Assessing Miscounts in the 2020 Census

Research Report

Assessing Miscounts in the 2020 Census


The decennial census, which aims to count every US resident each decade, is critical to our democracy. Census population counts guide appropriations and federal funding allocations, congressional redistricting, state and local budgets, and data-driven business and research decisions.

But the 2020 Census faces unprecedented challenges and threats to its accuracy. Demographic changes over the past decade will make the population harder to count. And underfunding, undertested process changes, and the last-minute introduction of a citizenship question could result in serious miscounts, potentially diminishing communities’ rightful political voice and share of funding.

So far, there has been little evidence about the effects of these factors on the 2020 Census, particularly for populations who have historically been at risk of being undercounted. To understand how these factors could alter the 2020 Census, we used demographic projections and Census Bureau data to create projections of the potential counts—overall, by state, and by demographic group—under three scenarios, reflecting the miscount risk as low, medium, and high.

Key Findings

  • The undercount of the US population overall in 2020 could range from 0.27 percent in the low-risk scenario to 1.22 percent in the high-risk scenario.
  • Some states face a greater risk of undercounts because they have large populations of historically undercounted groups. California has the greatest undercount risk, with projected 2020 undercounts ranging from 0.95 percent (low risk) to 1.98 percent (high risk). Other states at risk for serious undercounts are Texas, New Mexico, Nevada, Georgia, New York, and Florida.
  • The miscounts may disproportionately affect some groups more than others. Black and Hispanic/Latinx-identified individuals in the high-risk scenario could be undercounted nationally by 3.68 percent and 3.57 percent, respectively.
  • White, non-Hispanic/Latinx individuals could be overcounted nationally by 0.03 percent in the high-risk scenario. States with the greatest potential for overcounts include Vermont, West Virginia, Maine, New Hampshire, and Montana. These states have large populations of white, non-Hispanic/Latinx residents.
  • Children younger than 5, who have historically been undercounted, are at risk of being undercounted by as much as 6.31 percent in the high-risk scenario.

Potential Risks for the 2020 Census

An undercount in the 2020 Census is likely inevitable. The big question is by how much. Demographic changes over the last decade suggest that the population in 2020 will be harder to enumerate. Hard-to-count groups—including complex households, renters, young children, immigrants, and people of color—will represent a larger share of the population in 2020 than they did in 2010. Our projections show that even under the lowest-risk scenario—where we assume that the 2020 Census will perform exactly as the 2010 Census did—the national population count will be less accurate.

To counter the challenge of enumerating a nation that is harder to count, the 2020 Census will introduce operational changes like internet self-response to boost responses and the use of administrative records to fill in information for the missing population. Not only are these new additions insufficiently tested in a decennial census environment, but the best evidence suggests they will disproportionately improve the count of those who are already easiest to count, leaving the hard-to-count population a lingering challenge. In fact, our medium-risk scenario—where we assume the 2020 Census will perform exactly as anticipated—projects a less accurate count than that of 2010 precisely because of these changes.

Finally, political discourse about immigration and the citizenship question has created a potential chill among some groups in the country, including those who are Hispanic/Latinx-identified and immigrants. When we consider how the count could be affected by a chill to these groups’ participation in our high-risk scenario, in addition to lower participation among all demographic groups, we project that the count will be even more inaccurate.

Implications of an Unfair Census

The census needs to be not only accurate, but also fair. Fairness means that the census count truly reflects the diversity of the nation’s population, rather than undercounting some demographic groups while overcounting others. When entire communities are underrepresented in their count, they do not receive their rightful political voice or fair share of funding.

There is still time to invest in outreach and engage communities in a culturally sensitive way to ensure that representation in the 2020 Census is fair and accurate. In a democratic society, a fair and accurate count is critical and relies upon all of us to participate and complete the census in 2020.


See our interactive feature for the full data and to explore how these miscount risk factors could affect states and demographic groups in the 2020 Census.


This report was corrected on June 11, 2019, to delete “(log scale)” from the y-axis label in figure 1. The scale is linear, not logarithmic, so the numbered tick labels along the y-axis are correct as is.

Cross-Center Initiative

Cross-Center Initiative: 
To reuse content from Urban Institute, visit, search for the publications, choose from a list of licenses, and complete the transaction.