Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 1 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO LATHAM & WATKINS LLP Steven M. Bauer (Bar No. 135067) steven.bauer@lw.com Sadik Huseny (Bar No. 224659) sadik.huseny@lw.com Shannon D. Lankenau (Bar. No. 294263) shannon.lankenau@lw.com 505 Montgomery Street, Suite 1900 San Francisco, California 94111 Telephone: 415.391.0600 Facsimile: 415.395.8095 LATHAM & WATKINS LLP Richard P. Bress (admitted pro hac vice) rick.bress@lw.com Melissa Arbus Sherry (admitted pro hac vice) melissa.sherry@lw.com Anne W. Robinson (admitted pro hac vice) anne.robinson@lw.com Tyce R. Walters (admitted pro hac vice) tyce.walters@lw.com Genevieve P. Hoffman (admitted pro hac vice) genevieve.hoffman@lw.com Gemma Donofrio (admitted pro hac vice) gemma.donofrio@lw.com 555 Eleventh Street NW, Suite 1000 Washington, D.C. 20004 Telephone: 202.637.2200 Facsimile: 202.637.2201 LAWYERS’ COMMITTEE FOR CIVIL RIGHTS UNDER LAW Kristen Clarke (pro hac vice forthcoming) kclarke@lawyerscommittee.org Jon M. Greenbaum (Bar No. 166733) jgreenbaum@lawyerscommittee.org Ezra D. Rosenberg (pro hac vice forthcoming) erosenberg@lawyerscommittee.org Dorian L. Spence (pro hac vice forthcoming) dspence@lawyerscommittee.org Maryum Jordan (pro hac vice forthcoming) mjordan@lawyerscommittee.org Ajay Saini (pro hac vice forthcoming) asaini@lawyerscommittee.org Pooja Chaudhuri (Bar No. 314847) pchaudhuri@lawyerscommittee.org 1500 K Street NW, Suite 900 Washington, DC 20005 Telephone: 202.662.8600 Facsimile: 202.783.0857 Attorneys for Plaintiffs National Urban League; City of San Jose, California; Harris County, Texas; League of Women Voters; King County, Washington; Black Alliance for Just Immigration; Rodney Ellis; and Adrian Garcia [additional counsel on docket] UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION NATIONAL URBAN LEAGUE; LEAGUE OF WOMEN VOTERS; BLACK ALLIANCE FOR JUST IMMIGRATION; HARRIS COUNTY, TEXAS; KING COUNTY, WASHINGTON; CITY OF LOS ANGELES, CALIFORNIA; CITY OF SALINAS, CALIFORNIA; CITY OF SAN JOSE, CALIFORNIA; RODNEY ELLIS; and ADRIAN GARCIA, Plaintiffs, v. CASE NO. 20-cv-5799-LHK DECLARATION OF JOHN THOMPSON IN SUPPORT OF PLAINTIFFS’ MOTION FOR STAY AND PRELIMINARY INJUNCTION WILBUR L. ROSS, JR., in his official capacity as Secretary of Commerce; U.S. DEPARTMENT OF COMMERCE; STEVEN DILLINGHAM, in his official capacity as Director of the U.S. Census Bureau; and U.S. CENSUS BUREAU, Defendants. CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 2 of 23 1 2 EXPERT DECLARATION OF JOHN THOMPSON I. 3 Introduction 1. I served as the Director of the U.S. Census Bureau from August 2013 to June 4 2017. My responsibilities as Director included overseeing the research and testing that produced 5 the design for the 2020 Census. Prior to becoming Director, I worked at the Census Bureau for 6 27 years, culminating with my role as the career senior executive in charge of management of all 7 aspects of the 2000 Decennial Census. These experiences and more inform my assessment that 8 the Trump administration’s decision to accelerate the timelines for completing the 2020 Census 9 will likely result in significant and material degradation of the quality of the 2020 Census 10 relative to previous censuses. 11 2. On April 13, 2020 the Secretary of Commerce, Wilbur Ross, and the Director of 12 the Census Bureau, Dr. Steven Dillingham, issued a statement on 2020 Census operational 13 adjustments due to the COVID-19 pandemic. This statement concluded that “[u]nder this plan, 14 the Census Bureau would extend the window for field data collection and self-response to 15 October 31, 2020, which will allow for apportionment counts to be delivered to the President by 16 April 30, 2021, and redistricting data to be delivered to the states no later than July 31, 2021.”1 17 3. On August 3, 2020 the Director of the Census Bureau, Dr. Steven Dillingham, 18 issued a statement announcing that the Census Bureau would “accelerate the completion of data 19 collection and apportionment counts by our statutory deadline of December 31, 2020, as required 20 by law and directed by the Secretary of Commerce.”2 Based on my experience and expertise, I 21 believe that this August 3 decision will adversely affect the quality and accuracy of the 2020 22 Census. 23 4. The 2020 Census results will be of great importance to our nation. The 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 1 U.S. Department of Commerce Secretary Wilbur Ross and U.S. Census Bureau Director Steven Dillingham Statement on 2020 Census Operational Adjustments Due to COVID-19, April 13, 2020, https://2020census.gov/en/news-events/press-releases/statement-covid-192020.html?linkId=100000011751624. 2 Statement from U.S. Census Bureau Director Steven Dillingham: Delivering a Complete and Accurate 2020 Census Count, August 3, 2017, https://2020census.gov/en/news-events/pressreleases/delivering-complete-accurate-count.html. 1 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 3 of 23 1 Constitution requires that the census be used for reapportioning the House of Representatives 2 and the Electoral College. The 2020 Census will also be used for numerous other functions to 3 support good policymaking and economic growth including: redrawing congressional and state 4 legislative voting districts; allocating over $1.5 trillion of federal funds annually; informing 5 sound policy development; providing critical information for state, local and tribal government 6 planning; and supplying important data to large and small businesses to generate growth and job 7 creation. Inaccuracies or errors in the 2020 Census will have grave consequences on these uses 8 for the subsequent 10-year period. 9 5. I have carefully reviewed the 2020 Census Operational Plans as well as the 10 documentation that the Census Bureau has issued describing the actions it is taking in response 11 to the COVID-19 pandemic and its recently issued documentation regarding its plans to conclude 12 data collection by September 30, 2020. I have evaluated these documents by drawing on my 13 experience and expertise developed over 31 years working on four censuses under seven 14 different presidential administrations. Based on my experience and expertise, it is my opinion 15 that reducing the time for data collection at this late date will most likely have grave and material 16 consequences for the 2020 Census and public perceptions of its legitimacy, because: (1) the time 17 constraints will force the Census Bureau to modify the plans for their data collection operations; 18 (2) these modifications will significantly increase the likelihood of larger total and differential 19 undercounts for the hard-to-count populations, as well as increase the levels of erroneous 20 enumerations and reduce the overall quality of this census, relative to previous censuses; and (3) 21 the Census Bureau is not providing timely measures that will allow stakeholders to assess 22 whether the 2020 Census is succeeding in carrying out a fair and accurate enumeration, 23 undermining the legitimacy of the count. 