The shifting deadlines for the 2020 census shortchanges data reconciliation processes that ensure an accurate count, says Sara Curran, director of the Center for Studies in Demography & Ecology at the University of Washington and a professor of international studies, sociology, and of public policy,
These data reconciliation processes were statutorily set for four months after the close of enumeration, but now are being given only two and a half months. On Nov. 19, census officials said data anomalies were likely to jeopardize existing deadlines.
And on Nov. 30, the Supreme Court will take up yet another census issue raised by the Trump administration: whether it can exclude undocumented immigrants from the count it uses to apportion seats in the U.S. House of Representatives.
Curran answered the following questions about the census.
Does the outcome of the recent presidential election have any effect on the census?
It’s not so much the presidential election, but the results of the runoff elections in January that will determine control of the Senate. As with most instances over the last four years, my answer is going to get into some uncharted territory.
Normally, according to Title 2 of the U.S. Code, within one week of the opening of the next session of the Congress after Dec. 31, the president must report to the clerk of the House of Representatives the apportionment population counts for each state and the number of representatives to which each state is entitled. If the Nov. 30 hearing before the Supreme Court yields a decision that supports the administration, and allows for excluding undocumented immigrants in the enumeration, then I expect there will be a major fight in the U.S. Congress. States where there may be a significant loss in population by eliminating undocumented immigrants, such as California, Florida and Texas, would be on the hook to either lose seats or not gain more seats in the House of Representatives. This loss would also affect over $600 billion in federal funds allocated annually to states according to their apportionment.
What is the logistical – impact of a shorter timeline in getting census results to the president?
The earlier deadline for the census count, originally set for July, allowed for the continued and challenging efforts to count hard-to-reach populations and to conduct the substantial, systematic and thorough data reconciliation processing to ensure that the census does not yield under- or over- counts of any groups of persons. These processes took four months during the 2010 census and were expected to take longer this year, because of the new online and phone-based data collection efforts, the variability of follow-up rates to census nonresponses and the COVID-19 pandemic.
The Commerce Department claims that 99% of households have been counted. Why is that in question?
That count includes both the count of people or households who self-responded and a count of “nonresponse” people or households whom enumerators reported following up on. For the latter, Secretary of Commerce Wilbur Ross claimed that census enumerators used efficient ways to impute, or estimate, the counts for those persons and households. This imputation approach drew upon other sources of data for a nonresponse address, but it is not yet clear how data reconciliation processes will hash out the variability in imputation rates and the quality of those imputations. This year, Washington state had a pretty high self-response rate at 72.4%. For Puerto Rico more than 64% of the count is imputed, whereas in Minnesota, data is imputed on fewer than 25% of households.
This seems to be a particularly controversial census. But have there been other significant disputes over previous censuses?
The 1920 census was controversial because the country had just experienced dramatic increases in its immigrant population, especially immigrants to the country’s cities. While the actual numbers were validated, they were not used for apportionment purposes during 1920-1930 because of substantial and successful legislative resistance from congressional representatives in rural districts.
During the Second World War, census information was used to relocate Japanese Americans for internment, which also violated the promised confidentiality of the census. Following the war, Congress added statutory language to ensure complete confidentiality of census data. Since then, there has never been a data breach. Other debates in the census have included those around how we count race and ethnicity, which has been politically charged, and with each decade, yielded different ways of categorizing our population.
Additionally, the amount of information collected on the census has been contentious: By 2000, both to cut costs and increase response rates, the long-form version of the census was eliminated, and only a short-form census is used now. The American Community Survey, an annual, nationally represented sample of households, replaced the long form.
For those who worry about the accuracy of the final 2020 census, is there any opportunity to revisit the outcome, or do we wait for 2030?
I’m sure that there will be considerable efforts to advocate for, and support, an accurate final product. Census products are absolutely vital for the conduct of life in the U.S., from private to public to civil society sectors. When it comes to the allocation of federal dollars or the apportionment counts, deciding on which number is the accurate one is probably going to be both adjudicated in the courts and through legislative action. I am often an optimist, and as painful as these political debates and processes might be, I hope this will be a teachable moment and an opportunity for everyone to understand better why we need well-funded, scientifically informed, expert-led federal and state agencies to provide accurate and comprehensive data about the country.