as countable under the residence rules.
Would the census stop asking about race everywhere?
No. The proposed prohibition is for decennial enumeration questionnaires. It explicitly allows the American Community Survey and other non-enumeration surveys to ask demographic questions, and it allows administrative sources to provide demographic data.
Does this already change the number of House seats?
No. The 2020 apportionment remains in effect. The proposal concerns how the 2030 apportionment population would be built, and it has not been finalized.
Can we know which states would gain or lose seats?
Not reliably yet. The 2030 population, migration patterns and immigration-status distribution are still changing. Historical models can illustrate sensitivity, but they are not forecasts of the proposed 2030 system.
Would federal funding automatically use the narrower apportionment count?
No single rule controls every program. Funding formulas use different datasets and eligibility rules. The proposal directly changes congressional apportionment; downstream funding effects would have to be assessed program by program.
The bigger picture
A fight over the meaning of one national number
The census has always been both administrative machinery and democratic infrastructure. It counts households, creates geographies, supplies population baselines and, once every decade, helps reset political representation to match a country that has moved and changed. That is why seemingly technical definitions—usual residence, household membership, citizenship status—can become constitutional disputes.
The 2026 proposal is unusually consequential because it changes two foundational choices at the same time. First, it would narrow the set of people included in state apportionment totals by immigration status. Second, it would reduce the demographic questions asked on the universal enumeration form and shift more responsibility to surveys and administrative records. Either change would be significant on its own; together, they would reshape how the Census Bureau moves from individual records to the public numbers that states, courts and communities use.
For now, the most useful discipline is precision. The rule has been proposed, not adopted. It changes the apportionment population, not necessarily every resident-population dataset. It removes race and ethnicity from the enumeration questionnaire, not from every Census Bureau survey. It could shift political representation, but no responsible analyst can name the 2030 winners and losers with confidence four years before Census Day.
What can be said with confidence is that the choice of who belongs in the apportionment number will shape the next decade of political geography. The public-comment period is the first formal test of whether the administration’s legal theory and operational design can withstand scrutiny. The next tests—statistical, logistical and judicial—will determine whether the proposal becomes the architecture of the 2030 Census or another chapter in the long American argument over who counts.
Primary documents and reporting
- Federal Register: Residence and Demographic Questions Used in the Decennial Census (published Sept. 10, 2026)
- U.S. Census Bureau: 2020 Census Residence Criteria and Residence Situations
- U.S. Census Bureau: 2020 Census P.L. 94-171 Redistricting Data
- U.S. Census Bureau: American Community Survey sample-size definitions
- Pew Research Center: 2020 historical apportionment model excluding unauthorized immigrants
- U.S. Census Bureau: 2026 Census Test, concluded Aug. 31, 2026
- Reuters: independent reporting on the 2030 Census proposal
Comments
Post a Comment