The Trump Administration’s Catastrophic Census Proposal
Photograph by Nathaniel St. Clair
The Trump administration recently announced significant changes to the 2026 Operational Test for the 2030 Census. If implemented, the changes would undermine the utility of what should be a broad, multi-site rehearsal of census operations, potentially compromising the validity of the 2030 Census results. The changes transform the test from a safeguard into a source of additional risk and uncertainty.
The sweeping changes to the Operational Test include a reduction in the number of test sites from six to two; the replacement of the decennial short form with the much longer American Community Survey (ACS) questionnaire, with questions reordered so that a question on citizenship appears earlier; and a pilot to evaluate the use of United States Postal Service (USPS) employees in place of trained Census Bureau enumerators. Any one of these changes would compromise the scientific validity of the exercise. Taken together, they threaten to thoroughly degrade the test’s utility as a planning tool to inform preparations for 2030.
Operational tests are designed to evaluate the mechanics of conducting a nationwide census — canvassing, mailings, online response systems, field operations, enumerator hiring and training, outreach strategies, and quality control procedures. Testing helps identify weaknesses and improve accuracy ahead of the constitutionally mandated decennial census. The original design of the 2026 test reflected that mission. It included six geographically and demographically diverse sites: Colorado Springs, CO; Huntsville, AL; Spartanburg, SC; Western North Carolina; Western Texas; and tribal lands in Arizona. These locations were selected after years of research to stress-test census operations in environments that are historically difficult to count, including rural communities, tribal lands, military installations, and regions with limited cell phone service and few traditional mailing addresses.
The revised plan eliminates four of those six test sites, leaving only Huntsville and Spartanburg. This reduces the expected respondent pool from 631,850 to just 154,600. It also sharply reduces the diversity and representativeness of the subject pool. The census has a long history of undercounting specific populations, including people with disabilities, American Indians, young children, and predominantly Black and Latino communities. The elimination of these sites reduces the Bureau’s ability to evaluate their operations in communities where undercounts have historically been more pronounced, increasing the risk that such failures remain undetected and persist at the national level.
The Citizenship Question
The revised test also replaces the decennial census short form with the full ACS questionnaire. The ACS is far longer and collects different information than the decennial questionnaire. The ACS also takes respondents longer to complete, which in turn results in fewer people completing it. An earlier test design estimated that the short-form census questionnaire could be completed in about 10 minutes. By contrast, the Census Bureau estimates that the ACS takes people an average of 40 minutes to complete. And while that difference alone is likely to affect response behavior, the reordering of questions introduces an additional, and arguably more sinister, source of error.
The Trump administration made a concerted effort to add a citizenship question to the 2020 Census. They were ultimately thwarted in doing so by the Supreme Court, which ruled in Department of Commerce v. New York that the administration’s rationale for adding the question was contrived and pretextual. Research conducted during the litigation of that case demonstrated that the mere prospect of a citizenship question depressed participation intentions among immigrant households and mixed-status families, raising the risk of differential undercounts. This time, the Trump administration appears to be pursuing a similar objective through different means. They are substituting the ACS, which includes a citizenship question, and reordering the instrument so that the citizenship item appears earlier. Because the ACS has already been authorized under Title 13, substituting it into the test removes the need to separately justify adding a citizenship question to the decennial short form.
Revising the 2026 Operational Test to include a reordered ACS questionnaire does more than revive a controversy that federal courts had ostensibly already resolved in the lead-up to the 2020 Census. The substitution creates a methodological problem. The purpose of operational tests is to evaluate processes, not survey content. If response rates decline, it will be impossible to determine whether the cause was the operational systems the test was intended to evaluate or the questionnaire’s increased length and complexity. Reordering the questions compounds that problem. Introducing and foregrounding a sensitive citizenship question would make it impossible to isolate the participation effects of operational factors from those of fear and distrust stemming from the questionnaire’s design. Testing operational mechanics while simultaneously altering questionnaire length and structure makes it impossible to isolate the effects of each variable. A test that cannot clearly identify what it is testing is neither sound nor useful.
Turning USPS Workers Into Census Enumerators
Replacing dedicated census enumerators with USPS workers threatens to exacerbate the aforementioned fear and mistrust, with good reason. Census enumerators operate under Title 13 confidentiality protections that strictly limit data sharing and impose criminal penalties for violations; they are legally obligated to treat respondents’ information as confidential for statistical use only. Title 13 does not apply to USPS employees, who are governed by more general federal privacy laws. Practically speaking, this means that an administration that has already demonstrated a willingness to blur privacy firewalls between agencieswould be operating under a framework that lacks the census-specific statutory protections designed to keep respondents’ identifying information strictly walled off. Against a backdrop of the administration’s escalating rancor toward immigrants, the elevation of a citizenship question without Title 13 protections could understandably chill participation among immigrant households.
The idea of using mail carriers instead of census enumerators is not new. Policymakers previously expressed interest in leveraging the USPS’s existing footprint to reduce costs during planning discussions for the 2010 Census and the early design phases of 2020. However, Government Accountability Office findings from 2011 indicated that using mail carriers at USPS pay rates would be less cost-effective than hiring temporary census enumerators. USPS workers are also unionized, and their scope of work and compensation are strictly defined in their collective bargaining agreements. Expanding their job duties to include census enumeration could therefore require significant renegotiation with postal unions. While the Trump administration has attempted to curtail or disband many federal employee union protections in the name of “national security,” these efforts have been challenged in court and partially blocked.
To add to the chaos, the revised notice outlining these sweeping changes also shortened the public comment period from 60 to 30 days. This limits meaningful public scrutiny of changes to a test with long-lasting consequences for representation and the distribution of federal resources.
The goal of operational testing is to reduce uncertainty and correct errors before a nationwide launch. The revised 2026 test would do the opposite. Each proposed change reduces the utility of the test and, in turn, the integrity of the 2030 Census, with lasting consequences for who is seen, counted, and represented. An inaccurate and unrepresentative census would distort political representation and skew the allocation of trillions of dollars in federal funding that are tied to census counts.
Changing who gets counted in the census reshapes political power by design, not by chance. By sidelining historically underrepresented communities and weakening the integrity of the test, the Trump administration is treating the census as a means to consolidate political power through exclusion. Should these changes proceed, we risk a census that entrenches representational and resource disparities until at least 2040.
A fair and accurate census is a constitutional pillar of representative democracy in the United States. To achieve this in 2030, the administration must restore and recommit to the original scope and rigor of the 2026 Operational Test.
This first appeared on CEPR.
