Census data are foundational to democracy, research and equitable urban policy. In addition to supporting political reapportionment and redistricting, census data serve as the backbone of the federal statistical data system and are often considered the highest quality data—the ‘gold standard’—for scholarly and policy research. In the United States, undercounts, new data protection mechanisms, and other challenges are eroding data quality and threatening census data as a trusted public good. First, I review the accuracy of the 2020 Census. The data show that the 2020 Census, while accurate, excluded millions of Americans; undercounts were severe for people of color, immigrants, and other marginalized communities. Second, I explain the US Census Bureau’s decision to protect individual privacy by intentionally distorting census data with differential privacy, leading to the ‘digital displacement’ of certain populations. Together, the ‘double whammy’ of worsening coverage error and differential privacy disproportionately affects communities of color and other marginalized groups, underscoring issues of data justice. Navigating this uneven and increasingly precarious digital landscape, I argue, requires practicing statistical citizenship, a theoretically informed approach that properly situates data limitations and simultaneously recognizes the politics of census data to support data justice.