Copied to clipboard
About this project
2020 Census
Who’s At Risk of Being Miscounted?
June 4, 2019

The decennial census, which aims to count every US resident each decade, is critical to our democracy. It affects congressional seats and funding decisions at every level of government.

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.

To understand how these factors could affect the 2020 Census counts, we created projections under three scenarios—reflecting the miscount risk as low, medium, or high.

Choose a state or demographic group to see who’s at risk of being miscounted.

Filter by
1Filter by
2Choose a state
2Choose a demographic
3Sort by
Potential miscounts
Low risk
Demographic changes alone could result in greater miscounts than in the 2010 Census.
Read more
This scenario assumes the 2020 Census’s performance will be similar to that of the 2010 Census in terms of over- and undercount rates among subpopulations. It factors in changes in US population age patterns, race and ethnic distributions, and the proportion of renters between 2010 and 2020. Even in this idealistic scenario, demographic changes would lead to a greater overall miscount than in 2010 because hard-to-count groups have become a larger share of the overall US population.
Medium risk
Changes to the census process, in addition to demographic changes, could result in greater miscounts than in the low-risk scenario.
Read more
This scenario assumes, in addition to demographic changes, that process changes introduced in 2020—like the online self-response option—will perform as the Census Bureau expects. But even if self-response rates align with the Bureau’s projections, the 2020 Census’s new approach to following up with people who do not respond (including greater reliance on administrative records to fill data gaps and fewer in-person follow-ups) would result in greater miscounts.
High risk
Lower-than-expected self-response rates, in addition to census process changes and demographic changes, could result in even greater miscounts than in the medium-risk scenario.
Read more
This scenario assumes the 2020 Census self-response rates will perform at the lowest level that the Census Bureau expects among all demographic groups. It also assumes that participation among Hispanic/Latinx residents and immigrants will be lower because of political discourse around immigration and the citizenship question. These lower self-response rates, combined with demographic changes and census process changes (like fewer in-person follow-ups), would result in greater miscounts than in the low- and medium-risk scenarios.
About the data

Our miscount assessments are quantitative but not meant to be used for statistical inference. There are no margins of error or significance levels available. Our goal is to provide some sense of the 2020 Census’s potential outcomes under a few scenarios that we subjectively deemed realistic.

Population projections have been rounded to the nearest hundred. Population projections for the different racial and ethnic groups do not add up to the overall population projection because the Census Bureau’s projection data do not separate the Hispanic/Latinx designation from any race except for white. The term “Hispanic/Latinx” is used throughout this feature to reflect the different ways in which people self-identify.

Our miscount projections are based on findings about the 2010 Census performance from the Census Coverage Measurement Program’s Post Enumeration Survey. For our projections of children younger than 5, we also use demographic analysis.

In the low-risk scenario, we assume the 2020 Census will perform as the 2010 Census did. The 2010 Census was considered one of the most accurate in recent history, but it still miscounted subpopulations. We assume US age patterns, race and ethnic distribution, and the proportion of renters will all have changed by 2020, but the miscount rates for each group will be the same as they were in the 2010 Census.

In the medium-risk scenario, we assume that, beyond demographic changes, the Census Bureau will achieve the initial self-response rate (including via the internet) it expects and that all operations will happen as the Bureau projects in its 2020 Operational Plan. This scenario also assumes that nonresponse follow-up cases will follow the Bureau’s process changes for the 2020 Census, which include reducing the number of nonresponse follow-up contact attempts from six to one and relying more heavily on administrative records to predict housing status and fill data gaps.

In the high-risk scenario, we use the baseline assumptions about 2020 Census operations from the medium-risk scenario (demographic and process changes) but assume the lowest self-response rate for which the Census Bureau is planning. In this scenario, we also assume lower response rates among the Hispanic/Latinx and immigrant populations because of the political climate regarding immigration and the potential addition of the citizenship question. This means that a greater share of households would be handled through the Bureau’s new nonresponse follow-up processes, which would lead to greater miscounts overall.

See more about our methodology in the technical appendix of the full report here.

Download data
Read the report

This feature was funded by the Bill and Melinda Gates Foundation. We are grateful to them and to all our funders, who make it possible for Urban to advance its mission. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Funders do not determine research findings or the insights and recommendations of our experts. More information on our funding principles is available here. Read our terms of service here.

Steven Martin led the research on demographic projections and miscount estimates for this project.

Allison Feldman
Ben Chartoff
Liza Hagerman
Emily Peiffer