IX. Group Quarters and Other Special Living Situations

Not everyone lives in a house or apartment that can be reached through the ordinary census mailing process. Millions of people live in group quarters such as college residence halls, nursing facilities, and correctional institutions. Others stay in recreational vehicle parks, marinas, hotels, or other transitory locations. People experiencing homelessness may be counted through shelters, service locations, or outdoor enumeration. Each of these settings requires Census to identify the place correctly and use an operation suited to the people living or staying there.

These populations are not simply additions to the ordinary household count. A group quarters facility may house hundreds or thousands of people, and an error can affect the count and location of every resident. Census must determine whether a place is a housing unit, group quarters, or transitory location; identify the type of facility; obtain accurate information from its administrator; and select an appropriate enumeration method.

The work also raises distinct questions about individual participation. In some settings, residents may be able to respond for themselves. In others, Census may rely heavily on facility administrators or electronic records. The quality of the resulting data can depend on whether facility records are current, whether residents are counted at the correct location, and whether information about race, ethnicity, age, sex, and other characteristics comes from the individual or from an institutional file.

The consequences of an incomplete frame are especially serious. If Census fails to identify an ordinary housing unit, a resident may still respond without a Census ID or a fieldworker may discover the address. If an entire group quarters facility, encampment, or transitory location is absent from the appropriate frame, no specialized enumeration may be scheduled at all.

This chapter examines how Census plans to identify and classify these locations and how it will count the people associated with them. Although the settings differ, the central questions are consistent: Did Census find the place, classify it correctly, and give each person a meaningful opportunity to be counted at the right location?

A. Special Populations Address Frames

Census maintains specialized address frames for group quarters and transitory locations because these places cannot always be identified or classified accurately through ordinary residential address sources. Examples of group quarters include nursing facilities, adult group homes, and college housing. Transitory locations include recreational vehicle (RV) parks, marinas, and hotels or motels where people may be living without another usual residence.

For 2030, Census plans to strengthen these frames throughout the decade rather than constructing them primarily near the time of enumeration. The Bureau is evaluating federal records, facility data, business information, private sources, and information collected from the internet. It also plans a targeted Group Quarters Validation operation for locations whose status remains uncertain.

Federal and Facility Records

Government and facility records can provide large, regularly updated lists of places that house special populations. The Operational Plan identifies potential sources such as Centers for Medicare and Medicaid Services data on nursing homes and information about correctional facilities from the Census of Jails. Census also expects to obtain administrative and supplemental data from organizations that oversee networks of group quarters or transitory locations.

These records could help Census identify facilities earlier and reduce dependence on a short pre-census listing operation. They may also provide useful information about facility type, location, capacity, and organizational relationships.

But coverage will vary. Federal records are likely to be strongest for regulated facilities participating in federal programs. They may be weaker for small group homes, privately operated residences, temporary facilities, or places that have recently opened, closed, or changed their use. A facility can exist in a database while its address, classification, or operating status is out of date.

Census should evaluate each source separately by facility type and geography. A high overall matching rate could conceal significant omissions among the facilities least connected to federal reporting systems.

Web Scraping

Census is also exploring web scraping, which would use automated processes to extract publicly available information about group quarters and transitory locations from websites. The Bureau states that scraped information would be combined with administrative and supplemental data and verified before it is used to update the frame.

Web scraping could identify places that do not appear promptly in government records. A newly opened residential program, campground, or extended-stay property may advertise online before it appears in a formal dataset. Websites may also help Census determine whether an address remains active or has changed its name or purpose.

The method has predictable limits. Facilities with a strong and current internet presence will be easier to find than small, informal, or poorly resourced operations. Websites can contain outdated information, and search results may confuse an administrative office with the place where residents actually live. Online language may also describe a facility for marketing or regulatory purposes without using Census’s living-quarters definitions.

Census should publish how it tests the accuracy and completeness of scraped information. Web scraping should be used to identify possible additions or changes, not as an unreviewed substitute for facility contact or field validation.

Business Records

The Bureau is evaluating its Business Register and other government or commercial records as possible sources for the special-populations frames. It plans to examine matching rates, duplication, timeliness, and operational feasibility before deciding whether a source is fit for use.

Business data may help identify hotels, recreational vehicle parks, residential care businesses, and other establishments that could contain group quarters or transitory populations. These sources may also reveal closures, ownership changes, or new establishments.

But a business establishment is not necessarily a place where people live. A corporate headquarters may be listed instead of the residential facility, and a business code may not distinguish short-term customers from long-term residents. Conversely, some living arrangements may operate through religious, nonprofit, or informal entities that do not appear clearly in conventional commercial data.

