Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 1 of 37 1 2 3 4 5 6 7 8 9 10 JEFFREY B. CLARK Acting Assistant Attorney General DAVID MORRELL Deputy Assistant Attorney General ALEXANDER K. HAAS Branch Director DIANE KELLEHER BRAD P. ROSENBERG Assistant Branch Directors M. ANDREW ZEE ALEXANDER V. SVERDLOV Trial Attorneys U.S. Department of Justice Civil Division - Federal Programs Branch 1100 L Street, NW Washington, D.C. 20005 Telephone: (202) 305-0550 11 12 Attorneys for Defendants 13 14 IN THE UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION 15 16 17 NATIONAL URBAN LEAGUE, et al., 18 Plaintiff, 19 v. 20 21 22 WILBUR L. ROSS, JR., et al., Defendants. 23 24 25 26 27 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK Case No. 5:20-cv-05799-LHK DECLARATION OF ALBERT E. FONTENOT, JR. Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 2 of 37 I, Albert E. Fontenot, Jr., make the following Declaration pursuant to 28 U.S.C. § 1746, 1 2 and state that under penalty of perjury the following is true and correct to the best of my 3 knowledge and belief: 4 I. Executive Summary 1. 5 6 I am the Associate Director for Decennial Census Programs at the U.S. Census Bureau, and I submit this declaration to: x 7 Explain the magnitude, complexity, and planning involved in the 2020 decennial census, including the tightly integrated nature of census operations and processing; 8 9 x Detail the changes made to the original design in light of the COVID-19 pandemic; and 10 x Discuss the impacts of extending field operations past their current end date of September 30, 2020. 11 12 13 II. Qualifications 2. I am the Associate Director for Decennial Census Programs, in which capacity I 14 serve as adviser to the Director and Deputy Director of the Census Bureau on decennial programs. 15 In this role, I provide counsel as to the scope, quality, management and methodology of the 16 decennial census programs; provide executive and professional leadership to the divisions and 17 central offices of the Decennial Census Programs Directorate; and participate with other 18 executives in the formulation and implementation of broad policies that govern the diverse 19 programs of the Census Bureau. I have served in this capacity since November 12, 2017. 20 3. I began my career with the Census Bureau after retiring from a successful 40-year 21 career as a senior executive in the private sector with midsize manufacturing companies where I 22 was responsible for providing visionary leadership, developing innovative corporate growth and 23 development strategies. I served as Vice President of Marketing, Vice President of Research and 24 Development, and, for the last 14 years, as President and Chief Executive Officer. 25 4. In addition to a successful corporate career I served as Adjunct Professor in the 26 MBA program in the Keller Graduate School of Management from 2005–2013 where I taught 27 Leadership and Organizational Development, Marketing Management, Corporate Finance, 28 Statistics, and Marketing. I earned a BA in management and MBA in management and finance DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 1 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 3 of 37 1 from DePaul University and Doctor of Ministry in pastoral ministry from Bethel Theological 2 Seminary 3 5. I served as a as a commissioned officer in U. S. Army and was decorated in combat 4 in Vietnam. After leaving active service, I remained in the US Army reserve attaining the rank of 5 Major. 6 6. After retirement from private sector corporate management, I began my career with 7 the Census Bureau in 2009 as a Field Operations Supervisor in Southern California for the 2010 8 Census. I quickly rose through the ranks and managed the Non-response follow-up operations for 9 the 2010 Census as Area Manager responsible for census activities in Los Angeles County, the 10 State of Hawaii, San Bernardino County and Riverside County California. After 2010, I served in 11 positions of increasing responsibility as Survey Supervisor, Senior Supervisory Survey 12 Statistician, Assistant Regional Director for the Los Angeles Region, and Regional Director for 13 the Chicago Region. I moved from the field to the Census Bureau headquarters to assume the 14 position as Chief of the Field Division and subsequently Assistant Director of Field Operations, 15 Assistant Director for Decennial Census Operations, then Associate Director for the Decennial 16 Census. 17 7. 18 19 From 2012–2016, I represented the Field Directorate on the team that developed and wrote the Operations plan for the 2020 Decennial Census. 8. I have in-depth firsthand knowledge about the planning, management, and 20 execution of Census Bureau field operations and effective mission-oriented leadership. I serve as 21 the Chairman of the Census Crisis Management Team; I served as a member of the 2020 Census 22 Design Executive Guidance Group; I am a member of the Census Data Quality Executive 23 Guidance Group; and I chair the 2020 Census Operations Planning Group. Additionally, I 24 represent the Decennial Census Program in our engagement with two of the three committees that 25 advise the Census Bureau: the Census Scientific Advisory Committee and the National Advisory 26 Committee. 27 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 2 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 4 of 37 1 2 III. A Complex Design and Budget for the 2020 Census 9. The Census Bureau goes to extraordinary lengths to count everyone living in the 3 country once, only once, and in the right place, including those in hard-to-count populations. This 4 is the core mandate of the Census Bureau, and has been the most significant factor informing every 5 decision made in designing, planning, testing, and executing the decennial Census. 6 10. The Census Bureau’s mandate in conducting the decennial census is to count 7 everyone living in the United States, including the 50 states, the District of Columbia, and the 8 territories of Puerto Rico, American Samoa, Commonwealth of the Northern Mariana Islands, 9 Guam, and U.S. Virgin Islands. To that end, we expend significant funds, efforts, and resources 10 in capturing an accurate enumeration of the population, including those who are hard to count. In 11 particular, the 2020 Census operational design considers population groups that have historically 12 been hard to count, as well as population groups that may emerge as hard to count. 13 11. The planning, research, design, development, and execution of a decennial census 14 is a massive undertaking. The 2020 decennial census consists of 35 operations utilizing 52 separate 15 systems. Monitoring the status and progress of the 2020 Census—the operations and systems—is 16 managed in large part using a master schedule, which has over 27,000 separate lines of census 17 activities. Thousands of staff at Census Bureau headquarters and across the country support the 18 development and execution of the 2020 census operational design, systems, and procedures. In 19 addition, the 2020 Census requires the hiring and management of hundreds of thousands of field 20 staff across the country to manage operations and collect data in support of the decennial census. 21 12. The 2020 Census operational design is tailored to enumerate all persons, including 22 hard-to-count populations. Almost every major operation in the 2020 Census contains components 23 designed to reach hard-to-count populations. This includes: census outreach, census content and 24 forms design, finding addresses for enumeration, field infrastructure, multiple modes for self- 25 response, Non-Response Follow-Up (NRFU) operations that enumerate households that did not 26 self-respond to the census, and other operations designed specifically for the enumeration of 27 population groups that have been historically hard to count. The best explanation of the many 28 integrated operations designed to reach these populations is set forth in Appendix B to Version 4.0 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 3 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 5 of 37 1 of the 2020 Census Operation Plan, available at https://www.census.gov/programs- 2 surveys/decennial-census/2020-census/planning-management/planning-docs/operational- 3 plan.html. Examples include: 4 x Verifying address lists using address data provided by community organizations, 5 satellite technology, and in-person address listers checking addresses in communities 6 nationwide; 7 x In-person enumeration using paper questionnaires in areas such as Remote Alaska; 8 x Hand-delivering 2020 Census materials to areas impacted by natural disasters, such as those impacted by Hurricane Michael in Florida; 9 10 x Conducting a special operation to count persons in “Group Quarters.” Group Quarters 11 include places such as college or university student housing, nursing homes, and 12 corrections facilities; 13 x Working with local partners to identify locations, like shelters and soup kitchens, to best count people experiencing homelessness; and 14 15 x 16 13. Creating culturally relevant advertisements targeting hard-to-count communities. The Census Bureau obtained approval under the Paperwork Reduction Act from 17 the Office of Management and Budget for the data collections involved in the 2020 Census. The 18 Operational Plan is a project management document and, as in prior censuses, we did not obtain 19 clearance for it. We presented information about our plans as we developed them in quarterly 20 public Project Management Reviews, and we obtained input on our plans from both our Census 21 Scientific Advisory Committee and National Advisory Committee. We consulted with other 22 agencies throughout the decade about data security, postal delivery, acquisition of records, and the 23 like, though we did not ask other agencies to review or approve our project management plans. 24 14. We allocate vast resources to ensure as complete and accurate a count as possible. 25 Research and testing, in addition to the Census Bureau’s collective knowledge and experiences, 26 has resulted in an effective approach to reach all population groups. 