Your Mail Carrier May Ask Census Questions: Inside the Quiet Trial Shaping the 2030 Count

In two corners of the American South, the federal government is quietly testing a significant change to the way the United States counts its population.

Postal workers are no longer simply delivering census invitations. Some are knocking on doors, carrying Census Bureau identification and asking residents to complete lengthy government questionnaires.

The experiment covers approximately 154,600 selected homes in parts of the Huntsville, Alabama, and Spartanburg, South Carolina, metropolitan areas. Around 81,000 housing units were invited in Huntsville and 73,600 in Spartanburg. Households that did not respond online may receive an in-person visit before data collection ends on August 31. Some visits can occur as late as 9 p.m., including at weekends. (Census.gov)

Officially, this is the 2026 Census Test, the first of two major field tests intended to inform the design of the 2030 Census. Operationally, it is also a trial of something potentially much larger: whether the federal government can turn the country’s postal workforce into part of its census-taking infrastructure. (Census.gov)

The premise is intuitively attractive. Postal workers already travel through American neighbourhoods, know local streets and regularly encounter homes that can be difficult for temporary census staff to locate. They may also be more familiar to residents than an enumerator hired for a few months.

But familiarity is not the same as informed consent, and efficiency is not the same as accuracy.

Before this model is considered for nationwide use, the Census Bureau needs to establish whether postal enumeration improves response rates, reduces costs and reaches people who would otherwise be missed. It must also determine whether the arrangement creates confusion about who is collecting information, when they are acting as postal employees, and how residents can distinguish a legitimate visitor from an impersonator.

Those questions are not peripheral to the experiment. They are the experiment.

Two cities, two different versions of the postal census

The trial is sometimes described as though Huntsville and Spartanburg are running the same programme. They are not.

In Spartanburg, participating postal workers may collect census responses while completing their normal delivery routes. A household will generally be assigned either a postal worker or a conventional census taker, depending on its location. Residents are supposed to be told when the person collecting information is acting as a postal worker. (Census.gov)

Huntsville is testing a more complicated model. Postal workers are being hired directly by the Census Bureau to conduct follow-up visits outside their USPS working hours. During those visits, they identify themselves solely as Census Bureau employees. The area also has approximately 25 conventional census takers working among the same households, and the Bureau acknowledges that some homes may be visited by both a postal worker and a non-postal enumerator. (Census.gov)

That distinction matters.

Spartanburg tests whether census work can be integrated into the postal route. Huntsville tests whether the government can recruit people from the postal workforce and redeploy them as after-hours enumerators.

The first model relies on the visibility and familiarity of the mail carrier. The second relies on the carrier’s local knowledge while deliberately separating the census visit from the person’s postal duties.

Those approaches could produce very different outcomes in cost, productivity, public recognition and respondent trust. Any eventual evaluation should report the results separately rather than blending them into a single “USPS pilot” figure.

How residents can identify a legitimate visitor

The Census Bureau says every census taker and specially designated postal worker should carry two visible forms of official identification:

• a Census Bureau badge displaying the worker’s name and photograph;
• a Census Bureau-branded bag.

Postal workers participating in the pilot must undergo background checks and receive the same census training as conventional enumerators. The training includes confidentiality, data-security requirements and the use of Census Bureau-issued devices. They also take a lifetime oath to protect census information under Title 13 of the United States Code. (Census.gov)

That provides a basic verification framework. It does not eliminate the communication problem created by the trial.

In Spartanburg, a familiar postal worker may ask questions during a normal delivery round. In Huntsville, a person recognised as a postal employee may arrive later wearing Census Bureau identification and describe themselves only as a census worker. Some households may receive repeat visits from different people. Others may be approached during the evening by someone asking for sensitive demographic information on an electronic device.

Every element may be legitimate, yet the combination resembles the warning signs residents are often taught to associate with impersonation scams: an unexpected caller, an unusual role, personal questions and a visit outside ordinary business hours.

A successful security policy therefore requires more than issuing badges. It requires residents to know in advance what a valid encounter can look like.

The Bureau’s published material explains the general identification rules, but the larger test should examine whether residents actually understood them. An identification system that is technically correct but poorly recognised is not fully effective.

This is not a conventional census questionnaire

The questionnaire also deserves closer scrutiny.

Selected households were initially invited to respond through an online form accessible by computer, tablet or smartphone. No paper or telephone response option was offered, and the online questionnaire was available only in English. The Bureau estimated that completing it would take approximately 40 minutes. (Census.gov)

It was not the brief household form Americans usually associate with the decennial census. The test used questions from the American Community Survey, covering subjects including age, sex, race, Hispanic origin, citizenship, education and other demographic, social, economic and housing characteristics. (Census.gov)

That design creates an unusually demanding test of self-response.