Census should therefore avoid treating a successful business-record match as proof that a facility is correctly located and classified. Business records are useful leads, but they must be reconciled with information about how the property is actually used.

Private and Nonprofit Sources

Private and nonprofit organizations may maintain information that is not available in federal or business records. National associations, service networks, accrediting organizations, universities, advocacy groups, and facility operators may know about the locations they serve or represent. The Operational Plan specifically contemplates data from private organizations and from umbrella organizations covering groups of facilities.

These sources could be especially helpful for identifying small or specialized facilities. Community organizations may also recognize temporary shelters, residential programs, and other locations that government records overlook.

Census should use clear and transparent standards when assessing these sources. An organization’s list may cover only members, funded programs, or licensed facilities. It may also use definitions that differ from Census classifications. The Bureau should document which populations each source includes, which it excludes, and how conflicting information will be resolved.

Community and nonprofit input should not be limited to supplying databases. Census should create a practical way for organizations to flag a missing location or an apparent classification error, even when confidentiality restrictions prevent them from reviewing the full address frame.

Group Quarters Validation

Administrative records and web research cannot always determine whether an address is a housing unit, a group quarters facility, a transitory location, or a nonresidential property. For 2030, Census anticipates restoring a limited Group Quarters Validation field operation for locations flagged as needing additional review. Specially trained staff would visit those places before enumeration and determine how they should be classified.

This is an important safeguard. Misclassification can send a location into the wrong operation. A group home treated as an ordinary housing unit may receive only one household questionnaire. An apartment building incorrectly classified as group quarters may deny residents the ordinary opportunity to respond as separate households. A facility that has closed may waste resources and obscure the need to find where its former residents now live.

The Operational Plan describes the proposed 2030 effort as “GQV-lite,” focused only on group quarters and transitory locations tagged as requiring validation. That targeted approach may be efficient, but its success depends on the system used to identify uncertain cases. A location cannot be selected for validation if the underlying data sources fail to find it or incorrectly classify it with high confidence.

Census should explain the criteria used to flag locations, the facility types most likely to receive field review, and the results of validation. It should also track false confidence: places that were not flagged but were later found to be missing, closed, or misclassified.

The strongest design would use these sources as overlapping checks rather than substitutes for one another. Federal records may identify regulated institutions, web searches may reveal recent changes, and nonprofit partners may locate smaller facilities. Group Quarters Validation can then resolve cases that cannot be settled reliably from a desk.

Potential benefit

Continuous use of multiple data sources could produce a more timely and complete list of group quarters and transitory locations and allow Census to focus fieldwork on uncertain cases.

Credible risk

The frame performs well for large and highly regulated facilities but misses smaller, newer, informal, or less digitally visible places. Because those locations are never flagged, targeted validation does not reach them.

Open design questions

Which data sources will Census adopt, how will it measure their coverage across facility types, and what evidence will trigger Group Quarters Validation?

Why advocates should care

When a special living situation is missing or misclassified, the error can affect every person associated with that location and may prevent the correct enumeration operation from occurring at all.

B. Self-Response in Group Quarters

For 2030, Census is developing an internet self-response option for residents of certain group quarters, such as college and university housing. This would be a significant change from 2020, when group quarters residents could not respond for themselves online. Instead, facilities generally chose among methods that included individual interviews, paper questionnaires, or administrator-provided resident files.

Self-response will not be appropriate in every setting. Census research has identified college housing, workers’ quarters and Job Corps centers, and religious group quarters as particularly promising candidates. Institutional settings may present different questions about residents’ capacity to respond, access to technology, privacy, and facility control.

Potential Benefits

The clearest benefit is that residents can provide their own information. Facility records may contain names and basic demographics, but the completeness and detail of those records vary. An administrator may not have current or sufficiently detailed information about a resident’s race, ethnicity, or other characteristics, or the facility may limit which information it will share.

Direct response allows residents to describe their own identities rather than relying on categories recorded for another purpose. This may produce more accurate race and ethnicity information, particularly for people who identify with more than one group or use detailed identities that are not included in facility records.

Self-response may also improve information about the person’s living arrangement, usual residence, and other characteristics. Where the questionnaire asks about ties to another household or relationships relevant to determining where a person should be counted, the resident may know more than the facility administrator.

Just as importantly, self-response gives residents greater control. A person can see the questions, decide how to answer them, and submit information without asking an administrator to report on their behalf. The system should preserve that independence by ensuring that facility staff distribute invitations but cannot view or alter residents’ answers.

Residents Must Know That They Need to Respond

Self-response works only if residents understand that the invitation is legitimate and that completing the questionnaire is their responsibility. People living in a dormitory, religious residence, or workers’ quarters may reasonably assume that the facility will count everyone automatically.