27 28 15. The complexity and inter-related nature of census operations is echoed in the budget for the 2020 Census. The overall budget estimate for the 2020 Census—covering fiscal DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 4 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 6 of 37 1 years 2012 to 2023—is $15.6 billion. This represents enough funding to successfully complete 2 the 2020 Census in virtually all possible scenarios, including the current challenging 3 circumstances. In fact, the Government Accountability Office (GAO) recently reviewed this 4 budget estimate1 and determined, as of January 2020, that the estimate substantially or fully met 5 GAO’s standards and best practices for a reliable cost estimate in terms of credibility, accuracy, 6 completeness, and documentation quality. It is rare for civilian agencies to be so designated, and 7 we are proud that the Census Bureau has achieved this status. 16. 8 As of this writing, the Census Bureau has been appropriated in aggregate just under 9 $14 billion to use for the 2020 Census, covering fiscal years 2012 through 2020. This is $4.4 10 billion greater in appropriated dollars than the $9.6 billion actually expended from fiscal years 11 2002 to 2010 for the 2010 Census. 17. 12 Combined, prior to the COVID-19 pandemic operational adjustments, there remain 13 just over $2 billion in contingency funds that have been appropriated, but which we have not 14 needed to use. With only minimal exceptions, Congress appropriated these funds to allow us to 15 flexibly and quickly respond to any and all risks to the 2020 Census that might be realized and 16 have an impact on the operations. 18. 17 That is exactly what the Census Bureau has done in these challenging times. We 18 have always planned to exhaust any resources necessary to fulfill the Census Bureau’s mission in 19 counting everyone living in the United States once, only once, and in the right place. In all 20 scenarios, the focus of our resources includes the hard-to-count. We have designed and 21 implemented the 2020 Census to enumerate the most willing and able to respond in our most 22 efficient and cost effective manner, thereby freeing the majority of our resources to reach hard-to- 23 count communities using a bevy of in-person techniques specifically tailored to reach them. 24 IV. 25 26 27 28 Census Step 1: Locating Every Household in the United States 19. The first operational step in conducting the 2020 Census was to create a Master Address File (MAF) that represents the universe of addresses and locations to be counted in the 1 This is known as the 2020 Census Life Cycle Cost Estimate (LCCE) Version 2.0. An executive summary of that estimate is publicly available at https://www2.census.gov/programssurveys/decennial/2020/program-management/planning-docs/life-cycle-cost-estimate_v2.pdf. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 5 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 7 of 37 1 2020 Census. This operation constitutes a significant part of the 2020 Census, and our plans to 2 enumerate every resident once, only once, and in the right place. 3 20. A national repository of geographic data—including addresses, address point 4 locations, streets, boundaries, and imagery—is stored within the Census Bureau’s Master Address 5 File/Topologically Integrated Geographic Encoding and Referencing (MAF/TIGER) System. The 6 MAF/TIGER System provides the foundation for the Census Bureau’s data collection, tabulation, 7 and dissemination activities. It is used to generate the universe of addresses that will be included 8 in a decennial census. Those addresses are then invited to respond, typically through an invitation 9 in the mail. The MAF/TIGER System is used to control responses as they are returned to the 10 Census Bureau and to generate a list of nonresponding addresses that will be visited in person. 11 Finally, the MAF/TIGER System is used to ensure that each person is tabulated to the correct 12 geographic location as the final 2020 Census population and housing counts are prepared. 13 21. For all of these reasons, the Census Bureau implemented a continuous process for 14 address list development in preparation for the 2020 Census. There are two primary components 15 to address list development—in-office development and in-field development. 16 development involves the regular, on-going acquisition and processing of address information 17 from authoritative sources, such as the U.S. Postal Service (responsible for delivering mail to 18 addresses on a daily basis), and tribal, state, and local governments (responsible for assignment of 19 addresses to housing units), while in-field address list development involves individuals traversing 20 a specified geographic area and validating or updating the address list based on their observations 21 and, if possible, interaction with residents of the housing units visited. 22 22. In-office Between 2013 and 2019, the Census Bureau accepted nearly 107 million address 23 records from government partners. Over 99.5 percent of those records matched to addresses 24 already contained in the MAF, many of which were obtained from the U.S. Postal Services’ 25 Delivery Sequence File (DSF). The remaining 0.5 percent of address records from partner 26 governments represented new addresses and were used to update the MAF. In addition, partners 27 submitted over 75 million address points that were either new or enhanced existing address point 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 6 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 8 of 37 1 locations in TIGER. Over 257,000 miles of roads were added to TIGER using data submitted by 2 partners. 3 23. For the third decade, as mandated by the Census Address List Improvement Act of 4 1994, the Census Bureau implemented the Local Update of Census Addresses (LUCA) Program 5 to provide tribal, state, and local governments an opportunity to review and update the Census 6 Bureau’s address list for their respective jurisdictions. In 2018, participants from over 8,300 7 entities provided 22 million addresses, of which 17.8 million (81 percent) matched to addresses 8 already in the MAF. The Census Bureau added 3.4 million new addresses to the MAF, nationwide, 9 as a result of LUCA. 10 24. Between September 2015 and June 2017, the Census Bureau conducted a 100 11 percent in-office review of every census block in the nation (11,155,486 blocks), using two 12 different vintages of imagery (one from 2009, which was contemporary with the timing of address 13 list development and Address Canvassing for the 2010 Census, and one concurrent with the day 14 on which in-office review occurred) and housing unit counts from the MAF. The 2009-vintage 15 imagery was acquired from a variety of sources, including the National Agricultural Imagery 16 Program as well as publicly available imagery from state and local governments. Current imagery 17 was acquired through the National Geospatial Intelligence Agency’s Enhanced View Program, 18 through which federal agencies can access imagery of sufficiently high quality and resolution to 19 detect individual housing units and other structures, driveways, roads, and other features on the 20 landscape. 21 25. During the in-office review, clerical staff had access to publicly available street- 22 level images through Google Street View and Bing StreetSide, which provided the ability to see 23 the fronts of structures, as if standing on the sidewalk. The technicians categorized blocks as 24 passive, active, or on-hold. Passive blocks represented stability, meaning the technician verified 25 the currency and accuracy of housing data in the office. Active blocks represented evidence of 26 change and/or coverage issues in the MAF. On-hold blocks represented a lack of clear imagery. 27 In these latter two instances, In-Field Address Canvassing was required. At the end of the initial 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 7 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 9 of 37 1 review in June 2017, 71 percent of blocks were classified as passive, suggesting a need for in-field 2 review of only 29 percent of blocks. 3 26. However, since the 2020 Census was still several years away when In-Office 4 Address Canvassing completed its initial review of the nation, the Census Bureau continued the 5 in-office review to ensure the MAF was keeping up with changes on the ground. The Census 6 Bureau used information from the U.S. Postal Services’ DSF and partner governments to identify 7 areas experiencing recent change and triggered these areas for re-review. Between July 2017 and 8 March 2019, the additional review resulted in the categorization of nearly 87.9 percent of the 11.1 9 million census blocks as passive, indicating a need for in-field review of only 12.1 percent of 10 census blocks. 11 27. In-Field Address Canvassing occurred between August 2019 and October 2019. 12 Of the 50,038,437 addresses in the universe, fieldwork validated 44,129,419 addresses (88.2 13 percent). The remainder were removed from the universe as deletes, duplicates, or non-residential 14 addresses. There were 2,685,190 new addresses identified during fieldwork, of which 1,553,275 15 matched addresses already in the MAF as a result of contemporaneous in-office update processes. 16 In other words, even the hardest to count areas that required fieldwork to verify the addresses, 17 resulted in only a small percentage of additions to the existing MAF. 18 28. The design for address list development in the decade leading up to the 2020 Census 19 was the most comprehensive in history. Extensive partnerships with tribal, federal, state, and local 20 governments provided multiple opportunities to validate and update the MAF using the most 21 authoritative sources available. This process of continual assessment and update using partner- 22 provided data created a strong foundation on which to implement the use of satellite imagery to 23 validate existing addresses or detect change during In-Office Address Canvassing. This suite of 24 in-office methods allowed the Census Bureau to focus In-Field Address Canvassing resources in 25 the hardest to validate census blocks. 26 29. The MAF/TIGER System created the foundation for the 2020 Census. The Census 27 Bureau believes that the Census Bureau’s MAF/TIGER System is the most complete and accurate 28 in history. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 8 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 10 of 37 1 2 V. Census Step 2: Encouraging Self-response Throughout the 2020 Census 30. In order to encourage everyone in the United States to self-respond, the Census 3 Bureau designed, tested, and implemented and Integrated Communications Program, the IPC. The 4 two major components of this program are the ICC, the Integrated Communications Contract, and 5 the IPP, the Integrated Partnership Program. A. 6 7 31. Advertising and Media The ICC is the major contract that supports all components of the communications 8 campaign for the 2020 Census. For the 2020 Census, the push to educate people and motivate 9 response to the 2020 Census represented the largest advertising campaign in U.S. government 10 11 history. 