24 II. 25 Qualifications and Retainer Information 6. Below I briefly describe specific aspects of my qualifications and work 26 experience that establish my credentials as an accomplished statistician and an expert on the 27 Census Bureau and Decennial Census. I have also attached a copy of my CV to this declaration. 28 ATTORNEYS AT LAW SAN FRANCISCO 7. I have been retained to evaluate the likely impact of the administration’s decision 2 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 4 of 23 1 to compress data-collection and data-processing operations of the 2020 Decennial Census. My 2 compensation in this case is $150 per hour. 3 8. I have served both as the Director of the U.S. Census Bureau and as the career 4 senior executive in charge of management of all aspects of the 2000 Decennial Census. I am 5 also a distinguished professional in the areas of statistics and survey design. I have a deep 6 understanding of the processes that are necessary to achieve a complete and highly accurate 7 Decennial Census. 8 9 9. I started my career as a mathematical statistician in 1975. I spent the majority of my employment at the Census Bureau focused on the Decennial Census and ultimately served as 10 the Associate Director for the 2000 Decennial Census, with management responsibility for all 11 phases of the 2000 Decennial Census. As I mentioned above, I served as the Director of the U.S. 12 Census Bureau from August 2013 to June 2017 and worked at the Census Bureau for 27 years. 13 10. The Census Bureau is the country’s largest Statistical Agency and produces a 14 wide range of demographic and economic statistics including: the Decennial Census; the 15 American Community Survey; the Current Population Survey; the National Crime Victimization 16 Survey; the National Health Interview Survey; the Economic Census; 13 principal key economic 17 indicators released on a monthly or quarterly basis; and about 100 additional surveys. The 18 Director of the Census Bureau is appointed by the President and confirmed by the Senate. 19 11. Prior to being appointed Director of the Census Bureau, I was at National Opinion 20 Research Center (NORC) at the University of Chicago, serving as Executive Vice President from 21 2002 to 2008 and President from 2008 to 2013. NORC is an objective, non-partisan independent 22 research institution that delivers reliable data and rigorous analysis to guide critical 23 programmatic, business, and policy decisions. Clients include government, corporate, and 24 nonprofit organizations around the world who partner with NORC to transform increasingly 25 complex information into useful knowledge. NORC conducts research in five main areas: 26 Economics, Markets, and the Workforce; Education, Training, and Learning; Global 27 Development; Health and Well-Being; and Society, Media, and Public Affairs. NORC services 28 include designing and conducting surveys (telephone, internet, and in-person), as well as ATTORNEYS AT LAW SAN FRANCISCO 3 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 5 of 23 1 analytical studies. 2 12. From July 2017 to August 2018, I served as the Executive Director of the Council 3 of Professional Associations on Federal Statistics (COPAFS). COPAFS is an organization with 4 a membership consisting of professional associations and research organizations that depend on 5 and support high quality federal statistics. The Executive Director of COPAFS must have a deep 6 understanding of the Federal Statistical System and the wide range of data products that are 7 produced. Serving as the Executive Director of COPAFS reinforced my appreciation of the 8 importance of high-quality Decennial Census data to the entire Federal Statistical System. 9 13. In addition to the work experience described above, I am an elected Fellow of the 10 American Statistical Association and was selected to serve on the National Academies of 11 Science, Engineering, and Medicine Committee on National Statistics. 12 III. 13 Analysis A. The requirement to end data collection by the end of September 2020 will 14 force the Census Bureau to modify data collection procedures, resulting in a 15 less complete enumeration compared to previous censuses. 16 14. My responsibilities as Director of the Census Bureau included overseeing the 17 research and testing that produced the design for the 2020 Census. During my tenure, the 18 original operational plan for conducting the 2020 Census was released, as was an updated 19 version 2.0 of this plan.3 In addition, major field tests were conducted in 2013, 2014, 2015 and 20 2016. The results of these tests informed the final 2020 Census Design that was tested in the 21 2018 end-to-end test. This was the final large scale test in advance of the 2020 Census. It 22 combined the results of all previous tests and could be viewed as a dress rehearsal for the 2020 23 Census. Additionally, during the 2000 Census, I managed all aspects of census operations. 24 These experiences and the expertise that I developed in their course equip me to evaluate the 25 likely systemic effects of the August 3 decision to truncate the 2020 Census. 26 15. The COVID-19 pandemic forced the delay of key 2020 Census operations out of 27 28 ATTORNEYS AT LAW SAN FRANCISCO 3 U.S. Census Bureau, 2020 Census Operational Plan, A New Design for the 21st Century, version 2.0 issued, September 2016. 4 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 6 of 23 1 concerns for the safety of both census workers and the general public. The in-person 2 components of the local partnership program to increase response rates of the traditionally hard- 3 to-count populations were delayed, as was the operation to collect responses from those 4 households that do not self-respond. This operation is referred to as nonresponse follow-up or 5 NRFU. As of August 16, the national self-response rate is 63.8 percent, which means that over 6 36 percent—or over 50 million housing units and their occupants must still be enumerated.4 As I 7 will discuss below, the hard-to-count populations are disproportionately represented in the 8 nonresponse universe. A failure to obtain a complete enumeration in NRFU would result in 9 disproportionate undercounts of these populations. Therefore, I view a successful NRFU as the 10 most important census operation to ensuring a fair and accurate count. 11 16. The NRFU operation had been scheduled to start on May 15, 2020 and run 12 through July 31, 2020. However, as a result of the COVID-19 pandemic, the Census Bureau 13 rescheduled it to start in most of the United States on August 11, 2020 and initially planned to 14 complete it by October 31, 2020. 