Census should develop communications specifically for group quarters residents. The materials should explain:

• why the person is receiving an individual invitation;

• whether the facility will also provide a resident list;

• where the person should be counted;

• how to avoid being counted again at another address; and

• how to obtain language or disability assistance.

The invitation should not depend solely on an email from an unfamiliar sender. Messages may be filtered as spam, sent to an inactive account, or overlooked among routine facility communications. Census research has therefore recommended pairing the response system with programming that explains why the census matters and why residents should respond at the group quarters location.

Facilities Must Facilitate Participation

Even though residents would respond individually, Census will continue to depend on facility administrators. Under the model Census has been testing, administrators would provide information about the facility and supply residents’ names and email addresses so Census could send individual invitations.

Facilities may vary greatly in their ability and willingness to support the process. Some will maintain current electronic contact information and have staff available to answer questions. Others may have outdated records, limited administrative capacity, high resident turnover, or concerns about sharing contact information.

Census should establish clear responsibilities for facilities and provide training, technical support, and reminders. Administrators should be expected to distribute or facilitate access to Census invitations, provide private internet access where needed, and explain how residents can obtain assistance. They should not be allowed to discourage participation or substitute facility data simply because individual response requires more coordination and effort.

Census will also need an alternative for residents who lack an email address, reliable device, private internet access, or the ability to use the online questionnaire. Paper and in-person options should remain available, and facilities should not select internet self-response unless they can support meaningful access.

Duplicate Counting

Individual response may increase the risk that a person is reported more than once. A student could respond from a residence hall while also being included on a parent’s household questionnaire. A facility might submit a complete resident file even after some residents have responded individually. A person who moves during the enumeration period could receive invitations connected to more than one location.

Census should treat duplication as a processing problem to be managed, not as a reason to limit self-response. Invitations should clearly explain the residence rules, and the questionnaire should ask enough information to identify likely duplicate records. Facility lists, individual responses, and responses from other households must then be reconciled without automatically preferring one source.

A matching system should not discard a valid self-response merely because the person also appears in an administrator file. The individual response may contain more complete and accurate characteristics. Census should preserve the best available information while ensuring that the person contributes only once to the population count.

The Bureau should publish its rules for identifying and resolving possible duplicates and evaluate whether those rules work equally well across names, cultures, and family arrangements. Matching that relies too heavily on exact names or conventional household ties may fail for people who use multiple names, recently changed their names, or are reported differently by relatives and institutions.

Administrative Fallback Data

Some residents will not respond individually. Census will therefore need a fallback, which may include facility-provided files or other administrative data. These sources can be valuable for establishing that a person was present, but they may not contain all the information requested by the census.

A college file, for example, may include a student’s name and age but incomplete race or ethnicity information. Other records may use categories developed for program administration rather than the final census questionnaire. Information about sex, residence, or other characteristics may be outdated or supplied by someone other than the resident.

Census should distinguish between records that are sufficient to count a person and records that are sufficient to supply each characteristic. A reliable indication that someone lived in a facility should not be treated as evidence that every demographic field in the same file is equally current or accurate.

Fallback data should also be evaluated for differential quality. Residents who do not respond may differ systematically from those who do. If their records are also less complete, the resulting data could be most limited for the people already least likely to participate directly.

Census should report self-response rates by group quarters type and examine the completeness of administrator and administrative fallback data. It should also explain when missing characteristics will trigger follow-up, use of another source, or imputation.

Potential benefit

Self-response could produce more accurate race, ethnicity, residence, and other characteristic data while giving group quarters residents greater control over how they are represented.

Credible risk

Residents do not realize that they need to respond, facilities provide inconsistent support, and Census falls back on records that establish a person’s presence but contain incomplete or outdated characteristics.

Open design questions

Which group quarters will offer self-response, how will residents be invited, what alternatives will be available, and how will Census reconcile individual responses with facility files and reports from other households?

Why advocates should care

Self-response can make group quarters enumeration more individual and more accurate. But without clear communications, facility cooperation, accessible alternatives, and careful duplicate processing, it could instead create an uneven system in which some residents speak for themselves while others are represented by incomplete institutional records.

C. College Housing and Military Barracks

College housing and military barracks are promising settings for individual self-response because many residents have regular access to digital systems and can provide more complete information about themselves than an administrator may have. The Operational Plan identifies both as possible users of an internet questionnaire designed for group quarters residents.

For college housing, the proposed model is relatively well developed. A housing administrator would use a Census system to provide information about the facility and the names and email addresses of residents. Census would then send each resident an invitation to complete an individual questionnaire online. Research with group quarters administrators found college housing to be among the strongest candidates for this approach, although it also emphasized the need for communications explaining why students should respond at their college address.