32. The budget for the 2020 Integrated Communications Contract is currently funded 12 at a higher level than in the 2010 Census, adjusted for both inflation and population growth. The 13 cost of the 2010 Census Integrated Communications Contract, in 2020 constant dollars, would be 14 $456 million, while the Census Bureau currently plans to spend approximately $695 million on 15 the 2020 Census Integrated Communications Contract. 16 communications program will mean an 18% increase in per-person spending over the 2010 17 amount. 18 33. The $695 million spent on the To run the ICC in connection with the Census Bureau, a contract was awarded to 19 VMLY&R, a major legacy-advertising firm with over 80 years of experience. Known as Team 20 Y&R, or TYR, by the Census Bureau, the contracting team includes 13 subcontractors. TYR 21 includes firms with expertise in reaching and working with the major audiences that will receive 22 advertising through the media outlets directed toward their population groups, including the 23 Black/African American, Hispanic/Latino, Asian, American Indian and Alaska Native, and Native 24 Hawaiian and Other Pacific Islander populations. By relying on firms with these individual skill 25 sets, the Census Bureau was able to better tailor the media and messaging toward individual groups 26 and gauge the response before going live with the advertising. It also allowed for more creative 27 risk-taking, and less of a one-size-fits-all approach. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 9 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 11 of 37 1 34. Every part of the 2020 Census communications program was grounded in research. 2 Based on the commitment to being a data driven campaign, beginning in 2018, we extensively 3 researched how people perceived the census and what would motivate them to complete it. Models 4 were developed to predict areas and audiences of low response across the country. These models 5 were then translated into “low response scores” that help the Census Bureau anticipate respondent 6 behavior so that messaging, media, and other communications activities could be deployed to 7 maximize impact. 8 35. As a result of that research, we mounted a media campaign with stories in news 9 media across the country in print, social, and digital media. The campaign was tested in over 120 10 focus groups across the country, and driven by efforts to reach historically undercounted 11 audiences. More than 1,000 advertisements, in English and 43 other languages, were developed 12 to communicate the importance of responding to the 2020 Census. This compares to roughly 400 13 separate creative pieces created in 2010. A sample of these creative pieces can be seen on the 14 Census Bureau’s YouTube channel website. 15 36. On March 29, 2019, the Census Bureau launched 2020census.gov—a key 16 information hub about the census, how to complete it, and how it will affect communities across 17 the country. Three days later, on April 1, 2019, we held a press conference to unveil the campaign 18 platform: "Shape Your Future. START HERE." On January 14, 2020, we unveiled highlights of 19 the public education and outreach campaign. That same day, we began airing ads to reach 99 20 percent of the nation's 140 million households, including historically undercounted audiences and 21 those that are considered hard to reach. 22 37. The massive multimedia campaign sought to engage stakeholders and partners, 23 support recruitment efforts and the Statistics in Schools program, and communicate the importance 24 of the census through paid advertising, public relations, social media content, and the new web 25 site. This was the first census where we made a significant investment in digital advertising, and 26 spending time and resources targeting online sites including Facebook, Instagram, paid search 27 engines, display ads, and programmatic advertising. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 10 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 12 of 37 1 38. The push to have a greater digital presence allowed the Census Bureau to reach a 2 mobile audience, tailor messages, micro-target, and shift campaign ads and messages as needed. 3 Online media, particularly search engines and social networking sites, made up a significant 4 portion of digital connections. Nearly every person living in the United States was reached an 5 average of 40 times throughout the campaign, from television, radio, newspaper and online ads, as 6 well as outdoor locations such as billboards and bus stops. 7 39. The Census Bureau adapted its outreach strategies in response to delayed census 8 operations due to COVID-19, increasing advertising and outreach to specific areas of the country 9 with lower response rates. We quickly adjusted our messaging, pivoting from our original 10 campaign to encourage people to respond online from the safety of their own homes. The use of 11 micro-targeting allowed the Census Bureau to tailor its messaging, including directing appropriate 12 messages to hard-to-reach communities and those who distrust government, both of which have 13 been traditionally undercounted. This targeting continues through NRFU as we encourage the 14 public to cooperate with enumerators. This targeting has allowed us to make each dollar spent on 15 the advertising campaign more effective than in any previous census. B. 16 17 40. Partnerships with Community Organizations The second major element of the Integrated Communications Program is 18 partnerships. There are two prongs to the Partnership Program, the National Partnership Program 19 that works from Census Bureau headquarters mobilizing national organizations, and the 20 Community Partnership and Engagement Program, that works through the regions at the local level 21 to reach organizations that directly touch their communities. The National Partnership Program 22 and Community Partnership and Engagement Program are more integrated than ever before, and 23 numbers involved for both programs significantly exceed the totals reached in prior censuses. 24 41. Census partners include national organizations like the National Urban League, the 25 Mexican American Legal Defense Fund, the National Association of Latino Elected Officials 26 (NALEO), the National Association for the Advancement of Colored People (NAACP), and the 27 U.S. Chambers of Commerce. Major corporations also become census partners. At the local level, 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 11 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 13 of 37 1 partners can be churches, synagogues and mosques, legal aid clinics, grocery stores, universities, 2 colleges, and schools. 42. 3 Partners are the trusted voices in their communities; they have a profound impact 4 on those who listen when they say the census is important and safe. We depend on our partners to 5 seal the deal with communities that may be fearful or distrustful of the government. Even with all 6 the Census Bureau’s innovation and improvements to the self-response system, we have learned— 7 and confirmed through research—that when communities and leaders recognize the importance of 8 participating in the census, this message is better conveyed to households within those 9 communities. The best, most trusted information comes from a person of trust. 10 11 VI. Census Step 3: Self-Response 43. The design of the 2020 Census depends on self-response from the American public. 12 In an effort to ensure the most efficient process to enumerate households, the Census Bureau 13 assigns every block in the United States to one specific type of enumeration area (TEA). The TEA 14 reflects the methodology used to enumerate the households within the block. There are two TEAs 15 where self-response is the primary enumeration methodology: TEA 1 (Self-Response) and TEA 16 6 (Update Leave). 17 44. TEA 1 uses a stratified self-response contact strategy to inform and invite the public 18 to respond to the census, and to remind nonresponding housing units to respond. Invitations, 19 reminders, and questionnaires will be delivered on a flow basis unless a household responds. 20 These mailings are divided into two panels, Internet First and Internet Choice. Internet First 21 emphasizes online response as the primary self-response option. Mailings to the Internet First 22 panel begin with an invitation letter that alerts the housing unit to the beginning of the 2020 Census 23 and provides the Census ID,2 the URL for the online questionnaire, and information for responding 24 by phone. 25 26 27 28 45. Internet Choice is targeted to areas of the nation that we believe are least likely to respond online. Historical response rates from other Census Bureau surveys, internet access and 2 A Census ID is a unique identifier assigned to each address in a decennial census; the Census ID is used to track whether an address has self-responded or to track the address through nonresponse data collection and, ultimately through response processing and data tabulation. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 12 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 14 of 37 1 penetration, and demographics are used to determine those areas least likely to respond online. 2 Mailings to the Internet Choice panel begin with an invitation letter that alerts the housing unit to 3 the beginning of the 2020 Census and provides the Census ID and the URL for the online 4 questionnaire, information for responding by phone, and also a paper questionnaire. Housing units 5 in Internet Choice areas have the choice to respond on paper beginning with the initial contact. 6 All nonresponding housing units, regardless of panel, receive a paper questionnaire after the initial 7 mailing and two separate reminder mailings. 46. 8 Update Leave (TEA 6) is conducted in areas where the majority of the housing units 9 do not have mail delivery to the physical location of the housing unit, or the mail delivery 10 information for the housing unit cannot be verified. The purpose of Update Leave is to update the 11 address list and feature data, and to leave a 2020 Census Internet Choice package at every housing 12 unit. The major difference from TEA 1 is that a Census Bureau employee, rather than a postal 13 carrier, delivers the 2020 Census invitation to respond, along with a paper questionnaire. Housing 14 units also have the option to respond online or by phone. 