15 17. In order to accommodate this delay, the Census Bureau had requested, through the 16 Department of Commerce, a four-month extension of the deadlines5 to deliver apportionment 17 and redistricting data. For apportionment, the requested extension was from the current deadline 18 of December 31, 2020 to April 30, 2021. For redistricting, the requested extension was from 19 March 31, 2021 to July 31, 2021. 20 18. However, the Census Bureau has now announced that NRFU will be completed 21 by September 30, 2020.6 The Census Bureau will have to take steps to complete NRFU more 22 rapidly than it planned, given that it has already lost over a third of the schedule that the career 23 staff had developed under the original plan, all while managing the added difficulties that the 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 4 U.S. Census Bureau 2020 Census daily response rate tracker, https://2020census.gov/en/response-rates.html (last accessed August 16, 2020). 5 13 U.S.C. § 141(b), (c). 6 Statement from U.S. Census Bureau Director Steven Dillingham: Delivering a Complete and Accurate 2020 Census Count, August 3, 2020, https://www.census.gov/newsroom/pressreleases/2020/delivering-complete-accurate-count.html. 5 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 7 of 23 1 pandemic has created. 2 19. The Census Bureau recently released a review of the 2020 Census Operational 3 Plan Schedule7 that describes actions being taken to complete all data collection, including 4 NRFU, by September 30, 2020. According to the Plan, these actions include: 5 • Starting NRFU in all areas by August 9, 2020 6 • Sending enumerators to make up to 6 visits to attempt to obtain an interview with 7 occupied housing units 8 • Offering bonuses to NRFU enumerators to maximize staff production hours 9 • Making efforts at “Keeping Staff Levels Up” 10 • Implementing outbound telephone calling to supplement in-person contact attempts as a 11 means of enumerating hard-to-count populations 12 20. It is very unlikely that these actions will effectively address the constraints 13 imposed by the revised timelines for completing NRFU. My conclusion is informed by my 14 experiences in managing all aspects of the 2000 Census and by directing the research and 15 development necessary to plan the 2020 Census. The bases for my conclusion are as follows: • 16 The staffing levels will not be adequate to complete NRFU without accepting lower 17 quality enumerations and incompletely enumerating the traditionally hard-to-count 18 populations. The Census Bureau has lost over 30 percent of the time that had been 19 planned for NRFU, so it stands to reason that they will need more staff to complete this 20 critical undertaking. However, the plan being put forth to end data collection by 21 September 30 is to maintain staffing at levels determined before the advent of the 22 COVID-19 pandemic. The Department of Commerce Office of the Inspector General has 23 recently reviewed the progress of staffing for the NRFU and stated: 24 “Bureau management have stated that their target number of enumerators, needed 25 by the end of August 2020 to complete NRFU production, is just above 300,000. 26 As of August 17, 2020, the Bureau has just under 220,000 enumerators trained 27 28 ATTORNEYS AT LAW SAN FRANCISCO 7 U.S. Census Bureau, Review of 2020 Census Operational Plan Schedule, August 17, 2020, https://www.census.gov/content/dam/Census/newsroom/press-kits/2020/2020-operational-planschedule-review.pdf. 6 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 8 of 23 1 and ready to start working on the NRFU operation that is underway—this 2 represents approximately 73 percent of the estimated number of enumerators 3 needed to complete NRFU production. However, 132 out of 248 total Area 4 Census Office (ACOs) are less than 75 percent toward reaching their estimated 5 goals; of those 132 ACOs at less than 75 percent, 37 are less than 50 percent 6 toward reaching their goal.”8 7 Briefly, the Bureau has established Area Census Offices (ACO) to carry out the 2020 8 Census field operations, including NRFU. There are 248 ACOs, each of which has a 9 significant portion of the NRFU workload to carry out. On average, this would be 10 about 226,000 housing units from which a self-response was not received. But the 11 Census Bureau is already falling significantly behind in its plans for staffing NRFU, 12 and these hiring shortfalls for NRFU staff are not uniform. Approximately 15 percent 13 of the NRFU workload is in areas where the Census Bureau is falling 50 percent short 14 of hiring goals. While the Census Bureau stated in the Review of the Operational 15 Plan Schedule that it was making efforts at “keeping staff levels up,” it is falling well 16 behind in reaching the staffing levels it had determined were necessary for NRFU. 17 Insufficient staffing will significantly lower the quality and effectiveness of NRFU 18 operations, as I explain below. • 19 The self-response rates are not uniformly distributed and are disproportionally lower in 20 areas with higher proportions of Black and Hispanic populations, as well as in some rural 21 areas. As of August 6, 2020, there were 50.7 million people living in census tracts in the 22 23 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 8 Mark H Zabarsky, Principal Assistant Inspector General for Audit and Evaluation, 2020 Census Alert: The Census Bureau Faces Challenges in Accelerating Hiring and Minimizing Attrition Rates for Abbreviated 2020 Census Field Operations Final Memorandum No. OIG-20041-M., Memorandum for Steven D. Dillingham, Director, U.S. Census Bureau, August 18, 2020. 7 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 9 of 23 1 lowest fifth of self-response.9, 10 The overall self-response rate for these tracts is less than 2 51.3 percent, compared to a national average of over 63 percent. Furthermore, while 3 non-Hispanic Blacks make up 12.3 percent of the US population, they represent 22.2 4 percent of the population in these low response areas. For Hispanics, the corresponding 5 rates are 18.3 and 25.8 percent, respectively. The Census Bureau also noted that, as of 6 August 6, 2020, the self-response rate in update-leave (rural areas) was a little over 34 7 percent. In addition, since these areas have the lowest self-response rate, they will have 8 the largest NRFU workloads, making recruiting and hiring sufficient staff to achieve a 9 complete enumeration particularly challenging. As I will discuss below, the likely 10 outcome for these areas and populations will be increased undercounts relative to 11 previous censuses and decreased quality of the information collected. • 12 Given the current NRFU staffing levels, the Census Bureau will have to rely less on 13 direct in-person contact attempts and more on the following in order to try to meet the 14 new September 30, 2020 deadline, with deleterious consequences for the count: 15 a. Reduced in-person contact attempts with residents of the NRFU households, leading 16 to increased undercounts of the traditionally hard-to-count populations. While the 17 Census Bureau is planning for up to 6 attempts for most NRFU households, this will 18 not be enough to obtain complete interviews in many hard-to-count communities. 