A household needs the invitation containing its unique Census ID, internet access, sufficient digital confidence, the ability to read the form in English and enough uninterrupted time to complete a questionnaire expected to take around 40 minutes.

The consequences of that design should be measurable. The government should eventually disclose how nonresponse varied by age, language, disability, housing type, broadband availability and neighbourhood characteristics.

Without that analysis, a higher volume of in-person follow-up could be mistakenly interpreted as evidence that postal enumeration is necessary when some of the demand may have been created by the test’s own online-only, English-only entry point.

The Government Accountability Office has already identified language access as a significant change in the trial. Earlier plans called for Spanish and Chinese versions, but the final questionnaire was approved in English only. (GAO)

That makes it difficult to separate two questions:

1. Can postal workers successfully enumerate nonresponding households?
2. How many households became nonrespondents because the initial response system was comparatively restrictive?

A credible evaluation must answer both.

Why the experiment matters beyond two metropolitan areas

The decennial census determines far more than the number displayed beside a state’s name.

Census counts underpin the apportionment of seats in the House of Representatives and inform redistricting, federal statistics, public planning and the distribution of substantial government resources. The Constitution requires the recurring national count, making its field operations part of the United States’ democratic infrastructure rather than an ordinary administrative survey. (GAO)

Small operational choices can therefore produce political and financial consequences lasting a decade.

The Census Bureau has long struggled with differential coverage: some populations are more likely to be missed, while others are more likely to be counted more than once.

Its post-enumeration analysis found that the 2020 Census significantly undercounted the Hispanic or Latino population by an estimated 4.99% and the Black or African American population by approximately 3.30%. American Indian and Alaska Native people living on reservations were undercounted by an estimated 5.64%. By contrast, the non-Hispanic White population was overcounted by approximately 1.64%, and the Asian population by 2.62%. (Census.gov)

Children are another persistent problem. The Bureau estimated that approximately one million children aged four or younger were missed in 2020, representing an undercount of about 5.46%. Experimental estimates subsequently indicated that young children were undercounted in every state and in more than four-fifths of the counties covered by the analysis. (Census.gov)

The relevant standard is therefore not simply whether postal workers complete more interviews.

It is whether they reduce the disparities that repeatedly affect the census count.

A postal model that improves overall response while leaving historically undercounted communities behind would represent an operational gain but a democratic failure. Conversely, even a comparatively expensive model could be valuable if it substantially improves coverage among renters, young children, households with limited English proficiency or residents in hidden and unconventional housing.

Why postal workers might succeed

The Census Bureau’s theory is that postal workers possess two resources conventional enumerators often lack: local knowledge and established relationships.

That argument has practical force.

Postal employees may know where an address is difficult to reach, where multiple households receive mail through one entrance, where an apparently vacant property is occupied, or where an informal dwelling does not fit neatly into official address records.

The Census Bureau selected prospective test areas partly for characteristics associated with enumeration difficulty, including historically low response, rural geography, limited internet or mobile service, homes without direct mail delivery, seasonal vacancies, rapid housing growth, group quarters and hidden or complex housing units. (Census.gov)

USPS information has already played a role in census operations. During the 2020 Census, the Bureau used administrative records—including USPS mail-delivery failure information—to reduce unnecessary follow-up visits for approximately 13% of nonresponding addresses. For 2030, it plans to make earlier and more extensive use of administrative data when determining how to handle nonresponse. (GAO)

The new trial moves the relationship from data sharing and mail delivery into direct interviewing.

If it works, the postal network could provide an existing national workforce with unmatched geographic reach. It might reduce recruiting pressure, shorten travel, improve address recognition and make some residents more willing to open the door.

But those advantages remain hypotheses until the Bureau publishes comparative evidence.

The danger of measuring only what is easy

Government pilots often produce operational statistics—visits attempted, interviews completed, hours worked and cases closed—more readily than they produce evidence about trust, confusion and unequal access.

That would be insufficient here.

At a minimum, the evaluation should compare postal workers and conventional census takers on:

• completed interviews per assigned household;
• completed interviews per working hour;
• number of repeat visits required;
• refusals and partial interviews;
• cost per completed case;
• travel time and mileage;
• worker retention and absenteeism;
• safety incidents;
• complaints and suspected impersonation reports;
• response rates by neighbourhood and demographic characteristics;
• data quality, including missing or inconsistent answers;
• residents’ understanding of the visitor’s identity and employer;
• outcomes for households with limited English proficiency;
• outcomes before and after 8 p.m.

The evaluation should also distinguish Spartanburg’s on-route model from Huntsville’s after-hours model.