47. 15 Self-response began in March 2020 and will continue until the end of data 16 collection. The total self-response period for the 2020 Census will be longer than the 2010 self- 17 response period. 18 VII. 19 Census Step 4: Nonresponse Followup (NRFU) 48. NRFU is the field operation designed to complete enumeration of nonresponding 20 housing unit addresses. The primary purpose of NRFU is to conduct in-person contact attempts at 21 each and every housing unit that did not self-respond to the decennial census questionnaire. 22 49. After giving everyone an opportunity to self-respond to the census, census field 23 staff (known as enumerators), attempt to contact nonresponding addresses to determine whether 24 each address is vacant, occupied, or does not exist, and when occupied, to collect census response 25 data. Multiple contact attempts to nonresponding addresses may be needed to determine the 26 housing unit status and to collect decennial census response data. 27 28 50. The 2020 Census NRFU operation is similar to the 2010 Census NRFU operation, but improved. In both the 2010 Census and the 2020 Census, cases in the NRFU workload are DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 13 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 15 of 37 1 subject to six contact attempts. In both the 2010 and 2020 NRFU, the first contact attempt is 2 primarily an in-person attempt. In the 2010 Census, these six contact attempts could be conducted 3 as three in-person attempts and three attempts by telephone. By comparison, each contact attempt 4 in the 2020 Census NRFU will be either a telephone or an in-person contact attempt (however the 5 vast majority of attempts will be in-person). 6 51. In both the 2010 Census and 2020 Census NRFU, if upon the first contact attempt 7 an enumerator determines an address is occupied and the enumerator is able to obtain a response 8 for the housing unit, then the housing unit has been counted, and no follow-up is needed. 9 52. If upon the first contact attempt, the enumerator is not able to obtain a response, the 10 enumerator is trained to assess whether the location is vacant or unoccupied. Enumerators will 11 use clues such as empty buildings with no visible furnishings, or vacant lots, to identify an address 12 as vacant or non-existent. 13 53. In both the 2010 and 2020 Census, a single determination of a vacant or nonexistent 14 status was not sufficient to remove that address from the NRFU workload; a second confirmation 15 is needed. If a knowledgeable person can confirm the enumerator’s assessment, the address will 16 be considered vacant or non-existent and no additional contact attempts are needed. 17 knowledgeable person is someone who knows about the address as it existed on census day or 18 about the persons living at an address on census day. A knowledgeable person could be someone 19 such as a neighbor, a realtor, a rental agent, or a building manager. This knowledgeable person is 20 known as a proxy respondent. 21 54. A If a knowledgeable person cannot be found to confirm the status of vacant or non- 22 existent, use of administrative records may provide confirmation of the enumerator’s assessment. 23 The Census Bureau does not rely on a single administrative records source to determine an address 24 is vacant or non-existent. Rather, multiple sources are necessary to provide the confidence and 25 corroboration before administrative records are considered for use. When used in combination 26 with an enumerator’s assessment of vacant or non-existent, corroborated administrative records 27 provide the second confirmation that a nonresponding address is vacant or non-existent. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 14 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 16 of 37 1 55. If, upon the first in-person contact attempt, the enumerator believes the address is 2 occupied, but no knowledgeable person is available to complete the enumeration, the Census 3 Bureau will use consistent and high-quality administrative records from trusted sources as the 4 response for the household and no further contact will be attempted. We consider administrative 5 records to be of high quality if they are corroborated with multiple sources. Examples of high- 6 quality administrative records include Internal Revenue Service Individual Tax Returns, Internal 7 Revenue Service Information Returns, Center for Medicare and Medicaid Statistics Enrollment 8 Database, Social Security Number Identification File, and 2010 Census data. 9 56. Regardless of whether administrative records are used as a confirmation of vacancy 10 or non-existent status or for the purposes of enumerating an occupied housing unit, the Census 11 Bureau will, as a final backstop, send a final mailing encouraging occupants, should there be any, 12 to self-respond to the 2020 Census. 13 57. The vast majority of nonresponding addresses in the NRFU workload will require 14 the full battery of in-person contact attempts to determine the status of the nonresponding address 15 (vacant, occupied, does not exist) and to collect 2020 Census response data. The full battery of 16 in-person contact attempts also includes the ability to collect information about persons living in 17 a nonresponding housing unit from a proxy respondent. Nonresponding units become eligible for 18 a proxy response after a pre-determined number of unsuccessful attempts to find residents of a 19 nonresponding address. 20 58. The operational design for NRFU evolved over the course of the decade. Use of 21 administrative records, field management structures, systems, procedures, data collection tools and 22 techniques were proven in tests occurring in 2013, 2014, 2015, 2016, and 2018. 23 24 VIII. 59. Census Step 5: Quality Control The Census Bureau is committed to a quality NRFU operation and has in place 25 several programs to monitor and promote quality, such as the NRFU Reinterview Program, the 26 Decennial Field Quality Monitoring Operation, and the Coverage Improvement Operation. 27 28 60. The NRFU Reinterview Program involves contacting a small number of households to conduct another interview—to help us ensure that enumerators are conducting their jobs DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 15 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 17 of 37 1 correctly and are not falsifying responses. We have streamlined this operation, using information 2 collected from the mobile devices used by enumerators. The data from these mobile devices tell 3 us where the enumerators were physically located while they were conducting the interviews, how 4 long they spent on each question in the interview, time of day of the interview, and other detail 5 data about the interview process. Having this information—which is new for the 2020 Census— 6 provides management with information on how the census takers are doing their jobs, and allows 7 us to select reinterview cases in a targeted fashion. 8 61. A second quality check program, new for the 2020 Census, is the Decennial Field 9 Quality Monitoring operation. This operation monitors overall adherence to field procedures in 10 order to identify unusual patterns. We used this near real-time data analysis successfully during 11 the Address Canvassing operation in 2019, and it is currently active in the NRFU operation. The 12 goal of the program is to identify and investigate potential quality issues. In this program we 13 examine data from individual field representatives and larger scale data, scanning for the 14 possibility of both individual and systemic data quality problems. The program monitors outlier 15 metrics, and produces reports that we analyze on a daily basis. Management staff use these reports 16 to investigate suspicious activities and follow up as needed. 17 62. Another quality check operation, the Coverage Improvement Operation, seeks to 18 resolve erroneous enumerations (people who were counted in the wrong place or counted more 19 than once) and omissions (people who were missed) from all housing unit data. Coverage 20 Improvement will attempt to resolve potential coverages issues identified in responses from 21 the Internet Self-Response, Census Questionnaire Assistance, and NRFU operations, as well as 22 from the paper questionnaires. 23 63. The Census Bureau believes that these quality programs (Reinterview, Decennial 24 Field Quality Monitoring, and Coverage Improvement), taken together, provide a robust quality 25 check for our data collection operations. We believe that our quality program remains an effective 26 deterrent to poor performance, and an appropriate method to identify enumerators who fail to 27 follow procedures. None of these programs, to date, reveals a pattern of substandard data 28 collection. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 16 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 18 of 37 64. 1 The Census Bureau has also formed a Data Quality Executive Guidance Group that 2 brings together the Census Bureau’s experts in the fields of census operations, statistical 3 methodology, acquisition and utilization of administrative records, and in the social, economic and 4 housing subject areas. The group’s mission is to provide direction and approvals about quality 5 assessments of changes to the operational plans and of the 2020 Census data during and post data 6 collection. We plan to release Demographic Analysis estimates of the population in December, 7 prior to the release of the apportionment counts, as previously planned. 65. 8 Finally, as noted by the Director in his August 3, 2020 statement, the Census Bureau 9 intends to meet a similar level of household responses as in prior censuses, meaning that we will 10 resolve 99% of the cases in each state. In short, the Census Bureau has robust programs in place 11 to monitor data quality and has no indication that its NRFU operation is collecting “substandard” 12 data. 13 IX. 14 Census Step 6: Post-data Collection Processing 66. The next major step in the census, after the completion of data collection operations, 15 is post processing. Post processing refers to the Census Bureau’s procedures to summarize the 16 individual and household data that we collect into usable, high quality tabulated data products. 