The model could improve data quality by allowing students to report their own race, ethnicity, and other characteristics. It could also reduce the burden on colleges that otherwise must compile a complete resident file. But the process depends heavily on the institution. Colleges must provide current contact information, distribute reminders, make technology and assistance available, and explain that students should not also be reported at a parent’s home.

Census should not assume that every student checks an institutional email account or understands the residence rules. Invitations should be sent through more than one channel where possible, and students should receive a clear explanation that they are generally counted where they live and sleep most of the time. Matching systems must reconcile student responses with facility rosters and household responses without discarding the more complete information supplied by the student.

Military barracks present some similar opportunities but different operational risks. A military installation can provide an organized roster and communicate quickly with residents. At the same time, personnel may deploy, relocate, or move between barracks near Census Day. A strong chain of command can promote participation, but it may also make the census feel mandatory in a way that reduces privacy and individual control.

The public research released so far provides stronger support for online response in college housing, workers’ quarters, and religious quarters than for military barracks, even though the Operational Plan lists barracks as an example of a possible self-response setting. Census should therefore publish separate test results for military housing rather than assuming that the college model transfers directly.

Both settings need a reliable backup process. The Operational Plan retains in-person interviews, dropped-off paper questionnaires, electronic resident files, and paper resident lists. Military group quarters may also use facility self-enumeration, in which sworn facility administrators distribute and collect questionnaires.

The backup should fill gaps without displacing valid individual responses. Census should first identify who responded, then use an administrator roster or administrative record for the remaining residents. Where the fallback record lacks detailed characteristics, Census should not treat every missing field as equally reliable merely because the person’s presence has been confirmed.

Central design question

Can Census combine individual response and facility records so that each resident is counted once, at the correct location, using the most complete information available?

D. Correctional Facilities

Correctional facilities may require a different approach because residents have limited control over their communications, movement, and access to technology. The current Operational Plan allows correctional administrators to provide resident-level electronic or paper records and permits facility self-enumeration. It also plans to use person-level records from the Criminal Justice Administrative Records System, or CJARS, when quality problems are identified in collected data. CJARS receives group quarters and person-level criminal justice data on a recurring basis.

Administrative data could help identify missing facilities, check facility population totals, and recover people omitted from an incomplete submission. This is potentially important because the 2020 Census included thousands of prisons, jails, detention centers, juvenile facilities, and other correctional group quarters. Census research comparing correctional sources found high overall facility coverage, but an overall match rate does not establish that every resident or demographic characteristic was recorded accurately.

Residence Rules

Under the 2020 residence rules, adults in federal and state prisons were counted at the correctional facility where they were held on Census Day. Census has historically treated that location as their “usual residence” for the decennial count, even when the person maintains family, legal, or community ties elsewhere – and even when the person will be in the facility only for a very short duration (i.e., days, weeks).

That rule has major consequences for redistricting and local population counts. Some states subsequently reallocate incarcerated people for their own redistricting processes, but the underlying Census enumeration must still place each person somewhere accurately. Census should publish the 2030 residence rules early, explain how they apply to short-term detention and transfers near Census Day, and preserve information needed to identify duplicate or incorrect placements.

Individual Self-Response

The current plan does not clearly promise internet self-response for correctional facilities. The published self-response research has focused on noninstitutional settings, while the Operational Plan continues to list facility self-enumeration as an option for correctional institutions.

Nevertheless, Census should examine whether residents can provide at least some information privately and directly. An individual questionnaire may be feasible through paper, a secure device, or an interview conducted outside the hearing of correctional staff. Direct participation is particularly important for information that may not be recorded accurately in institutional files.

Self-response must be genuinely confidential. Correctional personnel should not review answers, stand close enough to hear an interview, or learn whether a resident declined to answer a particular question. Participation procedures should make clear that Census responses cannot affect classification, discipline, release, immigration status, or access to programs.

Race and Ethnicity Accuracy

Correctional records may use race and ethnicity categories developed for identification, booking, or facility administration. Those categories may be assigned by staff, carried forward from older records, or lack the detail available on the Census questionnaire.

Census race and ethnicity data are intended to reflect individual self-identification. Facility or CJARS records may be useful as a fallback, but Census should evaluate whether they reproduce the resident’s own response and whether their categories correspond to the standards in effect for 2030.

Census should report characteristic completeness by enumeration method. A facility may provide an accurate head count while supplying incomplete or systematically different race and ethnicity information. Those are separate dimensions of quality and should not be collapsed into one measure of a successful response.