Combining them could hide the very trade-offs the test is supposed to reveal. For example, route-based interviews may cost less but create time pressure during mail delivery. After-hours enumeration may allow longer conversations but require duplicate hiring, separate scheduling and additional supervision.

The best-performing model may also vary by neighbourhood. A postal carrier could be particularly effective in rural communities or stable residential areas, while a specialised census team with language support may perform better in multilingual apartment districts.

A national census does not need a single method applied everywhere. It needs an evidence-based system capable of matching different methods to different barriers.

A major test that became narrower

The emphasis on postal enumeration is partly the result of a broader contraction in the 2026 testing programme.

The Census Bureau initially planned six test sites and 19 operational activities. The programme was later reduced to two sites, while ten planned activities were either removed or substantially curtailed. According to the Government Accountability Office, the remaining work was refocused around USPS in-field enumeration and changes to field infrastructure, staffing and training. (GAO)

GAO warned that the reduced test may not generate enough evidence for several decisions the Bureau must make before finalising the 2030 design. It said the resulting evidence gaps could create cost and quality problems and undermine congressional or public confidence. The Department of Commerce agreed with recommendations to test the omitted features and assess workforce requirements before the 2030 Census design is completed. (GAO)

That context increases the importance of the postal trial.

It is no longer one experiment among a broad portfolio of equally developed tests. It has become one of the principal operational priorities in a significantly narrowed programme.

The country therefore needs more than a promotional conclusion that postal workers “helped” collect responses. It needs enough information to decide whether the approach merits nationwide investment—and where it does not.

The records the public should see

The Census Bureau and USPS should release, proactively or through public-records requests:

The operational agreement

The memorandum, interagency agreement or other document defining responsibilities, staffing, payment, liability, supervision and data access between the Census Bureau and USPS.

Training materials

The complete training provided to participating postal workers, including modules on identification, confidentiality, handling refusals, personal safety, language barriers, device security and suspected fraud.

Cost data

Wages, overtime, training expenditure, equipment, mileage, administration and supervision costs for postal and conventional enumerators, presented in a genuinely comparable format.

Performance data

Assignments, contact attempts, completed cases, refusals, partial interviews, repeat visits and cost per completion, separated by worker type and test location.

Safety and security records

Complaints, threats, injuries, lost devices, suspected impersonation, police referrals, privacy concerns and incidents involving visits after 8 p.m.

Language-access procedures

Instructions for what an enumerator should do when a household cannot complete an interview in English, including any use of interpreters, translated materials or follow-up staff.

Evaluation criteria

The thresholds the Bureau established before the trial for judging whether postal enumeration is viable, scalable and cost-effective.

Releasing success criteria only after the results are known would allow the government to redefine success around whichever outcomes appear most favourable.

What residents should know now

A legitimate census visitor participating in this test should carry a Census Bureau badge bearing their name and photograph and a Census Bureau-branded bag. Postal workers conducting census interviews are subject to background checks, confidentiality rules and census-specific training. (Census.gov)

Residents should not assume that every legitimate visit will look identical.

In Spartanburg, a postal employee may collect information during an ordinary delivery route. In Huntsville, a postal worker hired separately by the Census Bureau may visit outside postal hours and identify themselves as a Census Bureau employee. Visits may take place during the day or evening and can occur as late as 9 p.m. before the test ends on August 31. (Census.gov)

That complexity is precisely why the government’s public-information campaign should be evaluated alongside the enumeration results.

Residents cannot be expected to detect impostors reliably unless officials clearly explain the legitimate variations first.

The real question for 2030

The Census Bureau is not merely testing whether postal workers can operate census devices and complete questionnaires.

It is testing whether one trusted public institution can lend its relationships, workforce and physical reach to another.

The opportunity is substantial. A properly designed partnership could help locate difficult addresses, reduce duplicated travel and reach households that disregard letters from an unfamiliar federal agency.

The risks are equally real. Blurred institutional identities may create confusion. Evening visits may reduce trust. An English-only online process may push avoidable numbers of residents into field follow-up. Familiarity with a postal worker may make some households more comfortable, while making others less willing to discuss citizenship, income, housing and household composition with someone they encounter regularly.

The verdict should not be ideological. Nor should it rest on anecdotes about whether Americans “trust their mail carrier.”

It should rest on transparent comparisons: who responded, who was still missed, what each completed case cost, what residents understood, what incidents occurred and whether the model narrowed or reproduced the census’s long-standing coverage gaps.

Until those results are published, the postal census remains a promising idea—not a proven solution.

And because the numbers collected in 2030 will shape representation and public decisions for the following decade, “promising” is not enough.