Response to Comments Received on Address Canvassing

Responses to Comments Received on the 2020 Census 60-day Federal Register Notice

The 2020 Census 60-Day Federal Register Notice was posted for public comment and published in the Federal Register June 8, 2018. This document provides a response to comments received other than comments about the inclusion of the citizenship question on the 2020 Census questionnaire.

Topic of Comments: Address Canvassing

The Census Bureau received four sets of comments related to the 2020 Census Address Canvassing operation. The authors of these comments, organizations the authors represent, and the specific comments related to Address Canvassing are provided below. The comments focused on ensuring that efforts to reduce the overall costs of conducting the 2020 Census do not compromise the quality of the data, particularly as it relates to undercounts of hard-to-count populations. A general response to all four of these sets of comments follows the listing of the comments.

Department of Finance, State of California, Irena Asmundson, Chief Economist

California is committed to ensuring full public participation in the 2020 Census given that a complete and accurate count is critical for fair political representation and the distribution of billions of dollars in federal funds. To that end, we strongly recommend the U.S. Census Bureau withdraw the citizenship question, increase field canvassing efforts, expand language options for the Census form and broaden efforts to reach residents who do not initially respond.

In order for the Census Bureau to achieve its goal of counting everyone “once, only once, and in the right place,” it needs a complete address list for all residents. In 2010, Census Bureau field staff canvassed nearly every block in the nation. This comprehensive approach for validating the address list has been abandoned for the 2020 Census and replaced with a new, in-office validation strategy using satellite imagery and third-party data. This type of off-site canvassing has the grave potential to miss unconventional and secondary housing units that share a roofline, such as garages and basement conversions, and do not appear on satellite imagery. People residing in these housing units may not be counted, which is unacceptable. While the Census has agreed to do in-field canvassing in areas where addresses cannot be verified (approximately 30 percent of total addresses), this is inadequate to ensure that every person has an opportunity to participate.

Asian Americans Advancing Justice, John C. Yang, President and Executive Director

As stated in the federal register notice, “[t]he goal of the 2020 Census is to count everyone once, only once, and in the right place.” To achieve that goal, the Census Bureau must prioritize the quality of the data it collects above all else, including cost savings. By definition, counting those that are harder to count will cost more. And today’s anti-immigrant, hostile political climate means there will be more that are hard to count, and they will be harder to count than ever. Thus, the 2020 Census effort must be focused on counting those hardest to count. These comments will address how best to ensure a fair and accurate census that counts everyone by providing concerns and recommendations on the content and form design as well as design changes in four key areas: reengineered address canvassing, optimizing self-response, utilizing administrative records and third-party data, and reengineered field operations. In particular, the comments will focus on (i) whether the proposed collection of information is necessary for the proper performance of the function of the agency, including whether the information shall have practical utility; (ii) ways to enhance the quality, utility, and clarity of the information to be collected; and (iii) ways to minimize the burden of the collection of information on respondents, including through the use of automated collection techniques or other forms of information technology.

The Census Bureau’s reliance on technology in their reengineered address canvassing operations could negatively impact a fair and accurate count of Asian Americans through the use of technology to replace on-the-ground, in-person address canvassing for 70% of the addresses in its database. In-office address canvassing, which relies on tools such as administrative records and satellite imagery, is less able to detect nontraditional, complex households than people in the field. The reality is that the traditional concept of a household – one that only includes a married couple with children under 18 – has been giving way to nontraditional, more complex households over the last several decades. While three-fourths of all U.S. households in 1960 consisted of married couples with or without children, in 2000, just under 53 percent of all households consisted of married couples with or without children. The increase in nontraditional, complex households stems from “demographic trends such as: increases in immigration rates and the proportion of the population that is foreign born[…], and changing migration streams now coming predominantly from Asia and Latin America, rather than from Europe. Other factors include increases in cohabitation and blended families due to more divorces and remarriages; increases in the proportions of cohabitor households with children; and dramatic increases in grandparent-maintained households and nonrelative households.” In fact, multigenerational households – those that include two or more adult generations, or those that include grandparents and grandchildren – have been increasing, with a record 60.6 million people (or 19% of the U.S. population) living with multiple generations under one roof in 2014. Asian Americans are more likely to live in non-traditional, complex households. Sixty-seven percent of Asian Americans are immigrants, and those that are foreign-born are more likely to live with multiple generations of family. In 2014, 28% of Asian Americans lived in multigenerational family households, among the highest of any group. Additionally, we’ve seen in recent years that young adults are the age group most likely to live in multigenerational households. In fact, for the first time in more than 130 years, young adults 18 to 34 were more likely to be living with parents than any other living arrangements in 2014. Over 4.6 million Asian Americans are 18 to 34 years old, representing just over one in four Asian Americans. Asian Americans are often more likely to live in crowded conditions. For example, in New York City in 2009, Asian Americans had larger households than average (3.12 people v. 2.67) and were more likely to live in a household with more than one occupant (14% v. 8%). Asian Americans can also find themselves living in crowded housing with many unrelated individuals in an effort find affordable housing. All these households are complex and can be more difficult to properly count in a decennial census.