Privacy and Retaliation Concerns

Correctional settings create an unusually large power imbalance between the respondent and the institution facilitating the census. Residents may reasonably fear that providing information, requesting privacy, or declining assistance will draw unwanted attention. Even when Census data are legally protected, the method of collection may expose responses to facility staff or other residents.

Census should develop correctional-specific privacy protocols with formerly incarcerated people, civil rights organizations, facility administrators, and correctional oversight experts. Staff training should address private interviewing, literacy and disability access, language assistance, and the distinction between a census questionnaire and a correctional record.

Central design question

Will residents of correctional facilities be represented primarily through institutional records, or will Census provide a safe and meaningful opportunity for them to identify and describe themselves?

E. Homeless Shelters, Service-Based Enumeration, and Targeted Outdoor Enumeration

Census uses specialized procedures to count people experiencing homelessness because many cannot be reached through an ordinary household mailing. These procedures include enumeration at emergency and transitional shelters, community meal programs, regularly scheduled mobile food vans, and outdoor locations where people are known to live or sleep.

In 2020, Census commonly used Service-Based Enumeration (SBE) to refer to the count at shelters and service locations and Targeted Non-Sheltered Outdoor Locations (TNSOL) to refer to the count at preidentified outdoor locations. TNSOL was also treated administratively as part of the broader SBE operation.

The 2030 Operational Plan no longer identifies TNSOL as a separately named activity. Instead, it describes SBE as covering people in transitional shelters, people receiving services at meal programs or mobile food vans, people living in outdoor encampments, and people who are not counted elsewhere. Census should clarify whether the TNSOL name and distinct field procedures will continue within SBE and how the outdoor enumeration component will be designed.

Improving the Service-Location and Outdoor-Location Lists

The quality of the operation begins with its list of shelters, meal programs, mobile service stops, and outdoor locations. Census should update that list throughout the decade using information from HUD, Continuums of Care, tribal and local governments, service providers, street-outreach organizations, and people with direct experience of homelessness.

Formal databases will not capture every relevant location. Small faith-based programs, temporary warming centers, informal food distributions, newly established encampments, and places serving people outside formal homelessness systems may be absent. Census should create a simple process for organizations and residents to propose additions without requiring them to navigate a technical group quarters system.

Outdoor locations require particular attention. The places where people live can change quickly because of weather, construction, law-enforcement activity, encampment clearances, or changes in access to public space. A location identified months earlier may no longer be occupied, while a newer location may never enter the workload.

Engagement With Service Providers and Outreach Organizations

Service providers can help determine when residents are present, how the operation should be explained, and which staff members or outreach workers should participate. Organizations conducting street outreach can provide similar guidance for outdoor enumeration, including whether a proposed visit time is realistic and how Census staff should approach the location.

Engagement should begin well before Census Day. Providers need to understand what information will be requested, how confidentiality works, and whether Census plans to conduct individual interviews or use a facility-provided list. They should also have time to inform residents without creating the impression that access to shelter, food, or other assistance depends on census participation.

Census should distinguish between asking community organizations for operational guidance and asking them to disclose sensitive information. Organizations should not be pressured to provide detailed maps or records that could later expose where people sleep or receive services.

Coverage Outside Formal Service Locations

A homelessness count limited to shelters and service programs would miss people who do not use those systems. The outdoor component formerly identified as TNSOL is therefore essential, not an optional extension of shelter enumeration.

Census should work with street-outreach teams, public libraries, health providers, mutual-aid groups, and people with lived experience to identify places that formal homelessness systems may not know. It should also examine how to reach people who sleep alone or in small, dispersed locations that may not qualify as a recognizable encampment.

The purpose should not be to create a lasting federal map of where unhoused people live. Location information should be collected only to the extent necessary for enumeration, protected against unrelated use, and removed from operational systems when it is no longer needed.

Timing and Mobility

The plan anticipates conducting SBE over the same three days nationwide. A short collection period can limit duplication and provide a common reference point, but it can also miss people who move frequently or are not present when Census visits a location.

Timing should reflect differences among settings. Shelters may have predictable overnight rosters, while meal programs operate during the day and outdoor locations may be safest and most accurate to visit at particular hours. Weather, displacement, or changes in service schedules can alter where people are found from one day to the next.

Census should allow repeat visits or other remediation where a scheduled enumeration is disrupted. Quality measures should distinguish between completing an assigned visit and successfully reaching the people expected at that location.

Risks of Uniformed Federal Staff

Enumeration should not be conducted alongside law-enforcement activity or by personnel whose clothing, vehicles, or behavior could be confused with immigration enforcement, policing, or encampment-clearing operations. People may leave before an interview begins or avoid a service location altogether if they believe the federal government is present for another purpose.