19 The Government Accountability Office (GAO) evaluated the early testing that the 20 Census Bureau carried out to develop the current NRFU procedures. The GAO 21 stated: 22 23 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 9 A census tract is a small geographic area that is similar to a neighborhood. See https://www.census.gov/programs-surveys/geography/about/glossary.html#par_textimage_13. 10 Steven Romalewski, Mapping “Self-Response” for a Fair and Accurate Census, Center for Urban Research at the Graduate Center, City University of New York, August 7, 2020, https://www.gc.cuny.edu/CUNY_GC/media/CUNY-GraduateCenter/PDF/Centers/Center%20for%20Urban%20Research/Resources/Census2020-selfresponse-rates-thru-Aug-6-CUNY-Graduate-Center.pdf. 8 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 10 of 23 1 “according to preliminary 2016 Census Test data, there were 19,721 NRFU cases 2 coded as non-interviews in Harris County, Texas and 14,026 in L.A. County, 3 California, or about 30 and 20 percent of the test workload respectively. 4 According to the Bureau, non-interviews are cases where no data or insufficient 5 data were collected, either because enumerators made six attempted visits without 6 success (the maximum number the Bureau allowed) or visits were not completed 7 due to, for example, language barriers or dangerous situations.”11 8 The Census Bureau subsequently refined the NRFU procedures to allow for more 9 contact attempts, as is necessary to reach higher resolution rates comparable to 10 previous censuses.12 11 In addition, hard-to-count communities have significantly lower levels of self- 12 response, and a corresponding larger proportion of households that fall into NRFU. It 13 will not only be more difficult to recruit adequate staff for these areas, but making 6 14 attempts will be exceedingly difficult, and as I noted above, will not be enough to 15 obtain complete responses from all households in these areas. My experience has 16 shown that the proposed use of outbound telephone calling will be ineffective in 17 reducing the need for in-person interviewing. For example, the Pew Research Center 18 has documented that telephone survey rates have fallen from 36 percent in 1997 to 19 under 6 percent in 2018.13 In addition, outbound telephone calling for NRFU has not 20 been tested to determine whether it is even effective. Achieving a complete and 21 accurate count in the hard-to-count communities requires a lot of hard work by well- 22 trained enumerators who are very familiar with these areas. Limited staff and a 23 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 11 United States Government Accountability Office, 2020 CENSUS Additional Actions Could Strengthen Field Data Collection Efforts, GAO-17-191, a report to congressional requesters, January 2017. 12 U.S. Census Bureau, 2020 Census Detailed Operational Plan for: 18. Nonresponse Followup Operation (NRFU), Version 2.0 Final, July 15, 2019. 13 Courtney Kennedy and Hannah Hartig, Response rates in telephone surveys have resumed their decline, Pew Research Center report, February 27, 2019. 9 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 11 of 23 1 shortened time frame will likely result in serious and material increases in the 2 undercounts for these communities relative to previous censuses. 3 b. Increased proxy enumerations, resulting in increased levels of erroneous 4 enumerations. The limited NRFU workforce combined with the shortened schedule 5 will result in a higher level of proxy enumerations than in previous censuses. Proxy 6 enumerations are those obtained by asking people other than the actual residents of 7 NRFU households for information about those residents. These proxies can include 8 neighbors, apartment managers, or other knowledgeable persons. The Census Bureau 9 conducted the 2010 Census Coverage Measurement (CCM) program which included 10 an extensive evaluation of the accuracy and quality of the 2010 Census. The CCM 11 found that in the 2010 Census, proxy enumerations were obtained for about 21 12 percent of the NRFU returns. The erroneous enumeration rate for the proxy 13 enumeration was 6.7 percent—over twice the overall erroneous enumeration rate of 14 3.3 percent.14 15 c. Increased reliance on administrative records to complete NRFU enumerations, 16 leading to less complete enumerations for the hard-to-count populations. The Census 17 Bureau plans include the use of administrative records (e.g., records from the IRS, 18 Medicare, and the Social Security Administration) to reduce the NRFU workload, 19 where feasible, by using such records to enumerate occupied households that have 20 failed to respond after several contact attempts.15 The Census Bureau may be forced 21 to rely more heavily on such enumerations if NRFU cannot be completed as planned. 22 Based on the research that the Census Bureau conducted to develop the current 23 NRFU strategy, it had planned to enumerate 12.9 percent of the occupied NRFU 24 housing units after making one visit.16 Expanding the uses of administrative records 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 14 P. Cantwell, DSSD 2010 Census Coverage Measurement Memorandum Series # 2010-G-01, (May 22, 2012), https://www.census.gov/coverage_measurement/pdfs/g01.pdf. 15 Albert E. Fontenot, Intended Administrative Data Use in the 2020 Census, 2020 Census Program Memorandum Series: 2020.06, May 7, 2020. 16 U.S. Census Bureau, 2020 Census Detailed Operational Plan for: 18. Nonresponse Followup Operation (NRFU), Version 2.0 Final, July 15, 2019. 10 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 12 of 23 1 to enumerate a higher portion of the NRFU occupied housing units is not supported 2 by the research the Census Bureau has used to date, and the Census Bureau has not 3 released additional research to support such actions. Census Bureau research has 4 shown that the quality and completeness of administrative records is not expansive 5 enough to replace a decennial census.17 Therefore, the use of administrative records 6 beyond the planned levels for NRFU will be less representative of the hard-to-count 7 populations than a complete NRFU. 8 d. Likely an increased use of “whole person imputation” relative to previous censuses. 