If a household is not in the master address list, then they will mostly likely NOT be counted during the 2020 census. Because In-Field Address Canvassing will be particularly important for identifying and noting nontraditional, complex households, we recommend that the accuracy of the list should be the top priority and that the Bureau should increase its In-Field Address Canvassing workload regardless of the cost factor. As part of this analysis, the Census Bureau must determine the accuracy of those 70% of the nation’s addresses that are deemed “stable” and resolved by In-Office Canvassing. Because of the invisible nature of complex households, such as multiple families living in one dwelling, we have concerns that these types of households would be missed during In-Office Canvassing. And to the extent that these types of households often represent those traditionally hardest-to-count, it is imperative that the In-Office Canvassing does not exacerbate the likelihood of missing them.

Leadership Conference on Civil and Human Rights - Vanita Gupta, President and CEO, et al

• The Census Bureau should devote more resources to conducting In-Field Address Canvassing and expand the scope of on-the-ground canvassing to increase the likelihood of capturing nontraditional, hidden, and converted housing units in the MAF before the census starts.

• We previously offered general support for the use of administrative data to assist with verifying and updating the residential address file. At the same time, we urge caution with respect to the limits of these datasets, which are far less likely to include accurate information on nontraditional and hidden housing units that, more often than not, are home to individuals and families already considered hard-to-count based on a range of demographic characteristics identified in previous censuses and 2020 Census pretests. Over reliance on government and commercial data to build an accurate address list undoubtedly will result in failure to count a disproportionate number of households in historically undercounted communities.

New York City Department of City Planning Population Division, Joseph Salvo, Director

The Notice states that in preparation for the 2020 Census, “the Census Bureau is using In-Office Address Canvassing for the first time,” which “detects and identifies change using high-quality imagery, administrative data, and third-party sources to review and update the address list.” However, the process of identifying discrepant addresses in the absence of field inspections is potentially flawed in places like New York City, where DCP’s Local Update of Census Addresses (LUCA) research has shown that housing units, particularly basement and subdivided units in small multi-unit buildings, are undetectable using high-quality imagery, even in conjunction with administrative data. In these cases, field work needs to be conducted to verify the presence of additional units, including those provided by local governments through LUCA. The flagging of housing units during the In-Office Address Canvassing process for scrutiny in the actual enumeration should not be viewed as a substitute for in-field canvassing prior to the census.

Census Bureau Response:

The Census Bureau has implemented a number of mechanisms to ensure the quality of its Address Canvasing operations. The In-Office Address Canvassing process began in September of 2015 with the Interactive Review, which reviews all housing units across the nation using multiple sources of data such as aerial imagery, administrative data, and third-party data. The administrative and third-party data sets are used in conjunction with the imagery to give the Interactive Review clerk information about what is under each roof top. Interactive Review utilizes imagery in conjunction with data in the Master Address File (MAF) to determine whether individual census blocks are “passive” or “active.” Blocks are determined to be passive when the number of housing units visible in the current imagery matches the number of addresses in the MAF, and to be active when the comparison of imagery to the MAF detects undercoverage or overcoverage in the MAF. Census blocks for which the Interactive Review clerk could not make a determination, generally due to cloud cover in imagery, are placed “on-hold.”

