In a world driven by data, accurate population counts form the foundation for informed decision-making. As the U.S. Census Bureau prepares for the crucial task of counting our population in the next census, it’s imperative to shine a spotlight on the people and households that we have the most difficulty enumerating. In our efforts to develop the best methods to do just that, we have found it essential to conceptualize two population groups: the hard-to-count (HTC) populations, and, a critical subset of those, the historically undercounted populations (HUPs).

These are two distinct concepts that mean different things. Following recommendations received from stakeholders, we recognized that these terms are commonly misunderstood, sometimes used interchangeably and often used differently than we intend. By sharing how we at the Census Bureau distinguish between the two and how it leads to improved, more relevant approaches, I hope we can have more productive discussions and engagements.

In this blog, I’ll clarify how the Census Bureau defines HTC populations and HUPs. I’ll also explore why certain communities are harder to enumerate and how we are working to address underrepresentation in the count through our 2030 Census research.  

We refer to a population as HTC when our traditional methods of counting may not be sufficient to fully include them in the census. Understanding the needs helps us identify how we need to adapt to count everyone.

When describing a population as HTC, the Census Bureau considers a framework that places HTC populations and their households into four segments:

• Hard to locate. People who live in homes that are not currently on the Census Bureau’s list of addresses as well as people who want to remain hidden.
• Hard to contact. People who are highly mobile, are experiencing homelessness, or for whom physical access barriers prevent contact, such as in gated communities.
• Hard to persuade. People who are suspicious of the government or have low levels of civic engagement.
• Hard to interview. People whose participation is hindered by language barriers, low literacy, health issues, or technological barriers (e.g., lack of internet access).

It’s important to note that some populations may fall into one, more than one, or all four segments of the HTC populations framework.

Examples of people who may fall into one or more categories of HTC populations include (but are not limited to):

• Young children.
• Racial and ethnic minorities.
• People who speak languages other than English.
• Low-income populations.
• Undocumented immigrants.
• People with disabilities.
• People who do not live in traditional housing.

We've identified these groups as hard to count because the Census Bureau acknowledges these challenges to counting everyone accurately. This framework helps us understand the challenges, explore them through research, and then adapt our approaches to reach these populations.

We have tried to overcome these challenges in a wide variety of ways, such as working with local governments to update our address list, partnering with trusted community leaders to encourage response, providing content in numerous languages, and providing multiple ways to respond. More information about the HTC framework and examples of how we tried to reach everyone in 2020 are available in the appendix of the 2020 Census Operational Plan.

Now I’d like to shift focus to discussing a critical subset of the HTC — historically undercounted populations. For this population, we restrict the definition to those we can measure in our decennial quality assessments and therefore have census coverage data available for them. To identify these groups, we use a more technical definition that considers data from official measures of coverage, such as estimates from the Post-Enumeration Survey (PES) and Demographic Analysis, or other appropriately validated and reviewed research.

The PES and Demographic Analysis each develop estimates of the size of the total U.S. population and certain demographic groups. We then compare those independent estimates to the census counts. If the census count is lower than the PES or Demographic Analysis estimate, we call that a “net undercount.”

Using these coverage estimates, we define HUPs as populations with a definitive net undercount over time.

For example, according to the PES conducted for the 1990, 2000, 2010, and 2020 Censuses, the following groups have been persistently undercounted:

• The Black or African American population.
• The American Indian and Alaska Native population living on reservations.
• People who indicate that they are some other race than the categories offered.
• The Hispanic or Latino population.
• Young children, ages 0-4.
• Renters.
• Males, ages 18-29 and ages 30-49.

Given these distinct definitions, it’s important to note that the terms HUPs and HTC are not interchangeable. Again, HTC is an umbrella term, while HUPs are a subset of the HTC populations. Both provide useful lenses upon which to conduct specific research and planning for the 2030 Census. 

In other words, not all HTC populations are historically undercounted. While we may have challenges locating, contacting, persuading or interviewing members of a group, we may not consider the group an historically undercounted population because:

• We haven’t measured a definitive undercount — or we’ve sometimes even measured an overcount — across censuses. For example, while we may have challenges accurately counting members of the Asian population, they are not consistently undercounted in aggregate statistics. The PES did not measure a statistically significant undercount for the Asian population in 2010 and measured an overcount in 2020. However, we may still need to make additional effort to ensure that we accurately count the Asian population as a whole.
• We have not measured coverage for the population group. Groups may be undercounted but currently we have insufficient data for measurement.

As it becomes more challenging to locate individuals, connect with them, and ask for their participation through traditional methods, we are dedicated to understanding the needs of populations who have historically been undercounted or are hard to count. We want to overcome barriers to participation to ensure that everyone living in the United States is counted in the census.

Several of our 2030 Census research projects primarily focus on reaching, motivating and accurately counting hard-to-count and historically undercounted populations, and many more projects include research on these populations as a component. To learn more about research on HUPs and HTC populations, check out the latest blog from Director Robert Santos, “How an Interwoven Research Agenda Will Help Us Enumerate Historically Undercounted People in the 2030 Census.”

By the way, while much of our research for the 2030 Census is focused on addressing undercounts, many of these same research projects have the side benefit of addressing issues related to overcounts. For example, one of our enhancement areas is dedicated to real-time data processing, which will help us identify and resolve duplications in real time.

The importance of reaching HTC populations and HUPs cannot be overstated. The same is true of understanding and using these definitional concepts for effective research and planning for the 2030 Census. Counting HTC populations and HUPs is essential to our mission of serving as the nation’s leading provider of quality data about its people and economy. As we strive for a more inclusive future as a nation, let us endeavor to ensure that every voice is heard and that every person is counted. We welcome and appreciate your ongoing engagement in this effort.

As Director Santos said:“There is much research, testing and learning that lie ahead, and we’ll continue to communicate with you about our preparations.” It is crucial to gather a broad and varied range of perspectives to formulate the most effective strategy for the upcoming 2030 Census.

Until our next update, please visit our 2030 Census webpage to check out everything we are doing to prepare for the next census.