Census experts fear rush to finish tally will yield flawed data

doi:10.1126/science.369.6509.1285

GSTDTAP > 气候变化

DOI	10.1126/science.369.6509.1285
	Census experts fear rush to finish tally will yield flawed data
	Jeffrey Mervis
	2020-09-11
发表期刊	Science
出版年	2020
英文摘要	With the 2020 census in its final month, the U.S. statistical community fears rushed deadlines and political interference could lead to a seriously flawed head count. They want Congress to take two steps to avoid that fate: ensure that the Census Bureau has enough time to do the job right, and create an independent oversight body to track the agency's efforts. The primary purpose of the decennial census is to determine how many seats each state gets in the 435-member House of Representatives. The data are also used to allocate some $1.5 trillion per year in federal spending, and they fuel countless research studies of U.S. demographic trends. But many social scientists believe several recent actions by the Trump administration have undermined the bureau's ability to meet those obligations without sacrificing its rigorous standards for quality. Last month, the administration cut by nearly half the time the bureau had earlier said it needed for its final push to complete the census. Demographers fear that could result in a major undercount of people who are traditionally hard to reach—including immigrants, the poor, and people of color—and distort the country's demographic profile. And some observers charge that the recent insertion of three political appointees into new, high-level Census positions is part of a broader effort by the White House to produce a 2020 census that will benefit Republican-leaning states by giving them greater representation in Congress. “Forcing the bureau to meet the current deadlines will sacrifice the accuracy of the census, and waste $16 billion in taxpayer dollars,” says Arturo Vargas, head of the National Association of Latino Elected and Appointed Officials (NALEO) Educational Fund, one of several groups that have criticized the Trump administration's approach to the 2020 census. He says the administration's actions, which include a failed last-minute attempt to add a citizenship question to the census, have also tarnished the agency's “well-earned global reputation as a respected statistical agency, independent of political agendas.” The most expensive element of every census is tracking down the one-third of all U.S. residents who do not respond to repeated reminders to answer the 10 questions and submit the form. The bureau begins its nonresponse follow-up (NRFU) campaign roughly 6 weeks after the official 1 April start of the decennial census. But the COVID-19 pandemic delayed the NRFU and also led the bureau to ask Congress for a 4-month extension of its 31 December deadline for submitting the state-by-state numbers used for the apportionment of House seats. The Trump administration later rescinded that request, however, and on 3 August Census Bureau Director Steven Dillingham announced that the agency would meet its end-of-the-year deadline by halting field operations on 30 September, 4 weeks earlier than planned. (Last week a federal judge blocked the bureau's effort to wrap things up early pending a 17 September hearing.) Last week, the House committee that oversees the Census Bureau released an internal agency report warning that the compressed period “creates risks for serious errors” and that eliminating some operations “will reduce accuracy.” Census officials have also canceled an exercise this month designed to ensure enumerators don't miss so-called group quarters—places that are home to large numbers of residents, including college dormitories, prisons, and nursing homes. Such last-minute changes will most likely mean greater reliance on a process called imputation to fill in any data gaps. Imputation uses information on file with other government agencies to infer the demographic characteristics of nonrespondents. But experts say demographic groups with lower self-response rates are also less likely to be found in existing administrative records, increasing the odds they will be undercounted. In recent censuses, the nonresponse rate has been less than 1%—it was about 0.4% in 2010—leaving few holes to fill with imputation. But many experts believe the nonresponse rate could reach double digits in 2020. “And you can't impute 15%” without seriously jeopardizing the accuracy of the overall count, warns Kenneth Prewitt, a former Census director. To reduce that number, Prewitt and other census advocates want to give the bureau the 4-month extension it originally requested. In May, the Democrat-controlled House included the extension in a pandemic relief package. But that bill has stalled in the Republican-led Senate. Prewitt and three other former Census directors also want Congress to establish an independent body to monitor the census in real time. On 29 July, Prewitt told the House panel that the group would “assess if the final 2020 numbers reasonably match what the bureau knows they should be and, if not, what steps the country should take.” The panel's chair, Representative Carolyn Maloney (D–NY), supports the idea but hasn't settled on a legislative strategy to make it happen, according to a committee aide. And advocates worry the administration might balk at providing the outside body with the access it will need to do its job. Last month, the bureau took a small step toward greater transparency by posting daily updates on the percentage of addresses in each state from which it has collected information. But such state-level data don't tell the whole story. Advocates for an outside panel say it could dig deeper by examining response rates and NRFU data from smaller geographic areas, such as several city blocks or a portion of a rural community. Such analyses might result in the panel issuing “a red flag for the possibility of a disproportionate undercount,” says Robert Groves, another former Census director who supports outside oversight. The expert body could also examine how often field workers have used proxies—interviewing a neighbor, for example—to obtain information. Such proxies are less reliable than self-responses in determining whether a particular dwelling is occupied, as well as the race, sex, and age of every resident. Other helpful indicators of census quality might include the share of returned forms that didn't answer one or more questions. This summer's arrival of three political appointees holding newly created positions at the bureau has also spurred calls for more oversight. Social scientists fear that the appointees—including Benjamin Overholt, an Army veteran with a 2013 Ph.D. in applied statistics and research methods as the deputy director for data—might bring a political agenda to how the bureau completes its work and releases the data. Census officials declined to make Overholt available for an interview, and the bureau has not spelled out his duties. But Thomas Louis, a former chief scientist at the agency and emeritus professor at Johns Hopkins University, says that “Of all the things the Census Bureau doesn't need at this point, a deputy director for data is at the top of my list. … Ensuring data quality is the job of everybody involved in the collection, curation, and dissemination of census products.” Social scientists also worry that a 21 July Trump order requiring the Census Bureau to exclude undocumented residents from the state-by-state count will damage the overall quality of the 2020 census ( Science , 7 August, p. [611][1]). Civil rights groups have sued to block the order, which they say violates a constitutional requirement to count every resident. Given all these unanswered questions, some observers are already speculating about a possible early do-over. A badly flawed census, says one census expert who requested anonymity, could create “a groundswell for a middecade census.” [1]: http://www.sciencemag.org/content/369/6504/611
领域	气候变化 ; 资源环境
URL	查看原文
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.173/C666/handle/2XK7JSWQ/294076
专题	气候变化资源环境科学
推荐引用方式 GB/T 7714	Jeffrey Mervis. Census experts fear rush to finish tally will yield flawed data[J]. Science,2020.
APA	Jeffrey Mervis.(2020).Census experts fear rush to finish tally will yield flawed data.Science.
MLA	Jeffrey Mervis."Census experts fear rush to finish tally will yield flawed data".Science (2020).