The Census Bureau began In-Office Address Canvassing in September 2015 and completed the Interactive Review initial review of all 11.1 million census blocks in June 2017. The Census Bureau’s triggering process, which began in June 2017 after completion of the initial review, identifies census blocks with changes in the inventory of addresses due to processing of the United States Postal Service Delivery Sequence File (DSF) or other address sources, census blocks that may have potential classification changes, and census blocks that were put on hold with the hope of assigning either an active or passive status. The process then sends those census blocks back into the Interactive Review component of In-Office Address Canvassing for a new review, which can result in the following:

• A passive block remains passive (i.e., the MAF has kept up with change on the ground).

• An active block becomes passive (i.e., updates to the MAF have resolved the coverage issue detected in the previous IOAC review).

• A passive block becomes active (i.e., updates to the MAF did not account for all of the change detected in imagery).

• Clearer imagery allows for determination of passive or active status for an on-hold block.

Census blocks may also be triggered directly into In-Field Address Canvassing when there are known address issues within the block.

Another component of the In-Office Address Canvassing is Ungeocoded Resolution. This project identifies the census block to which an address should be assigned when the automated geocoding process for assigning block numbers from the MAF/TIGER System is unable to make a determination. Since the project began in the spring of 2017 at the Census Bureau’s National Processing Center, over 770,000 addresses (73 percent of addresses reviewed to date) have been geocoded to census blocks by clerical staff who use reference data and resources for adding missing features, feature names, and/or addresses to the MAF/TIGER database. Only geocoded addresses are included in the census address frame since it is imperative to know in which block every housing unit is located.

This In-Office Address Canvassing process allows for a continuous update of MAF data so that the Census Bureau can target resources to identify hard-to-list and hidden housing units before conducting In-Field Address Canvassing. In-Field Address Canvassing can then target known gaps in the MAF, which the Census Bureau expects will minimize undercounting. The 2020 Census recruiting and hiring strategies will ensure that field listers have the knowledge and skills (including language skills) necessary to work effectively in areas targeted for In-Field Address Canvassing. Also, In-Field Address Canvassing training instructs listers to identify and inquire about hidden housing units.

Although the Address Canvassing operation is a key address list development activity leading into the 2020 Census enumeration, it does not work alone to assure a complete and accurate list. The 2020 Census address list development started soon after the Census Bureau finalized the 2010 Census count of living quarters. The 2010 Census addresses are retained in the MAF, and they are subject to continuous updates from the DSF, as well as new sources, to support the 2020 Census. In contrast to the previous decade, the 2020 Census did not wait until just prior to enumeration to supplement the address list nationwide. For example, an initiative for the Geographic Support System (GSS) program—which is an integrated program of improved address coverage, continual spatial feature updates, and enhanced quality assessment and measurement—started work to improve the address frame in Fiscal Year 2011. The GSS and the DSF, in conjunction with the Address Canvassing operation, continue to update the frame prior to the 2020 Census.

The Census Bureau developed the Local Update of Census Addresses (LUCA) operation to meet the requirements of the Census Address List Improvement Act of 1994, Public Law 103-430. The Census Bureau uses LUCA to help develop the housing unit and group quarters (e.g., college dormitory, nursing home, correctional facility, etc.) address information that it will need to conduct the 2020 Census, similar to the LUCA operations that were conducted for Census 2000 and the 2010 Census. LUCA is a voluntary operation. Participating governments, after signing a confidentiality agreement to maintain the confidentiality of the Census Bureau address information, may review the Census Bureau's Title 13 United States Code confidential list of individual living quarters addresses and provide to the Census Bureau address additions, corrections, and deletions, as well as structure point coordinates and road updates. Participating governments also may provide spatial and attribute updates for addresses and roads. LUCA is available to tribal, state, and local governments, the District of Columbia, and Puerto Rico (or their designated representatives). LUCA includes federally recognized American Indian tribes with reservations and/or off-reservation trust lands, states, and general-purpose local governments, such as cities and townships, for which the Census Bureau reports data. The Census Bureau does encourage the LUCA participants to conduct a targeted review of areas that contain changes that have occurred and hidden or hard-to-find housing units, in the event the participants cannot conduct a full review of the address list. The LUCA updates are incorporated into the MAF in time to determine the In-Field Address Canvassing workload.