17 Our post processing procedures and systems are meticulously designed, tested and proven to 18 achieve standardized, thoroughly vetted, high quality data products that we can stand behind. 19 67. Post data collection processing is a particularly complex operation, and the steps of 20 the operation must generally be performed consecutively. It is not possible, e.g., to establish the 21 final collection geography for the nation prior to processing housing units and group quarters that 22 are added or corrected during NRFU. Similarly, it is not possible to unduplicate responses prior 23 to processing all non-ID responses. In this sense, the post data collection activities are like building 24 a house – one cannot apply dry wall before erecting the walls, any more than one could lay floor 25 tile before the floor is constructed. There is an order of steps that must be maintained. 26 68. As part of developing the Replan Schedule, we looked at the possibility of starting 27 the post data collection processing activities on a flow basis and reaffirmed that there is no 28 opportunity to begin the post data collection processing until data collection operations close DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 17 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 19 of 37 1 everywhere. For example, we cannot begin processing in one region of the country while another 2 region is still collecting data. This is true because the first post processing step is geographic 3 processing, which cannot begin until the entire universe is determined. Geographic processing is 4 key because we must tabulate census results at the block level and then build to higher levels of 5 geography such as block groups, tracts, counties, and states. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 69. The information below provides additional detail about the post data collection activities under the Replan Schedule. A. Incorporate address updates from the field data collection operations into MAF/TIGER Original Dates: February 10 – August 10, 2020 Replan Dates: February 6– September 24, 2020 During the data collection operations, the census field staff can update address and physical location information and add addresses. These updates are incorporated into our address and geo-spatial MAF/TIGER databases. Once updated, each address must be associated to the correct state, county, tract, block group and block. Since it is critical to associate each address to the correct geography, we verify that the address and geo-spatial updates are incorporated correctly. B. Produce the Final Collection Geography MAF/TIGER Benchmark Original Dates: August 14 – September 1, 2020 Replan Dates: September 5 – 25, 2020 In preparation for the producing the final collection geography data files needed for producing the apportionment counts and redistricting data products, we create a benchmark of MAF/TIGER, which is a snapshot of the databases. C. Produce the Final Collection Address Data Products from MAF/TIGER 26 27 28 Original Dates: September 2 – 14, 2020 Replan Dates: September 26 – Oct 14, 2020 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 18 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 20 of 37 1 Once the benchmark has been created, the final collection geographic data files 2 are produced and verified. 3 4 5 6 7 8 9 D. Produce and review the Decennial Response File 1 (DRF1) Original Dates: September 15 – October 14, 2020 Replan Dates: October 14– November 8, 2020 The verified final collection geography data are integrated with the response data. Integration of these data is also verified to ensure accuracy. The next set of activities involves the standardization of the collected information. x 10 First we determine the final classification of each address as either a housing units or a group quarters facility. Addresses can change from a housing unit 11 to group quarters and vice versa. Initial status is set at the start of the data 12 collection operations as either a housing unit or group quarters. During the 13 enumeration operations, we collect information that informs us on the 14 classification. For a small number of addresses the classification may change, 15 for example a housing unit may have been turned into a small group home. 16 Based on the information collected we determine the status of every address 17 as either a housing unit of group quarters. 18 x Next, we identify each unique person on the housing unit returns. 19 x As part of NRFU operation, we conduct a reinterview of a sample of cases to 20 21 ensure quality. We incorporate the results of the reinterview. x 22 respondents can provide their data without their Census Identification Number 23 (ID). These cases are assigned an ID which associates them to the final 24 25 collection geography. x 26 27 As part of the Internet self-response option and telephone operation, Some group quarters will provide the information electronically. These files can contain duplicate records, so we need to remove the duplicates. x We also determine the population count for all group quarters. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 19 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 21 of 37 1 x We collect data in many ways, for example on-line, over the phone, on a 2 paper questionnaire, electronic administrative files, and in person using an 3 electronic questionnaire. As a result, we need to standardize the responses 4 across the modes of collection. 5 x Finally, for the operations that collect data on a paper questionnaire, some 6 housing units have more people than can fit on one paper questionnaire. The 7 census field staff will use multiple paper questionnaires to enumerate the 8 house. These continuation forms are electronically linked to form one 9 electronic form. 10 11 12 13 14 15 16 17 18 19 20 21 22 E. Produce and review the Decennial Response File 2 (DRF2) Original Dates: October 14 – November 4, 2020 Replan Dates: November 9 – 30, 2020 Once the previous step has been verified, we incorporate the results from the SelfResponse Quality Assurance operation. As part of the group quarters operations, we enumerate domestic violence shelters. Their locations and data are high sensitive and are handled with special procedures both in the field and in processing. Their data are incorporated at this point in the process. Finally, for a small number of addresses we receive multiple returns, for example where one person in a house completes the form on-line, and other completes the paper questionnaire. For these cases, we select a form that will be used as the enumeration of record. F. Produce and review the Census Unedited File (CUF) 23 24 25 26 27 28 Original Dates: November 4 – 30, 2020 Replan Dates: December 1 – 14, 2020 Once the previous step has been verified, we incorporate administrative records data as the response data for housing units were we do not have an enumeration and have high quality administrative records data. Next we determine the status DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 20 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 22 of 37 1 for every housing unit as occupied, vacant or non-existent. Non-existent units are 2 removed from future processing. For every occupied housing unit, the population 3 count is determined. For each person with write-in responses to the race and 4 Hispanic origin questions, we merge in the information from automated and 5 clerical coding operations. The coding operations assign a numerical value to the 6 write-in responses. At this point in the post-data collection activities, for every 7 housing unit and group quarter their location (state, county, tract, block group and 8 block) is assigned, their status (occupied, vacant or non-existent) is determined, 9 and in occupied addresses the number of persons is known. In addition, at the 10 person level the demographic information (relationship, age, date of birth, sex, 11 race and Hispanic origin along with write-in code values) and at the housing unit 12 level housing information (tenure) is determined. For the majority of these items, 13 the respondent provided the information. However, for a small number of people 14 and addresses the information may be missing or inconsistent with other provided 15 information, for example the Person 1’s spouse is five years old. The result of 16 these processes is a file that contains records for every housing unit and group 17 quarters along with person records for the people associated with the addresses. 18 Note that some of the demographic information and response to the tenure 19 question may be missing. 20 21 22 23 24 25 26 27 28 G. Produce, review and release the Apportionment Counts Original Dates: December 1 – 28, 2020 Replan Dates: Dec 15- 31, 2020 Once the CUF has been verified, the process goes down two paths. The first path is to determine the apportionment counts. Since every housing unit and group quarters has a population count and linked to a state, we can tabulation the state level population counts. In addition, we merge in the count of the Federally Affiliated Overseas population and the results of the Enumeration of Transitory Locations for each state. To ensure accuracy in the apportionment numbers, the DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 21 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 23 of 37 1 state counts including the overseas population and apportionment numbers are 2 verified by multiple independent ways. The results of the independent 3 verifications are compared and reconciled, if necessary. 4 5 X. Census Step 0: Research and Testing of the 2020 Census Design 70. The operational design of the 2020 Census, discussed above, has been subjected to 6 repeated and rigorous testing. Given the immense effort required to conduct the census, the 7 importance of the results, and the decade of work by thousands of people that goes into planning 8 and conducting the decennial census, the Census Bureau expends a significant amount of effort to 9 evaluate its planning and design to ensure that its operations will be effective in coming as close 10 as possible to a complete count of everyone living in the United States. Design and testing of the 11 2020 Census was an iterative process: after each test, we revised our plans and assumptions as 12 necessary. 