9 Such imputations will not correct for any undercounts that have resulted from an 10 incomplete NRFU. In conducting NRFU in previous censuses, situations have arisen 11 where, despite the best efforts of NRFU enumerators, either minimal or no 12 information was obtained for some housing units by the conclusion of the NRFU. 13 The Census Bureau uses statistical techniques, referred to as imputation, to correct for 14 this missing data problem. The statistical processes are used to estimate—or 15 impute—all of the characteristics of the persons in these housing units. The Census 16 Bureau applies “Count Imputation” for situations where no information is available 17 for a housing unit. This methodology will first estimate whether the unit is occupied, 18 and if so, will estimate or impute a household size – meaning, the number of people 19 in that household. The process will then use “whole person imputation” to estimate 20 characteristics for persons in a household of this size. The Census Bureau also uses 21 whole person imputation in situations where only the count of people residing in a 22 housing unit could be determined. In the 2010 Census, about 2.0 percent of the 23 enumerations fell into the category of whole person imputation – 0.4 percent were the 24 result of count imputation and 1.6 percent resulted when only the population count 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 17 Rastogi, Sonya and Amy O’Hara, 2010 Census Match Study, 2010 Census Planning Memorandum Series, No. 247, November 19, 2012. 11 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 13 of 23 1 was known.18 It should be noted that of the 16.3 million persons enumerated by 2 proxy in the 2010 Census, 23.1 percent required whole person imputation.19 I believe 3 that the levels of housing units requiring whole person imputation will be much larger 4 in 2020 than in 2010, due to the reduction in time and staff limitations for NRFU 5 enumerators to get a complete response. Unfortunately, the statistical methods that 6 the Census Bureau uses for whole person imputation rely on using information from 7 the resolved housing units to estimate or impute for the unresolved housing units. 8 Therefore, any undercounts that are in the resolved housing units will be carried 9 forward and not corrected. 10 B. 11 The reduced schedule for NRFU will have serious accuracy and quality implications for the 2020 Census 12 21. Undercounts, particularly for traditionally hard-to-count populations, are likely to 13 increase significantly in 2020 relative to previous censuses as a result of the Bureau’s new, 14 reduced schedule. As I discussed above, the NRFU workloads will be relatively higher in areas 15 with lower self-response rates. The Census Bureau uses low self-response as a key measure in 16 determining whether an area is hard-to-enumerate,20 so by definition the challenge for NRFU to 17 obtain a complete count is in these areas. In addition, these areas also contain higher proportions 18 of Black and Hispanic populations relative to the White non-Hispanic population. The end result 19 for these communities is likely to be incomplete NRFU enumeration due to staffing and time 20 limitations, as well as more use of proxy enumerations and whole person imputation. This will 21 lead to increased undercounts relative to previous censuses. For example, in the 1990 Census the 22 undercount of Black or African American population was 4.6 percent and for the Hispanic 23 population the undercount was 5.0 percent.21 It is important to understand that in 1990, the 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 18 P. Cantwell, DSSD 2010 Census Coverage Measurement Memorandum Series # 2010-G-01, May 22, 2012, https://www.census.gov/coverage_measurement/pdfs/g01.pdf. 19 Ibid. 20 Response Area Outreach Mapper, Census.gov, www.census.gov/roam, July 2018. 21 P. Cantwell, DSSD 2010 Census Coverage Measurement Memorandum Series # 2010-G-01, May 22, 2012, https://www.census.gov/coverage_measurement/pdfs/g01.pdf. 12 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 14 of 23 1 Census Bureau had the flexibility to extend the NRFU beyond its planned end date until it had 2 reached a completion rate of 99 percent for NRFU enumeration.22 However, even with this high 3 completion rate for 1990, serious undercounts were measured. The Census Bureau does not have 4 the flexibility to extend NRFU for the 2020 Census – it has a hard stop at September 30, 2020. 5 In my opinion, there is a high risk that the measures the Census Bureau will be forced to take to 6 complete NRFU by this unmovable deadline (as I discussed above relying more on proxy or 7 count-only enumerations and administrative records), even potentially falling short of the 99 8 percent completion goal, will likely result in undercounts that will be materially larger than were 9 observed in the 1990 Census. 10 22. The overall quality of the 2020 Census data will very likely be materially lower 11 than in previous censuses. As I noted above, it is very likely that the Census Bureau will have to 12 rely more on proxy enumeration and whole person imputation that in previous censuses. While 13 this will be a particular problem for the hard-to-count areas, these less accurate enumeration 14 methods will also most likely be used more across the board in the 2020 Census relative to 15 previous censuses. In addition to the increased use of proxy enumeration, as I discussed above, 16 employing a higher level of administrative records and whole person imputation will result in 17 lower quality than would have been achieved through direct in-person contact. 18 23. The impacts of undercounts and poor quality data will not just be a problem for 19 the immediate uses of the 2020 Census (e.g., apportionment and redistricting), but will remain 20 for the 10 years until they can be corrected in the 2030 Census. 21 C. 22 Increased transparency is essential to assure stakeholders of the legitimacy of 2020 Census data collection 23 24. At this point, there is little information available to assess the conduct of the 2020 24 NRFU. The Census Bureau has been very forthcoming about the self-response portion of the 25 2020 Census: detailed and granular data have been made available to allow for public 26 assessment of self-response for many areas, including census tracts. This is not true for the 27 28 ATTORNEYS AT LAW SAN FRANCISCO 22 U.S. Census Bureau, 1990 Census of Population and Housing – History Field Enumeration 636, Report Number CPH-R-2, 1996, https://www.census.gov/library/publications/1996/dec/cphr-2.html. 13 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 15 of 23 1 NRFU portion of the 2020 Census. 2 25. The current Census Bureau plan is to release only NRFU resolution rates at the 3 State level. These rates are not helpful in assessing the actual progress of NRFU in achieving a 4 complete enumeration of all population groups and areas. In order to demonstrate that the NRFU 5 is meeting the goal of a complete and accurate enumeration, it is essential that the Census Bureau 6 provide additional data beyond just the resolution rate of housing units in NRFU. These data 7 should include information such as the rate of proxy and count-only enumerations at similar 8 levels of geographic aggregation as the self-response data. The absence of more granular data 9 will compromise public perception of the legitimacy of any final results that the Bureau does 10 release. 