The next Census could undercount the number of disabled Americans by 20 million

About 20 million disabled people will be erased if the U.S. Census Bureau moves forward with changes to disability data collection methods. That is because many disabled people will no longer be counted as disabled by the new questions the Census is proposing to use starting in 2025 with the annual American Community Survey (ACS).

As disabled people, and as scholars who study disability measurement and use disability data for our research, we have grave concerns about this proposed change. If the Census changes its disability questions it will artificially reduce national estimates of disability almost by half, by the bureau’s own estimate. Any effort to undercount disabled people is alarming as the prevalence of disability is rising, not declining, as a result of the Covid-19 pandemic.

Researchers and policymakers rely heavily on Census data from the decennial Census, as well as from the yearly ACS. Census data are part of informed decisions about federal, state, and local government allocation of resources and are also used to develop evidence-based policies for education, housing, infrastructure, health, support services, and emergency response.

The Census has historically undercounted certain minoritized groups, particularly Black, Hispanic, and American Indian populations. Recognizing the problem that these undercounts present, the Census invests substantial time, energy, and dollars into trying to ensure that they “count everyone.”

Yet the proposed changes to the disability questions, which were initially presented during a Census Scientific Advisory Committee in September 2023, will, as the proposers acknowledge, lead to an undercount of disabled people — the largest minority group in the United States.

The Census currently uses the ACS disability questions, which is the standard set by the Department of Health and Human Services. The ACS disability questions were created to measure disability status among the U.S. population and added to the Census in 2008 in response to concerns from the disability community. Since then, these disability questions have been used across 17 nationally representative surveys and have provided consistent estimates of the disabled population in the U.S. They ask respondents to answer yes or no to whether they have difficulty with six functional tasks (vision, hearing, mobility, cognition, self-care, independent living). Those who answer yes to difficulty with any of these six functional tasks are automatically counted as disabled. Any changes to this approach to counting disability must be considered carefully, as it will undercount disabled people and remove opportunities to examine trends in data over time.

The Census is proposing to use a different set of disability questions, the Washington Group (WG) questions, which were first developed in the early 2000s and are currently used, despite evidence that they undercount the disabled population, in two U.S. surveys: the National Health Interview Survey and the National Nutrition Examination Survey. Instead of asking a straightforward yes or no question, the WG asks respondents to report the level of difficulty they have with the same six functional tasks. The WG then uses a pre-determined cut-point to indicate what level of functional difficulty is considered disabled. Thus, a disabled person who reports having difficulty with one of the six functional tasks included in the WG questions, but at a lower level than the pre-determined cut point assigned by the WG, would not be considered as disabled. In effect, the WG questions allow those who own and use the data to act as gatekeepers for deciding how disabled is disabled enough to be counted.

The WG’s approach to collecting data disability is contradictory to the Census’ goal of counting everyone. The WG explained that when creating their questions: “It was not our purpose to identify every person with a disability within every community.” It is therefore not surprising that multiple studies have found that the WG questions perform worse than the ACS disability questions in counting disabled people. In fact, the Census found similar results during testing of the WG questions that was conducted in 2020-2022 — the WG undercounted disabled people, reducing the estimated size of this population by 42% — from 14% to 8% (see slide 12). If that holds true in the Census, that would mean removing 20 million Americans from the ranks of the disabled. That will result in a drastic slashing of resource allocation for a population that faces profound and unaddressed inequities and health disparities. However, the Census’ plan was to only “communicate” these changes to the disability community after making them (see slide 12), which is far from community engagement.

The ACS questions are not perfect, either. Like the WG approach, the ACS questions fail to accurately count large groups of people with disabilities, including people with mental health, learning, and intellectual and developmental disabilities, and people with chronic disabling conditions such as long Covid. The exclusion of these groups of disabled people result in inaccurate estimates of disability and must be a priority for Census to address. Ultimately, if the disability questions used by the Census accurately counted all of these groups, the Census should be counting more Americans as disabled, not fewer.

Under the Affordable Care Act, disabled people must be considered a core demographic group. Census policy is in alignment with that approach and specifically uses disability data to “plan and fund programs for people with disabilities. Disability data are also used to evaluate other government programs and policies to ensure that they fairly and equitably serve the needs of all groups, as well as enforce laws, regulations, and policies against discrimination.” Undercounting disabled people will negate these efforts by preventing full understanding of the interventions and services needed to ensure equity for disabled people in the U.S., and lead to a reduction in funding and resources needed to reach that goal.

At a recent public meeting of Census National Advisory Council (NAC) it became clear that the U.S. disabled community was not engaged regarding these changes. This meeting also highlighted the lack of scientific justification for this change, especially considering the Census’ goal of accurately counting demographic groups. During their debriefing session, the NAC members indicated shock that the disability community had not been more involved and recommended against adopting the proposed change to the WG questions prior to or without first consulting with a representative cross section of diverse and intersectional members of the disability community.

We share the NAC’s concerns and urge the Census to stop efforts to change the disability data questions and instead invest in developing an improved approach to disability data collection. That requires actively engaging with the U.S. disability community to develop and roll out new measures. Disabled people count. We must be actively included to ensure that we are accurately counted.

Bonnielin Swenor, Ph.D., MPH, is the director of the Johns Hopkins Disability Health Research Center. Scott D. Landes, Ph.D., is an associate professor of sociology at Syracuse University.