Census workers must be identifiable as legitimate, but identification should be paired with trusted outreach and a clear separation from enforcement agencies. Whenever possible, a known service provider or outreach worker should introduce the Census team and explain the limited purpose of the visit.

Census should train workers assigned to outdoor locations in trauma-informed engagement, de-escalation, disability access, and local conditions. The objective should be to offer people a safe opportunity to be counted, not simply to complete a geographic assignment.

Open design question

Will Census retain a distinct TNSOL-style outdoor enumeration component within SBE, and how will its procedures differ from enumeration at shelters and service locations?

Credible risk

Folding outdoor enumeration into a broadly defined SBE activity could obscure the separate planning, partnerships, training, timing, and privacy protections required to reach people who do not use formal services.

Why advocates should care

People living outdoors cannot be counted adequately through shelter and service-provider lists alone. Census must preserve a visible, well-resourced operation designed specifically to find and safely enumerate people outside formal service locations.

F. Intimate Partner Violence Shelters

Shelters for survivors of intimate partner violence require their own enumeration protocols. Their locations may be confidential, residents may be actively avoiding an abusive person, and an unexpected visit, mailing, or disclosure can create a direct safety risk.

The Operational Plan discusses shelters within the broader group quarters and service-based operations but does not identify a separate protocol for confidential survivor housing. Earlier Census ethnographic research found that these facilities can be difficult to classify, that their administrative records vary in reliability, and that staff may be reluctant or legally unable to share information. It recommended tailored procedures and communication through a single Census point of contact.

Confidential Location Information

Census should maintain a restricted list of confidential shelters that is separate from generally accessible operational address files. The precise address should be available only to personnel who need it to perform the enumeration. It should not appear in public maps, partner lists, training examples, contractor reports, or published descriptions of group quarters locations.

Census should coordinate all contact through an authorized shelter representative. No worker should arrive without advance agreement, and no follow-up should be sent to a survivor’s former residence, family home, or another address that could reveal their location.

Staff Training and Resident Safety

Workers assigned to these facilities should receive specialized, trauma-informed training. They should understand that residents may use alternate names, may be uncertain where they will live next, and may not want to provide an address associated with a spouse or family member.

The enumeration method should be chosen with the shelter, but resident safety and autonomy should remain central. A facility-provided list may reduce contact with outsiders, while private individual response may give survivors more control over identity and household information. Census should support both options and avoid pressuring a shelter to use the method that is easiest administratively.

Administrative-Record Use

Administrative records at a shelter may be intentionally limited. They may contain aliases, omit prior addresses, or separate a resident’s identifying information from service records for safety reasons. Census should not assume that an incomplete match means the resident is not eligible to be counted or that conflicting records should be resolved through broader data searches.

Any use of administrative data should be discussed with survivor-service experts and assessed for the possibility that matching could reconnect identities, addresses, family members, or prior locations that the shelter deliberately keeps separate. The safest method may sometimes be a minimal facility count with resident-provided characteristics rather than extensive linkage.

Postal Employees and Other External Personnel

Census should state explicitly whether postal employees, contractors, commercial data vendors, or other outside personnel could receive or infer the locations of confidential shelters. Even when another organization normally helps update the address frame or deliver census materials, survivor shelters may require an exception.

Location information should not be shared merely because an external partner is involved elsewhere in address maintenance. Census should identify who is legally bound by Title 13, who can access shelter-level data, and what procedures prevent the location from entering an ordinary mailing, canvassing, or contractor workload.

Advocacy priority

Census should publish a distinct protocol for confidential survivor housing, developed with survivor-led and domestic-violence organizations, before the special-populations frame and field procedures are finalized.

G. Transitory Locations, SROs, Maritime Vessels, and Overseas Populations

These populations are grouped together operationally, but they present very different enumeration problems.

Transitory locations include recreational vehicle parks, campgrounds, marinas, hotels, motels, hostels, and similar places. Census counts people at these locations when they do not have another place where they live or stay most of the time. For 2030, Census plans to move this field collection to an electronic instrument. The main challenges are identifying occupied spaces, applying the usual-residence rule consistently, and distinguishing a temporary visitor from someone for whom the site is effectively home.

Single-room occupancy units, or SROs, are classified as housing units under the current 2030 living-quarters definitions. They should therefore enter the ordinary housing-unit address frame rather than be treated automatically as hotel rooms or group quarters. The risk is misclassification, particularly where a building contains a mix of permanent SRO residents, short-term guests, and shelter placements.