In addition to collecting address information though LUCA, the Census Bureau also conducts the New Construction Program. The purpose of the New Construction Program is to obtain city-style addresses for newly built housing units, group quarters, and transitory locations for which construction is in progress during or after March 1, 2018, that are expected to be closed to the elements (final roof, windows, and doors) and potentially inhabitable by Census Day, April 1, 2020, in blocks where census questionnaires or mailing packages are delivered and households are expected to use a self-response mode to complete the census. The Census Bureau conducts the New Construction Program to assure the completeness and accuracy of the Census Bureau's address list. Participating governments have the opportunity to provide input to improve the Census Bureau's address list and to ensure accurate and complete enumeration of their communities.

The Census Bureau also conducts the Count Review operation, which is conducted in collaboration with the state demographer members of the Federal-State Cooperative for Population Estimates (FSCPE). This program allows an additional opportunity for stakeholders to review counts of housing units prior to the 2020 Census and to provide updates as necessary.

In addition, some areas of the country are designated for the Update Leave and Update Enumerate operations. The Update Leave operation is designed to occur in areas where the majority of housing units do not have either mail delivered to the physical location of the housing unit, or the mail delivery information for the housing unit cannot be verified. The Update Enumerate operation is designated to occur in areas where the initial visit requires enumerating while updating the address frame. The majority of this operation will occur in remote geographic areas that have unique challenges associated with accessibility. In both of these operations, the addresses are visited in the field, and the address list is updated at the time of the census.

The Census Bureau is using all of these programs together to ensure the quality of the address frame for the 2020 Census.

Commenters also raised concerns about the ability for the Census Bureau’s reengineered address canvassing process to locate and assure that hidden housing units and complex households are included on the address frame for the 2020 Census. As part of the GSS Program, the Census Bureau conducted a pilot project focused on assessing the degree to which hidden housing units were missing from the MAF, as well as field methods for identifying locations of hidden units. This study was conducted in a selection of census blocks in an urban setting (Queens, New York) and a rural setting (Southern California). In-field canvassers verified 70 percent of the nearly 9,000 housing units in the Queens study site, adding only 71 (0.8 percent) and deleting 13.4 percent. Of the 9,000 housing units in the Queens site, 87.8 percent were on the address list as a result of updates from address sources, including the DSF and New York City’s Local Update of Census Addresses (LUCA) submissions for Census 2000 and the 2010 Census. Only 9.8 percent of the addresses in the study site had been acquired through previous in-field canvassing operations. It is important to note that canvassers in this study used the same job aid as was used during address canvassing for the 2010 Census to help identify potential hidden units. In other words, canvassers in the Queens study site were successful in validating addresses included on the address list but found very few additional hidden units. In-field canvassers in the rural site added a larger percentage of new units (6.7 percent) but deleted 31.5 percent of all units on their address list and verified only 28 percent of all units.

Additional research utilizing a statistically reliable sample of addresses and using 2010 Census household-level responses indicated that in-field canvassers were as likely to delete hidden housing units that had been acquired through processing of address sources as they were to add hidden units. Based on this research, it is estimated that 2010 Census operations were responsible for adding 214,370 good hidden units that would otherwise have been omitted from the enumeration frame (with 432,038 population). But it also estimated that census operations were responsible for deleting 217,558 good hidden units that otherwise would have survived to the enumeration frame (note that the population in these units is unknown since they were not enumerated). It is also worth noting that census operations added 163,241 hidden units that residents themselves did not consider to be separate units (such as basement or attic apartments, garage apartments, and other external units located on the same lot) and, as a result, returned duplicate household rosters on each of the census questionnaires delivered to the units. Based on the results from these two studies, while fieldwork can be valuable for locating hidden units, given the nature of hidden housing, it also is likely that fieldwork will result in the deletion of hidden units when canvassers cannot locate them on the ground. Thus, fieldwork to locate additional hidden units imperils hidden units that have been added to the address frame from sources provided by tribal, state, and local government partners. During the 2020 Census, the Census Bureau will work closely with local officials and community advocates, through the Integrated Partnership and Communications operation, to advertise so that households that have not received a questionnaire are aware of how to request a questionnaire, respond on-line, or by telephone through a call center. The Census Bureau advocates that this method will prove more effective for assuring the enumeration of individuals and households living in hidden units.

2