13 71. Below are eight significant tests conducted prior to the 2020 Census. Seven of the 14 tests listed below directly contributed to the support of the NRFU operational design or the 15 infrastructure needed to support it. The eighth test pertained to In-Field Address Canvassing. 16 A. 2013 Census Test. The 2013 Census Test explored methods for using 17 administrative records and third-party data to reduce the NRFU workload. 18 Key objectives of the 2013 Census Test included: 19 i. Evaluate the use of administrative records and third-party data to 20 identify vacant housing units and remove them from the NRFU 21 workload; 22 ii. Evaluate the use of administrative records and third-party data to 23 enumerate nonresponding occupied housing units to reduce the NRFU 24 workload; 25 26 27 28 iii. Test an adaptive design approach for cases not enumerated with administrative records and third-party data; and iv. Test methods for reducing the number of enumeration contact attempts as compared with the 2010 Census. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 22 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 24 of 37 1 B. 2014 Census Test. The 2014 Census Test built upon the results from the 2013 2 Census Test specific to administrative records and third-party data usage to 3 reduce the NRFU workload. 4 included: Key objectives of the 2014 Census Test 5 i. Testing various self-response modes, including the Internet, 6 telephone, and paper, and response without a preassigned census 7 identifier; 8 ii. Testing the use of mobile devices for NRFU enumeration in the field; 9 iii. Continuing to evaluate the use of administrative records and third- 10 party data to remove cases (vacant and nonresponding occupied 11 housing units) from the NRFU workload; 12 13 iv. Testing the effectiveness of applying adaptive design methodologies in managing the way field enumerators are assigned their work; and 14 v. Examining reactions to the alternate contacts, response options, 15 administrative record use, and privacy or confidentiality concerns 16 (including how the Census Bureau might address these concerns 17 through micro- or macro-messaging) through focus groups. 18 C. 2014 Human-in-the-Loop Simulation Experiment (SIMEX). Key findings 19 20 21 22 23 24 25 26 27 28 included: i. Determination that the field management structure could be streamlined and the supervisor-to-enumerator ratios increased; ii. Messaging and alerts within the operational control system provided real-time and consistent communication; and iii. Smartphones were usable by all people—even those with little technology experience were able to adjust and adapt. D. 2015 Optimizing Self-Response Test. The objectives of this test included: i. Determining use of digital and target advertising, promotion, and outreach to engage and motivate respondents; DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 23 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 25 of 37 1 ii. Offering an opportunity to respond without a Census ID (Non-ID 2 Processing) and determine operational feasibility and potential 3 workloads around real-time Non-ID Processing; and 4 iii. Determining self-response and Internet response rates. 5 E. 2015 Census Test. The 2015 Census Test explored reengineering of the 6 roles, responsibilities, and infrastructure for conducting field data collection. 7 IT also tested the feasibility of fully utilizing the advantages of planned 8 automation and available real-time data to transform the efficiency and 9 effectiveness of data collection operations. The test continued to explore the 10 use of administrative records and third-party data to reduce the NRFU 11 workload. Key objectives included: 12 i. Continue testing of fully utilized field operations management system 13 that leverages planned automation and available real-time data, as 14 well as data households have already provided to the government, to 15 transform the efficiency and effectiveness of data collection 16 operations; 17 ii. Begin examining how regional offices can remotely manage local 18 office operations in an automated environment, the extent to which 19 enumerator and manager interactions can occur without daily face-to- 20 face meetings, and revised field staffing ratios; 21 iii. Reduce NRFU workload and increase productivity with the use of 22 administrative records and third-party data, field reengineering, and 23 adaptive design; and 24 iv. Explore reactions to the NRFU contact methods, administrative 25 records and third-party data use, and privacy or confidentiality 26 concerns. 27 F. 2016 Census Test. The 2016 Census Test tested different supervisor-to- 28 enumerator staffing ratios and incremental improvements and updates to the DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 24 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 26 of 37 1 field data collection software that guided an enumerator through interviews. 2 The 2016 Census Test also allowed the continued evaluation of the use of 3 administrative records to reduce the NRFU workload. Key NRFU objectives 4 included: i. Refining the reengineered field operations; 5 ii. Refining the field management staffing structure; 6 iii. Testing enhancements to the Operational Control System and field 7 data collection application; and 8 iv. Testing scalability of Internet and Non-ID Processing during self- 9 response using enterprise solutions. 10 11 Objectives related to self-response included: 12 i. Testing provision of language support to Limited English Proficient populations through partnerships and bilingual questionnaires; 13 14 ii. Testing the ability to reach demographically diverse populations; 15 iii. Testing deployment of non-English data collection instruments and contact strategies; and 16 17 18 iv. Refining Real-Time Non-ID processing methods, including respondent validation. 19 G. 2018 End-to-End Census Test. The 2018 End-to-End Census Test focused 20 on the system and operational integration needed to support the NRFU 21 operation. Nearly all 2020 system solutions supporting the NRFU operation 22 were deployed. The test also allowed continued evaluation of the NRFU 23 contact strategy. The objectives of this test included: 24 i. Testing and validating 2020 Census operations, procedures, systems, 25 and field infrastructure together to ensure proper integration and 26 conformance with functional and nonfunctional requirements. 27 H. Address Canvassing Test (conducted in the fall of 2016). The Address 28 Canvassing Test examined the effectiveness of the In-Office Address DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 25 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 27 of 37 1 Canvassing through the results of the In-Field Address Canvassing. The 2 objectives of the test included: 3 i. Implementing all In-Office Address Canvassing processes; 4 ii. Evaluating the effectiveness of online training for field staff; iii. Measuring the effectiveness of In-Office Address Canvassing through 5 In-Field Address Canvassing; and 6 7 iv. Integrating multiple information technology applications to create one 8 seamless operational data collection, control, and management 9 system. 10 11 XI. Current Status of 2020 Census Operations 72. As of September 2, 2020, over 96 million households, 65 percent of all households 12 in the Nation, have self-responded to the 2020 Census. Combining the households that self- 13 responded with those that field staff have enumerated under NRFU reveals that as of September 14 1, 2020 the Census Bureau has enumerated 84 percent of the nation’s housing units. 15 73. The Census Bureau is now roughly 3 ½ weeks into the 7 ½ week schedule for 16 conducting the NRFU operation. Under the Replan Schedule, NRFU is scheduled to last 7 ½ 17 weeks, not 6 weeks as some of Plaintiffs’ declarations state. As of September 1, 2020, we have 18 completed roughly 60% of the NRFU workload. We were helped in achieving this result by the 19 fact that we got a “head start” on data collection by beginning NRFU at select offices in July at a 20 “soft launch.” When we began NRFU in all areas on August 9 we had already enumerated over 3 21 million households. Additionally, over 80% of the households in 40 states have been enumerated 22 74. While the number of enumerators hired and deployed has not been at the level 23 anticipated, current progress indicates that we will nonetheless be able to complete NRFU before 24 September 30. 25 conducting continuous replacement training sessions to increase that number. 26 27 75. We currently have over 235,000 enumerators actively deployed, and we are The productivity rate for our enumerators thus far is substantially above the planned rate. Our plans assumed a productivity rate of 1.55 cases/hour, and 19 hours/week average hours 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 26 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 28 of 37 1 worked, whereas as of September 1, 2020 we have experienced a productivity rate of 2 approximately 2.32 cases/hour, and 20.1 hours/week averaged work hours. 76. 3 In sum, at our current rate we anticipate being able to conclude NRFU data 4 collection no later than September 30, 2020. 5 XII. 6 Replanning the Census – Multiple Times 77. The Census Bureau’s planning for the 2020 Census was, in my professional 7 opinion, excellent. Our plan was comprehensive and thoroughly tested. In March 2020, however, 8 it became clear that COVID-19 was a serious health issue, and we were forced to change our plans 9 around the time we began our self-response operation. 10 78. On March 18, 2020 the Census Bureau initially announced a two-week suspension 11 of field operations to protect the health and safety of our employees and the American public 12 because of the COVID-19 Pandemic. Self-response continued during this period through Internet, 13 telephone and paper questionnaires. On March 28, 2020 the Census Bureau announced an 14 additional two week suspension, until April 15, 2020. 