11 26. The public’s perception of the legitimacy of the census is already imperiled. The 12 Census Bureau has recently announced three new political appointees, including a new Deputy 13 Director for Policy23 and a new Deputy Director for Data.24 Having political appointees with 14 vague responsibilities at the Deputy Director level of the Census Bureau (which has always been 15 a career position) is unprecedented and is raising serious concerns among stakeholders. 16 Perceptions that the results of the 2020 Census have been manipulated for political purposes will 17 erode public and stakeholder confidence, not only in the 2020 Census, but also in our democratic 18 processes more generally. Therefore, it is critical that the Census Bureau release the data that I 19 have described above to demonstrate that it is achieving a complete and fair enumeration through 20 NRFU. 21 22 23 24 25 26 27 28 ATTORNEYS AT LAW SAN FRANCISCO 23 Statement from Census Bureau Director Steven Dillingham, Release Number CB20-RTQ.20. 24 Statement from Census Bureau on Deputy Director for Data, Release Number CB20-RTQ.24, August 17, 2020. 14 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISO PLTFS.’ MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 16 of 23 1 2 IV. Conclusion 27. It is my conclusion that the current deadlines for delivering the 2020 Census / 3 apportionment and redistricting data place unreasonable time constraints on the Census Bureau. 4 These constraints will not allow the Census Bureau to carry out data collection operations that 5 will deliver high quality results These timing constraints will very likely lead to large 6 undercounts for the 2020 Census, and much larger undercounts than measured in previous 7 censuses. 8 9 Pursuant to 28 U.S.C. § 1746, I declare under penalty of perjury that the foregoing is true and correct. 10 11 Executed on Augusf~, 2020 at Bend, Oregon. 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 LATHAM &WATKINS u, ATT ORNEYS AT LAW SAN FRA N CISCO 15 CASE NO. 5:20-CV-05799-LHK THOMPSON DECL. ISOPLTFS.' MOT. FOR STAY AND PRELIMINARY INJUNCTION Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 17 of 23 JOHN H. THOMPSON BRIEF CAREER HISTORY Extensive Senior Executive leadership in the non-profit and federal sectors, with experience in social science research and statistics, congressional advocacy, building coalitions, operational management, business development, stakeholder relations, innovation, and strategic vision. Independent Consultant, August 2018 to present Consulting service focusing on survey methodology, executive leadership, the Federal Statistical System, and decennial census. Activities have included: • • • Expert witness for the plaintiffs in two court cases opposing the addition of a citizenship question to the 2020 Census o New York Immigration Coalition, et al v. United States Department of Commerce and Wilbur Ross, U.S. District Court for the Southern District of New York, and o Robyn Kravitz et al., v. United States department of Commerce, et al Training news media journalists on the 2020 Census with Georgetown University, the Poynter Center, and the Harvard Shorenstein Center. Providing consultation services to NORC at the University of Chicago Executive Director, Council of Professional Associations on Federal Statistics – July 2017 to August 2018 The Council of Professional Associations on Federal Statistics (COPAFS) was founded in 1981 to coordinate activities of a number of Associations, Organizations, and Businesses that rely on federal statistics to support good governance and economic growth. COPAFS now represents a growing body of stakeholders that support the production and use of high quality statistics. The Executive Director represents these stakeholders in realizing their mission to Advance Excellence in Federal Statistics. Activities include: • • • • • • Advocated on behalf of federal agencies. For example, COPAFS is a co-chair of the Friends of the Bureau of Labor Statistics, and the Friends of the National Center for Health Statistics; Worked with stakeholder coalitions to support proper funding for the 2020 Census and the American Community Survey; Ensured members of Congress, COPAFS members, and other stakeholders were informed of critical issues facing agencies that produce federal statistics; Alerted members and stakeholders of breaking issues that needed immediate support and attention; Organized and supported ongoing educational efforts for members of Congress and their staff on the value and importance of federal statistics both nationally and in their own states and districts; Created and joined in powerful coalitions of organizations and businesses to advocate on behalf of federal agencies that produce statistics, building broad support across a wide spectrum of data users; Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 18 of 23 John H. Thompson Page 2 • • • Built partnerships with foundations that help fund critical research in the statistical agencies and academia to ensure the on-going modernization of how statistical data are created and made available to the public and researchers, and to fund educational efforts; Worked closely with the Chief Statistician of the United States and the statistical agencies to help inform and promote modernization efforts underway and assist agencies in keeping abreast of new stakeholder data needs; and Hosted events to demonstrate the importance of federal statistics such as the 2018 Federal Committee on Statistical Methodology Research and Policy Conference. Director, United States Census Bureau – August 2013 to June 2017 Appointed by the President as Director of the largest federal statistical agency, with a staff of over 5,000 headquarters employees and approximately 10,000 to 15,000 staff spread across the United States in six regional offices and a major production facility in Indiana, with an annual budget exceeding $1 billion. Key accomplishments include: • • • • • • • • Worked successfully with the executive and legislative branches of the federal government, including the White House, the Office of Management and Budget, Cabinet officials, and members of Congress and congressional staff, to accomplish a major transformation of the Census Bureau into a forward-looking 21st century statistical agency. Testified at 6 congressional hearings on the Census Bureau; Provided a conceptual vision and lead a redesign of the 2020 decennial census that is estimated to save $5 billion through effective use of operations research-driven reengineering of field operations, innovative use of technology, and partnership with key stakeholders; Lead outreach to key stakeholders including representatives of state local and tribal governments; advocacy organizations; professional associations, business groups, various media; and academic researchers; Put in place a robust research program to support mission critical