Maritime vessels depend on complete vessel lists and cooperation from operators. Under the current plan, Census contacts maritime agencies and vessel operators, sends questionnaire kits to the operators, and relies on them to distribute and return individual forms. This creates questions about reaching vessels that are at sea, identifying who actually lives aboard, and allowing crew members to respond without the operator reviewing their answers.

Federally affiliated people overseas are counted through administrative records supplied by federal agencies. This operation covers U.S. military personnel and federal civilian employees stationed or assigned overseas, along with dependents living with them. The records are used to assign that population to a home state for congressional apportionment. It is not a census of every U.S. citizen living abroad. Data quality depends on agency records containing current dependent counts and the correct home state.

Across these operations, Census should avoid treating facility or operator cooperation as a substitute for individual access where self-response is feasible. It should also report separately on frame completeness, residence-rule errors, duplicate resolution, and missing characteristics.

H. Island Areas

The Island Areas Censuses cover American Samoa, the Commonwealth of the Northern Mariana Islands, Guam, and the U.S. Virgin Islands. They are conducted separately from the stateside census and Puerto Rico, using questionnaires similar to the American Community Survey. Their responses are processed and tabulated separately, producing data products specific to each Island Area.

Census conducts each operation through a contract agreement with the local government. The Census Bureau provides funding, methods, materials, guidance, and technical systems, while the local government recruits, hires, and trains the workforce. Census plans earlier involvement by Island Area governments, improved communications and staffing plans, updates to questionnaire content, and some use of stateside electronic systems.

Local-Government Capacity

This structure makes local capacity central to census quality. Each government must recruit and manage a temporary workforce, secure offices and equipment, conduct field operations, and meet federal data-security requirements. Census should assess capacity early and provide sustained technical support rather than expecting each area to solve staffing or systems problems after the contract is signed.

Questionnaire Content and Language

Because the Island Areas do not receive the ACS, their decennial questionnaires collect much more detailed social, economic, and housing information than the short stateside census form. Local governments and data users should have a meaningful role in determining which questions are needed and whether stateside wording fits local institutions and living arrangements.

Language planning must also reflect each Island Area’s population rather than simply importing the stateside language program. Questionnaires, training, outreach, and respondent assistance should be available in the languages residents use, with adequate testing of local terminology.

Procurement and Funding

Contract and procurement delays could compress hiring, training, technology deployment, and community engagement. Census should complete agreements early, identify which systems or vendors are shared with the stateside census, and establish contingency plans if those procurements slip.

Funding must account for island-specific transportation, shipping, communications, workforce, and technology costs. A nominally equal or historically based budget may not be sufficient where infrastructure costs or local labor conditions have changed.

Timely Data Products

Separate processing should not mean that the Island Areas wait years for usable census results. Census should publish a detailed product and documentation schedule, provide progress reports, and ensure that local governments have timely access to the data needed for planning and federal funding.

The product program should also be developed with local users. A technically complete set of tables may still be inadequate if it does not provide the geographic detail, cross-tabulations, or documentation needed in a particular Island Area.

Why advocates should care

The Island Areas Censuses are not small versions of the stateside operation. They are separately negotiated, locally implemented censuses that require early federal commitments, locally appropriate content, sufficient funding, and equal attention to the timely release of useful results.

I. Worst-Case Scenario

The 2030 design creates real opportunities to improve the enumeration of people living in group quarters and other special situations. Better facility frames, electronic data collection, resident self-response, and administrative records could make the count more complete and efficient. The risk is that these tools become substitutes for meaningful resident participation rather than ways to support it.

In the worst-case scenario, Census successfully identifies most large facilities and obtains a plausible population total from administrators, federal records, business data, or other institutional sources. Automated quality checks confirm that the number reported for a college, prison, nursing facility, shelter, or military installation is close to the expected population. From the perspective of the total count, the operation appears successful.

But the information about the people included in that total is much weaker. Facility records may establish that a person was present without containing complete or current race, ethnicity, age, sex, or other characteristics. The categories may have been assigned by an employee, copied from an older record, or collected under standards that differ from the 2030 Census questionnaire.

Individual self-response is available in some noninstitutional group quarters but is not implemented consistently. Some colleges and facilities distribute invitations effectively, while others provide outdated email addresses, offer little explanation, or assume that an administrator file is sufficient. Residents who do not understand that they must respond are represented through the backup record rather than through their own answers. Census’s research agenda recognizes both the need to determine whether residents understand the self-response process and the need for a reliable method of remapping incomplete cases to another collection method.

In correctional institutions and other highly controlled settings, residents receive little or no private opportunity to answer for themselves. Facility records and CJARS data provide names and population counts but reproduce institutional classifications of race and ethnicity. People who would describe themselves differently have no practical way to correct the record. The current plan expressly anticipates using CJARS person-level data when quality problems are found in correctional-facility responses.