15 79. At that time the career professional staff at the Census Bureau undertook the project 16 of replanning each of the field operations based on our best predictions of when we could safely 17 begin sending staff into the field to interact with the public. On April 13, 2020 staff finalized the 18 plan to adjust field operations, and I presented the plan to the Secretary of Commerce and 19 Department of Commerce management. The plan involved delaying our key high personal contact 20 operations by 90 days. Update Leave, which had started on March 15 and been stopped because 21 of COVID-19 on March 17, would resume pursuant to a new schedule beginning on June 13 and 22 concluding on July 9. In-person Group Quarters operations which had been scheduled from April 23 2 – June 5 would be rescheduled from July 1 – September 3, and our largest field operation, NRFU, 24 which was scheduled from May 13- July 31, would be moved to August 11- October 31. We 25 rescheduled self-response to conclude with the end of Field Operations so instead of ending on 26 July 31 as indicated in the original plan, it was extended to October 31. This schedule required 27 Congress to provide legislative relief from the statutory deadlines of December 31, 2020, for the 28 submission of the Apportionment counts to the President, and March 31, 2021, for the delivery of DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 27 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 29 of 37 1 redistricting data to the states. A request statutory relief from Congress was made for 120 days to 2 enable us to complete the field operations and post enumeration processing. 3 80. On April 13, 2020, the Secretary of Commerce and the Director jointly announced 4 the new Census Schedule and stated that they would seek statutory relief from Congress of 120 5 additional calendar days. This new schedule set a completion date for field data collection and 6 self-response of October 31, 2020. For clarity, I will refer to this as “the COVID Schedule.” The 7 COVID Schedule assumed Congressional action and called for the delivery of apportionment 8 counts to the President by April 30, 2021 (120 days after the statutory deadline) and redistricting 9 data files to the states no later than July 31, 2021. 10 81. Once it became apparent that Congress was not likely to grant the requested 11 statutory relief, in late July the career professional staff of the Census Bureau began to replan the 12 Census operations to enable Census to deliver the apportionment counts by the Statutory deadline 13 of December 31, 2020. On July 29, the Deputy Director informed us that the Secretary had directed 14 us, in light of the absence of an extension to the statutory deadline, to present a plan at our next 15 weekly meeting on Monday, August 3, 2020 to accelerate the remaining operations in order to 16 meet the statutory apportionment deadline. I gathered all the senior career Census Bureau 17 managers responsible for the 2020 Census at 8:00 a.m. on Thursday, July 30 and instructed them 18 to begin to formalize a plan to meet the statutory deadline. At that time I consulted with the 19 Associate Director of Communications and we directed that the COVID Schedule be removed 20 from our website while we replanned. We divided into various teams to brainstorm how we might 21 assemble the elements of this plan, and held a series of meetings from Thursday to Sunday. We 22 developed a proposed replan that I presented to the Secretary on Monday August 3. 23 82. In developing the proposed replan we considered a variety of options and evaluated 24 risk for each suggested time-saving measure. We evaluated the risks and quality implications of 25 each suggested time-saving measure and selected those that we believed presented the best 26 combination of changes to allow us to meet the statutory deadline without compromising quality 27 to an undue degree. The challenge was to shorten the field data collection operation by 30 days, 28 and to conclude the post processing operation in only 3 months, as opposed to 5 months in prior DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 28 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 30 of 37 1 schedules. We began with a review of the status of all field outreach operations, and assessed the 2 impacts of possible revisions on the Census Bureau’s ability to complete those operations within 3 the compressed timeline. The six million housing units in the Update Leave Operation (which 4 provides Census invitations to housing units that do not receive regular US mail) had been 5 completed in early July, and we had received over two million self-responses and the remaining 6 housing units would be moved into the NRFU operation to be visited by enumerators for personal 7 interviewing. The Group Quarters enumeration operation which had begun on July 1st was on 8 track to be completed on schedule by September 3, 2020 and would not be negatively affected by 9 compressing the balance of the Field Schedule. The enumeration of persons staying in transitory 10 locations (Campgrounds, RV parks, marinas and hotels without a home elsewhere) was scheduled 11 to be conducted from September 3 – September 28. That operation could be conducted as planned 12 within the replan schedule timeline. 13 83. The COVID-19 pandemic had precluded the Census Bureau from sending staff to 14 conduct our Service Based Enumeration (SBE) operation. SBE is conducted at emergency and 15 transitional shelters, soup kitchens and regularly scheduled food vans and targeted non-sheltered 16 outdoor locations (TNSOL), and is designed to insure that people experiencing homelessness are 17 counted); it was originally scheduled to be conducted March 30-April 2. We had conducted an 18 extensive consultation in May and early June with a panel of 67 national service providers, federal 19 and state agencies to determine the best time frame to conduct this operation to best replicate the 20 weather, migratory behaviors and other factors affecting this population. The overwhelming 21 consensus of the stakeholders, and the input from Census experts, was that the best time to conduct 22 this operation would be mid-late September. Based on that stakeholder consultation we selected 23 September 22-24 to conduct the SBE and TNSOL operations with appointments made with service 24 providers in early September. A review of this operation indicated that we could conducted it in 25 the replan as currently scheduled without disruption. 26 84. We also reviewed NRFU, our largest and most critical operation. The Census 27 Bureau had conducted soft launches of all our major operations (during a soft launch a small 28 portion of the operation starts early to insure that all the planned and tested systems work as DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 29 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 31 of 37 1 designed under real field conditions with real respondents and actual newly hired temporary 2 employees). The NRFU Soft Launch was planned with six offices that could be safely started 3 based on COVID risk profiles (developed using CDC, HHS, State and Local health guidance), 4 availability of staff, and provisioning of Personal Protective Equipment. The original plan was to 5 begin the operation in one office from each of our six regions starting on July 16th (Cycle 1a) and 6 to follow on July 23rd (Cycle 1b - one week later) with six additional offices picked from coastal 7 areas that would be prone to Hurricane risk. As the plan developed we were unable to take offices 8 from all of the areas in the original plan because of high COVID risk and state and local stay at 9 home orders, however we were able to select 6 offices for each cycle and these offices commenced 10 NRFU field operations without incident on the planned dates. In early to mid July, as the pandemic 11 controls began to be lifted, and our concerns grew over lack of action on a waiver of the December 12 31, 2020 apportionment statutory deadline, we decided to expand NRFU operations to all offices 13 that could meet the safety, health, and staffing requirements – to start those offices in advance of 14 the initial planned start date of August 11, 2020. We deployed NRFU operations in 35 additional 15 offices on July 30, 2020 and 39 additional offices on August 6, 2020. We then made the decision 16 to pull forward all remaining offices from August 11 to August 9. All ACOs had begun NRFU 17 operations by August 9 and we had enumerated over 7.4 million housing units before the Replan 18 Schedule’s official start date of August 11. 19 85. Concurrent with the early start of NRFU operations, we observed higher levels of 20 overall staff productivity resulting from the efficiency of the Optimizer (a software program that 21 both schedules work for our enumerators and then routes them in the most effective routing). The 22 increased productivity that we observed during the soft launch period was a factor in our ability to 23 design the replanned field operations to end by September 30, 2020. The bonus plan to increase 24 hours also contributed to our ability to create a replan to meet this deadline. We presented the 25 Replan Schedule to the Secretary on August 3, he accepted it, and the Director announced it that 26 same afternoon. For clarity, I will refer to this schedule as “the Replan Schedule.” 27 28 86. The Replan Schedule intends to improve the speed of the NRFU operations without sacrificing completeness. Under the Replan Schedule, the Census Bureau has responded to the DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 30 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 32 of 37 1 shortened calendar period for NRFU operations by taking steps to increase the ability of its 2 employees in the field to work as efficiently as possible. This involves increased hours of work 3 per enumerator, spread across the total workforce, to get the same work hours as would have been 4 done under the original time frame. We incentivize this behavior by providing monetary bonuses 5 to enumerators in who maximize hours worked, and retention bonuses to those who continue on 6 staff for multiple successive weeks. Successful completion of NRFU is dependent on hours 7 worked, not days worked. 87. 8 We have aimed to improve the effectiveness of our count by continuing to maintain 9 an optimal number of active field enumerators by conducting additional training sessions, and 10 keeping phone and tablet computer devices for enumeration in use for the maximum time possible, 11 thereby decreasing the inefficiency created by training new enumerators. 88. 12 The Census Bureau was able to adopt the Replan Schedule because the design of 13 the 2020 Census allows a more efficient and accurate data collection operation in a shorter 14 timeframe than was possible in the 2010 Census. Improvements that make this possible include 15 use of our route and case optimization software, use of handheld devices, and streamlined 16 processing. Additionally, it is worth noting that largely because of the schedule delays, the self- 17 response period for the 2020 Census will be longer than the self-response period for the 2010 18 Census. 