activities, such as linking administrative records, disclosure avoidance methods, economic studies, statistical research, survey methodology, big data, and data dissemination; Lead efforts to maintain congressional support and funding for the American Community Survey, a critical data asset of the federal government, including mobilizing a diverse group of key stakeholders to effectively advocate in support of the survey, personally visiting almost all of the House of Representatives and Senate members of the Census Bureau appropriations and oversight committees, and establishing a program of research directly related to the concerns that had been raised; Improved economic statistics through research on using alternatives to direct survey data collection to produce statistics that are timelier and have increased granularity, and carrying out three initiatives to advance the release of principal economic indicators on trade, retail sales and services, which allowed the Bureau of Economic Analysis to significantly reduce revisions to Gross Domestic Product (GDP) estimates; Recruited outstanding research staff including new senior leadership for Research and Methodology, the Director of a newly established big data center, and seven former Presidential Innovation Fellows; and Improved data dissemination to the public, including development of a platform to deliver data in ways that will meet the rapidly evolving demands of a growing body of users. In addition, Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 19 of 23 John H. Thompson Page 3 in order to meet immediate targeted demands two new tools were released: City SDK (Software Development Kit) to allow easy developer access; and Census Business Builder a tool that combines small area demographic and economic data in a way that is easily accessible for entrepreneurs and small business owners. President and Executive Vice President, NORC at the University of Chicago – July 2002 to August 2013 NORC is a national non-profit organization that conducts high quality social science research in the public interest. As President, I had responsibility for all NORC corporate activities and for the quality of all NORC research efforts. I provided vision for NORC to establish the organization as a leader in the social science research industry. My accomplishments included: • • • • Strengthened the organization’s high-quality, diverse staff; Broadened the scope of the collaborations between NORC and the University of Chicago; Realized nearly 50 percent growth in revenue and greatly expanding NORC’s portfolio of business and research programs; and Provided leadership in the social science research community - selected to be a Fellow of the American Statistical Association (ASA), elected to serve a term as Chair of the Social Statistics Section of the ASA, and chaired the 2009 ASA Committee on Fellows. Also elected as a member of the Committee on National Statistics, serving on two National Academy of Sciences panels addressing 2010 and 2020 Census concerns. As Executive Vice President of Survey Operations (2002 – 2008), I provided oversight and direction to the Economics, Labor Force, and Demography Research Department, the Statistics and Methodology Department, and Survey Operations for field and telephone data collection. My major accomplishments included: • • • • Provided leadership and guidance for a major corporate initiative, the National Immunization Survey, which is conducted on behalf of the Centers for Disease Control and Prevention, and is the largest telephone survey in the United States conducted via random digit dialing for scientific purposes. Significantly increased the productivity and cost effectiveness of NORC’s overall data collection activities; Successfully utilized skills in directing large project start-ups, and in managing large complex operations, directing the project through the completion of the first contract phase, which included the first year of data collection and the delivery of the first data set; and All survey operations were completed on schedule, and within budget including the delivery of an extremely complex data set, and a public use file. Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 20 of 23 John H. Thompson Page 4 Principal Associate Director and Associate Director for Decennial Census Programs, United States Census Bureau – 1997 to July 2002 Served as the senior career executive responsible for all aspects of the 2000 Decennial Census. This was the largest peacetime mobilization undertaken by the U.S. government, with a budget of $6.5 billion, establishment of over 500 field offices, a temporary workforce that peaked at over 500,000, and establishment of telephone capacity to receive over 5 million calls over a period of one month. I was also chairman and director of the Executive Steering Committee for Accuracy & Coverage Evaluation Policy for the 2000 Census. This Committee was charged with making a recommendation as to whether or not to adjust the 2000 Census redistricting data for coverage errors, an issue fraught with political disagreement and controversy. This work was widely recognized as superb – with the Committee’s recommendation supported by numerous reviews, including the National Academy of Sciences Panel on evaluating Census 2000. EDUCATION M.S. Virginia Polytechnic Institute and State University, 1975 Mathematics Graduate course work in statistics - George Washington University 1977-1981 B.S. Virginia Polytechnic Institute and State University, 1973 Mathematics PROFESSIONAL SERVICE AND ASSOCIATIONS American Statistical Association, 1975 to Present Chair, Social Statistics Section – 2011 Chair, ASA Committee on Fellows - 2009 National Academy of Sciences, Member of the Committee on National Statistics – 2011 - 2013 Member of the Panel on the Design of the 2010 Census Program of Evaluations and Experiments Member of the Panel to Review the 2010 Census HONORS AND AWARDS Virginia Tech College of Science Hall of Distinction inaugural class, 2013 Presidential Rank Award of Meritorious Executive, 2001 Department of Commerce, Gold Medal, U.S. Bureau of the Census, 2000 Elected Fellow of the American Statistical Association, 2000 Department of Commerce, Silver Medal, U.S. Bureau of the Census, 1998 Department of Commerce, Bronze Medal, U.S. Bureau of the Census, 1988 Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 21 of 23 John H. Thompson Page 5 PAPERS AND PUBLICATIONS 2018 Thompson, John H and Yablon, Robert. Issue Brief: “Preparing for the 2020 Census Considerations for State Attorneys General”. American Constitution Society., October 10, 2018 2012 Thompson, John H. (Panel Member). “Panel Discussion: Considering Changing Sectors in the Research Industry?: Advice From Those Who Have Done It!” AAPOR 67th Annual Conference, Orlando, Florida, May 19, 2012 2012 Thompson, John H. (Discussant). “Future is Now: Realignment of Current Survey Management and Operations at the Census Bureau”. Population Association of America 2012 Annual Meeting, San Francisco, California, May 4, 2012. 