Population censuses have never been and never will be perfect. But the challenges and obstacles to conducting the 2020 Census were especially numerous and varied—from its politicization to the pandemic—and accuracy and fairness were likely affected.
The challenging environment in which the 2020 Census was conducted suggests the need for additional benchmarks to understand its performance. The Urban Institute undertook this study to address questions about the 2020 Census’s quality and provide additional data about its accuracy and fairness. Urban created an innovative methodology—a simulation of the 2020 Census—to better understand the decennial census’s performance.
The undercount was not as severe as expected, but who was undercounted and overcounted varied.
These are key takeaways from our 2020 Census simulation:
- There likely was an overall 0.5 percent net undercount of the US population. Although it was different from the 2010 count, which had nearly perfect net accuracy, it was perhaps not as severe an undercount as some feared.
Considerable variation exists in who was undercounted and overcounted overall in the 2020 Census. Net accuracy is important, but fairness also matters.
- We find that the true total populations of Mississippi and Texas were undercounted in our simulated 2020 Census by 1.3 and 1.28 percent, respectively, while Minnesota’s population was net overcounted by 0.76 percent. For the next decade, such differences matter for these states. Mississippi and Texas residents will receive less of their fair share of federal funding for infrastructure, health care, and children’s programs. In contrast, Minnesota residents will receive more.
- If the residents had been counted accurately in the 2020 Census, Texas would receive over $247 million more and Minnesota would receive $156 million less in 2021 federal Medicaid reimbursements. A fair and accurate census impacts people’s well-being, and these outcomes can be disparate across the nation.
Those hardest to count in recent decennial censuses were again likely undercounted in the 2020 Census. For each hardest-to-count group, equity issues arise with the count’s fairness, how resources will be distributed, and who will miss out on their fair share of political representation and funding:
- Black and Hispanic/Latinx people had a net undercount of more than 2.45 and 2.17 percent, respectively, in our simulated 2020 Census.
- Young children, or those younger than age 5, were likely net undercounted by 4.86 percent.
- Nationwide, renters were likely undercounted by 2.13 percent overall.
- Households with a noncitizen present were likely undercounted by 3.36 percent overall.
Understanding potential miscounts and their outcomes is the first step in improving the Census.
By simulating the 2020 Census and a hypothetical full count, we can better understand potential miscounts across states and metropolitan areas and for demographic groups. This is especially important for a decennial census conducted in such a challenging environment. These findings also help illuminate how potential outcomes, such as apportionment and the allocation of federal Medicaid funding, are impacted by differential counts.
Although it is impossible to change the 2020 Census outcomes, with adequate planning and innovation, the 2030 Census can be improved for the hardest-to-count groups and places. Such actions may include
- implementing well-researched operational changes;
- encouraging states and cities to elevate the importance of Census Bureau data-collection activities; and
- ensuring consistent and strong funding in early years of the decade when testing and planning for innovations occurs.
These efforts are all possible for the 2030 Census but require us to collectively recognize how critical it is to invest in the decennial census and value it as a core component of our democracy.