89. 19 20 21 The Replan Schedule also necessitated some changes to the content and timing of our post processing operation. These changes include: • We shortened address processing from 33 to 20 days. This required eliminating 13 days 22 of processing activities that will be deferred until the creation of the redistricting data 23 products. 24 • used to tabulate 2020 Census data (what we call “the MAF Extract”). 25 26 27 We cancelled the internal independent review of the final list of addresses that will be • We eliminated redundant quality control steps, and the multiple file deliveries that supported those steps, in order to enable a state-by-state flow of deliveries for processing. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 31 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 33 of 37 1 (Previous procedures delivered data to the next step only when the entire country had 2 been reviewed by multiple teams). 3 • We optimized employee assignments to ensure maximum staff resource usage during this shortened production period – i.e., implemented a seven-day/week production schedule. 4 5 • We compressed the time allotted for subject matter expert review and software error remediation, cutting 21 days from the schedule. 6 7 8 9 90. These changes increase the risk the Census Bureau will not identify errors during post processing in time to fix them. 91. Nevertheless, the Census Bureau is confident that it can achieve a complete and 10 accurate census and report apportionment counts by the statutory deadline following the Replan 11 Schedule. The 2020 Census operational design is tailored to enumerate all persons, including hard- 12 to-count populations. 13 92. The Census Bureau has kept the Office of Management and Budget informed about 14 schedule developments for both the COVID Schedule and the Replan Schedule, and has filed 15 nonsubstantive changes that have been published in the Federal Register. OMB was not required 16 to approve the changes to the operational plan, nor did it. As with the 2018 Operational Plan, we 17 did we not ask other agencies to review or approve either the COVID Schedule or Replan 18 Schedule. 19 20 XIII. 93. Impacts of Granting a Preliminary Injunction If the Court grants an injunction, the Census Bureau will need to replan the 21 remaining census operations again. We cannot speculate at this point exactly how we will replan 22 the remainder of the census, as the specific actions we take will depend on when the Court rules 23 and the specifics of the ordered actions. 24 94. The timing of any Court order changing the schedule is particularly important, as 25 stated in our filing on Wednesday, September 2, 2020, where we explained that the Census Bureau 26 has already taken steps to conclude field operations. As I will explain further, the fact that we are 27 concluding field operations in ACOs that have completed their workload is a normal part of the 28 NRFU operation, and is not specific to the Replan Schedule. DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 32 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 34 of 37 1 95. The Census Bureau manages its nonresponse follow up operation (NRFU) out of 2 “Census Field Supervisor areas” or “CFS areas” within each of the nation’s 248 ACOs. As of 3 September 3, 2020, roughly 11% of CFS areas nationwide are eligible for what we call “the 4 closeout phase,” over 1,220 are actually in the closeout phase, and roughly 50 have actually 5 reached conclusion. The closeout phase refers to the process of focusing our best enumerators to 6 resolve the remaining cases in that area. CFS areas are eligible for closeout procedures when they 7 cross the 85% completion mark. All CFS areas become eligible for closeout procedures on 8 September 11. This does not mean that all CFS areas will be moved to closeout procedures on 9 that date, only that regional directors can make this decision. Prior to that date no CFS area can 10 be moved into closeout procedures until it reaches 85% completion. The Census Bureau is 11 continuing to work across the nation to obtain responses from all housing units, and has not 12 begun closeout procedures for any CFS area with under 85% completion. 13 96. It is a normal and planned part of the NRFU operation for an ACO to move into the 14 closeout phase and complete operations. We used closeout procedures in NRFU in the 2010 15 Census and always planned to do the same for the 2020 Census. If we have not wound down in 16 some areas, it is because we are still counting. Some ACOs have greater initial workload, and some 17 started earlier than others –therefore, moving to completion varies by ACO and is a reflection of 18 workload and local conditions and results in the allocation of enumerator resources from areas that 19 are complete to areas that require more work. 20 97. We are currently finished with over 64% of the NRFU field work and over 85% of 21 the total enumeration of all housing units in the nation and those numbers increase daily. More 22 than 13 states have over 90% of their housing unit enumeration completed, and in 18 additional 23 states we have completed over 85% of the housing units in those states. As we complete areas, 24 staff are offered an opportunity to assist by enumerating in other areas that are not yet complete. 25 Some staff elect that option, others choose not to go outside of their home area, and as their area 26 is completed, they are released. As we complete more field work, the number of staff that are still 27 active declines, and our ability to ramp up is severely hampered. 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 33 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 35 of 37 1 98. Lack of field staff would be a barrier to reverting to the COVID Schedule were the 2 Court to rule later in September. The Census Bureau begins terminating staff as operations wind 3 down, even prior to closeout. Based on progress to date, as is standard in prior censuses, we have 4 already begun terminating some of our temporary field staff in areas that have completed their 5 work. It is difficult to bring back field staff once we have terminated their employment. Were the 6 Court to enjoin us tomorrow we would be able to keep more staff on board than were the Court to 7 enjoin us on September 29, at which point we will have terminated many more employees. 8 99. Were the Court to enjoin us, we would evaluate all of the changes we made for the 9 Replan Schedule and determine which to reverse or modify. For example, we notified participants 10 of the cancellation of the Count Review 2 operation, originally scheduled for September 15. If our 11 schedule were extended, we would evaluate whether to re-schedule this operation. We would go 12 through each and every aspect of remaining operations and determine how best to use the 13 remaining time to maximize the accuracy and completeness of the census results. 14 100. Finally, we wish to be crystal clear that if the Court were to extend the data 15 collection period past September 30, 2020, the Census Bureau would be unable to meet its statutory 16 deadlines to produce apportionment counts prior to December 31, 2020 and redistricting data prior 17 to April 1, 2021. The post processing deadlines for the Replan Schedule are tight, and extending 18 the data collection deadline would, of necessity, cause the Census Bureau to fail to be able to 19 process the response data in time to meet its statutory obligations. We have already compressed 20 the post processing schedule from 5 months to only 3 months. We previously planned and tested 21 our post processing systems assuming that we would follow a traditional, sequential processing 22 sequence, and the 3-month schedule necessary for the Replan Schedule has already increased risk. 23 We simply cannot shorten post processing beyond the already shortened 3-month period. 24 101. As I have tried to make clear in this Declaration, the decennial census is a massive, 25 complex, and interrelated endeavor. Particularly troubling is the prospect of continual, conflicting, 26 and evolving court orders from this this and other courts, including appellate courts. While Census 27 Bureau staff have demonstrated considerable resilience and flexibility during this difficult year, 28 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 34 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 36 of 37 1 some certainty as to the amount of time available to conclude data collection and post processing 2 will increase the likelihood of a successful outcome. 3 XIV. Commitment to Transparency and High Quality Enumeration 4 102. In my role as Associate Director, I remain committed to transparency about 2020 5 Census operations. The Census Bureau has been posting detailed information on its website about 6 both self-response and NRFU completion progress: 7 https://2020census.gov/en/response-rates/self-response.html 8 https://2020census.gov/en/response-rates/nrfu-completion.html 9 https://2020census.gov/en/response-rates/nrfu.html 10 103. The 2020 Census is the first to post NRFU workload information, which is now 11 available at the state and ACO level and may be seen at https://2020census.gov/en/response- 12 rates/nrfu-completion.html. I have briefed staff for House and Senate leadership every Friday 13 since April (except for August 7), and I have provided a transcribed briefing to Congress. We 14 produce a massive amount of documents and other information to the Office of the Inspector 15 General and the General Accounting Office every week, and these organizations interview Census 16 Bureau staff on almost a daily basis. 17 104. In my role as the Associate Director, I remain committed to conducting a high- 18 quality field data collection operation as explained above, and the ultimate goal of a complete and 19 accurate census. 20 21 22 23 I have read the foregoing and it is all true and correct. 24 DATED this ___ day of September, 2020 25 26 Digitally signed by Albert E Albert E Fontenot Date: 2020.09.05 00:14:42 -04'00' Fontenot ____________________________________ 27 Albert E. Fontenot, Jr. 28 Associate Director for Decennial Census Programs DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 35 Case 5:20-cv-05799-LHK Document 81-1 Filed 09/04/20 Page 37 of 37 1 United States Bureau of the Census 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 DECLARATION OF ALBERT E. FONTENOT, JR. Case No. 5:20-cv-05799-LHK 36