2012 Thompson, John H. (Discussant). “Use of Administrative Records in the 2020 Census.” Federal Committee on Statistical Methodology, Washington, DC., January 10, 2012 2011 Weinberg, Daniel H. and Thompson, John H., “Organization and Administration of the 2010 U.S. Census.” In Margo J. Anderson, Constance F. Citro, and Joseph J. Salvo (eds.) Encyclopedia of the U.S. Census, Second Edition, CQ Press., July 2011 2010 Thompson, John H., “Challenges, Innovation and Quality for the 21st Century” Keynote Speech at the 2010 FCSM Statistical Policy Seminar, Washington, DC, December 14, 2010. 2010 Thompson, John H., “The Future of Survey Research: Opportunities and Challenges” Paper presented at the Applied Demography Conference, San Antonio, Texas., January 11, 2010 and at the Population Association of America 2010 Annual meeting, Dallas, Texas, April 15, 2010. 2008 Thompson, John H. (Panel Member). “Panel Discussion: The American Community Survey: Promise, Products and Perspectives.” Population Association of America Annual Meeting, New Orleans, Louisiana, April 17, 2008. 2006 Thompson, John H. (Discussant). “Census 2010: A New Census for the 21st Century.” Population Association of America Annual Meeting, Los Angeles, California, March 30, 2006. 2004 Thompson, John H., “Interviewer Falsification of Survey Data.” Paper presented at the Joint Meetings of the American Statistical Association, Toronto, Canada, August 11, 2004. 2003 Thompson, John H., “Is Interviewer Falsification Scientific Misconduct?” Roundtable paper presented at the American Association for Public Opinion Research 58th Annual Conference, Nashville, Tennessee, May 16, 2003. 2002 Thompson, John H. (Discussant). “Eliminating the 2010 Census Long Form? – Current Status of the American Community Survey.” Population Association of America Annual Meeting, Atlanta, Georgia, May 9, 2002. Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 22 of 23 John H. Thompson Page 6 2001 Thompson, John H., “Decision on Release of Statistically Corrected Redistricting Data.” Invited paper presented at the Joint Meetings of the American Statistical Association, Atlanta Georgia, August 6, 2001. 1999 Thompson, John H., “Census 2000 – Innovations and New Technology.” Paper presented at the Economic Commission for Europe’s Conference of European Statisticians Meeting, Geneva, Switzerland, February 15-17, 1999. 1998 Thompson, John H. and Robert E. Fay, “Census 2000: The Statistical Issues.” Paper presented at the Joint Meetings of the American Statistical Association, Dallas, Texas, August 9-13, 1998. 1996 Thompson, John H. and Karen Mills, “Census 2000 Content: Tradeoffs on Cost, Quality, and Quantity.” Paper presented at the Annual Meeting of the Population Association of America, New Orleans, Louisiana, May 9-11, 1996. 1995 Thompson, John H., Mary H. Mulry, Susan M. Miskura, “Census 2000: Statistical Issues in Reengineering the Decennial Census.” Paper presented at the Annual Meeting of the American Statistical Association, Orlando, Florida, August 13-17, 1995. 1992 Fay, Robert E. and John H. Thompson, “The 1990 Post-Enumeration Survey: Statistical Lessons in, Hindsight.” Paper presented at the Annual Research Conference, March 22-25, 1992, Arlington, Virginia. 1989 Edson, Robert G. and John H. Thompson, “1990 Decennial Census Coverage Improvement Program.” Paper presented at the Annual Winter Meetings of the American Statistical Association, San Diego, California, January, 1989. 1988 Navarro, Alfredo, John H. Thompson, and Linda Flores-Baez, “Results of Data Switching Simulation.” Paper presented to the Census Advisory Committees at the Joint Advisory Committee Meetings, Oxon Hill, Maryland, April, 1988. 1987 Griffin, Richard A. and John H. Thompson, “Confidentiality Techniques for the 1990 Census.” Paper presented to the Census Advisory Committees at the Joint Advisory Committee Meetings, Oxon Hill, Maryland, October, 1987. U.S. Bureau of the Census, “Programs to Improve Coverage in the 1980 Census,” by John H. Thompson. Evaluation and Research Reports, PHC80-E3. 1986 Thompson, John H. and David Franklin, ‘Test Census Results and Applications for the 1990 Planning.” Paper presented at the Census Bureau Second Annual Research Conference, Reston, Virginia, March, 1986. 1984 Miskura, Susan M., John H. Thompson, Henry F. Woltman, “Uses of Sampling for the Census Count.” Paper presented at the Annual Meeting of the American Statistical Association, Philadelphia, Pennsylvania, August, 1984. Fan, Milton C., Martha L. Sutt, and John H. Thompson, “Evaluation of the 1980 Census Precanvass Coverage Improvement Program.” Paper presented at the Annual Meeting of the American Statistical Association, Philadelphia, Pennsylvania, August, 1984. Keeley, Catherine and John H. Thompson, “The 1980 Census Nonhousehold Sources Program.” Paper presented at the Annual Meeting of the American Statistical Association, Philadelphia, Pennsylvania, August, 1984. Case 5:20-cv-05799-LHK Document 36-2 Filed 08/25/20 Page 23 of 23 John H. Thompson Page 7 1983 Miskura, Susan M. and John H. Thompson, “1980 Census Findings and Their Implications for 1990 Census Planning.” Presented at the Joint Statistical Meetings, Toronto, Canada, August, 1983. Taeuber, Cynthia and John H. Thompson, “1980 Census Data: The Quality of the Data and Some Anomalies.” Paper presented at the Annual Meeting of the Population Association of America, April, 1983. 1982 Fan, Milton C., John H. Thompson, Jay Kim, and Henry F. Woltman, “Sample Design, Estimation and Presentation of Sampling Errors for the 1980 Census Early Publications National Sample.” Paper presented at the Annual Meetings of the American Statistical Association, Chicago, Illinois, August, 1982. 1981 Woltman, Henry F., Susan M. Miskura, John H. Thompson, and Peter A. Bounpane, “1980 Census Weighting and Variance Estimation Studies, Design and Methodology.” Paper presented at the Annual Meetings of the American Statistical Association, Detroit, Michigan, August, 1981. Kim, Jay, John H. Thompson, Henry F. Woltman, and Stephen M. Vajs, “Empirical Results from the 1980 Census Sample Estimation Study.” Paper presented at the Annual Meetings of the American Statistical Association, Detroit, Michigan, August, 1981. Fan, Milton, C., John H. Thompson, and Susan M. Miskura, “1980 Census Variance Estimation Procedure.” Paper presented at the Annual Meetings of the American Statistical Association, Detroit, Michigan, August, 1981. Thompson, John H., “Convergence Properties of the Iterative 1980 Census Estimator.” Paper presented at the Annual Meetings of the American Statistical Association, Detroit, Michigan, August, 1981. 1978 Thompson, John H., “The Nonhousehold Sources Program.” Paper presented at the Annual Meetings of the American Statistical Association, San Diego, California, August, 1978.