Shelters and outdoor enumeration locations are visited during a compressed three-day Service-Based Enumeration period. The operation reaches enough people to produce a defensible national estimate, but mobility, incomplete location lists, limited repeat visits, and fear of federal personnel leave substantial gaps. Information collected in brief or public encounters is also less complete than information residents might provide through a private questionnaire. The Operational Plan currently places shelters, meal programs, mobile food vans, outdoor encampments, and people “not counted elsewhere” within SBE.

Similar problems arise in transitory locations, on maritime vessels, and among federally affiliated people overseas. Census obtains files or questionnaires through facility administrators, vessel operators, or federal agencies, but residents have limited visibility into what was reported about them. The process produces a count while leaving uncertainty about usual residence, duplicate enumeration, and the completeness of individual characteristics.

The resulting weaknesses are difficult to see in national quality measures. A facility may have a complete population count even when half of its residents lack self-reported characteristics. A state may appear to have strong group quarters coverage even though the data for small facilities, confidential shelters, and people living outdoors are much less complete.

Automated processing then fills remaining gaps through edits, substitution, imputation, or administrative records. These procedures may be necessary to produce a complete dataset, but they can create the appearance of precision without reflecting what residents would have said about themselves.

The final census therefore counts approximately the right number of people in institutions and nontraditional living situations while misrepresenting who those people are. Published data understate some racial and ethnic groups, contain less detail for people counted through institutions, and make it difficult for data users to distinguish self-reported information from institutional or administrative substitutes.

Because these errors are concentrated in settings occupied disproportionately by people with disabilities, students, incarcerated people, people experiencing homelessness, older adults, and others whose living arrangements already limit their control, the quality gap is not random. The census effectively provides one level of representation to people living in conventional households and a weaker level to people whose information is supplied by an institution.

Worst-case scenario

Automation, facility records, and administrative data allow Census to produce plausible population totals for group quarters and special living situations, but meaningful resident participation becomes secondary. The resulting characteristics are incomplete, outdated, or institutionally assigned, and national quality measures conceal the unequal quality of the data.

Why advocates should care

Being included in the total population is essential, but it is not the same as being represented accurately. Census should evaluate whether people counted through institutions, shelters, and other special operations have the same meaningful opportunity as household respondents to describe who they are.

Where to Look: See the 2030 Census Operational Plan, particularly section 3.1.2, “Special Populations Address Frames,” pp. 21–22, for the use of government, facility, business, private, and online sources to identify group quarters and transitory locations, along with the proposed targeted Group Quarters Validation operation. Section 3.2.4, “Group Quarters and Special Populations Enumeration,” pp. 34–37, covers resident self-response, facility files, correctional-facility records, Service-Based Enumeration, Enumeration at Transitory Locations, Maritime Vessel Enumeration, and the Federally Affiliated Count Overseas. Section 3.2.5, “Island Areas Censuses,” pp. 37–39, describes the separate agreements, local-government responsibilities, address updating, enumeration, and possible automation for the four Island Areas.

The 2030 Census Research Project Explorer includes an entire enhancement area on Modernizing Group Quarters Enumeration. Particularly relevant projects include Stakeholder Engagement, Living Quarters Definitions, and Service-Based Enumeration; Group Quarters and Transitory Locations Frame Sources; Improving Group Quarters Validation; Automating Group Quarters and Transitory Locations Operations; Create Internet Self-Response Option for Non-Institutional Group Quarters; and Electronic Response Enhancements to Facilitate Group Quarters Administrators’ Response. The Explorer also includes separate projects on Island Areas enumeration systems and questionnaire content. Advocates should watch whether project findings report results separately by group quarters type rather than drawing conclusions from overall performance.

For the most useful 2020 evidence, see the operational assessments for Group Quarters Advance Contact, Group Quarters Enumeration, and Service-Based Enumeration, Enumeration at Transitory Locations, and the Island Areas Censuses. The group quarters assessment includes the 2020 experience with shelters and TNSOLs, while the transitory-locations assessment recommends improving the location frame, clarifying definitions, conducting field testing, and automating collection. The Island Areas assessment documents lessons from the locally implemented, entirely paper-based 2020 operation.

Census’s working papers on online individual response in group quarters provide the clearest public research on which facility types may be suitable, how administrators would supply resident contact information, and what communications and backup procedures are needed. Advocates should also watch future versions of the Operational Plan for distinct protocols governing correctional facilities, confidential survivor shelters, outdoor enumeration, privacy protections, and the quality of characteristics obtained through administrative fallback data.