United States Gimeral Accounting Oface /qzg: q3 Testimony GAO For Release Decennial Census: Status of Housing Coverage on Delivery Check and Postcensus Local Review Programs Expected at lo:00 a.m. EDT Tuesday September 25, 1990 Statement of L. Nye Stevens, Director Government Business Operations Issues General Government Division Before the Subcommittee on Census and Population Committee on Post Office and Civil Service House of Representatives GAO/T-m - 90 - 63 GAOFormt60(12/87) DECENNIAL CENSUS: STATUS OF HOUSING COVERAGECHECK AND POSTCENSUSLOCAL REVIEW PROGRAMS SUMMARYOF STATEMENTBY L. NYE STEVENS DIRECTOR, GOVERNMENTBUSINESS OPERATIONS ISSUES The Census Bureau's coverage improvement programs are primarily intended to reduce the historic population undercount by improving census coverage. The Bureau noted in July that its census address list contained about 102 million housing units compared to its independent estimate at that point that 104 million units may actually exist. The Census Bureau currently is implementing two of its final coverage improvement programs. Both of these programs--the Housing Coverage Check and Postcensus Local Review--are designed to identify omissions from the 1990 address list. GAO believes that these major coverage improvement programs, while important, are not likely to identify enough additional housing units to reconcile the current address list with the independent estimate. The question whether the gap is real, that is whether the estimate was incorrect, whether the address list remains incomplete, or whether the gap is a product of both, remains open. The Housing Coverage Check was developed this summer when the Bureau's research indicated that housing units were missed during census address list development efforts. The Bureau recanvassed about 15 million housing units in targeted areas nationwide. GAO believes that the Bureau was wise to modify its plans and undertake this major recanvassing effort. Nationwide, about 313,000 housing units have been proposed for addition to the Bureau’s address list as a result of the Housing Coverage Check, which is virtually complete. In contrast to the Housing Coverage Check, the Postcensus Local Review Program has long been scheduled as a 1990 census coverage improvement program and is still in progress. However, based on its preliminary observations, GAO believes that the program's potential to significantly augment housing unit counts is limited. First, the Bureau has budgeted to recanvass blocks containing about 2 million housing units for the program, substantially fewer than for the Housing Coverage Check. Second, widespread local government participation in the program is unlikely. Mostly because they lacked the resources or independent data for analysis, less than half of the eligible governments participated in an earlier phase of the local review program. In response to a GAO survey, about 60 percent of those that did not participate indicated they also would not participate in Postcensus Local Review. Finally, challenged housing unit counts on blocks that already have been recanvassed as part of the Housing Coverage Check will not be reviewed again. Mr. Chairman and Members of the Subcommittee: We are pleased to be here today to discuss the status of two of the Census Bureau's final coverage improvement programs for the 1990 decennial census: Housing Coverage Check and Postcensus Local Review. Both of these programs represent _ initiatives by the Bureau to reduce population undercounts resulting from missed housing units. The Bureau acknowledged in July 1990 that its address list contained about 102 million housing units compared to its independent estimate that 104 million may actually exist. My comments are based on our continuing effort, as requested by the Subcommittee, to monitor census operations at Bureau headquarters and in the field. Both of these programs are designed to identify omissions from the Bureau's 1990 address list. In the Housing Coverage Check, the Bureau targeted census blocks for recanvassing where count review and other research indicated evidence of missed housing units. Under Postcensus Local Review, all local governments in the United States were offered the opportunity to review and identify discrepancies between census housing counts and local estimates of housing units at . the block level and target those blocks for possible recanvass by the Bureau. As missed units are identified Y under either program, the Bureau intends to correct its 1 address list and enumerate the residents, provided those housing units existed and were occupied on April 1, 1990. Neither of these programs, however, is designed to correct for inaccurate counts of individuals residing in identified housing units, or for misclassification of identified housing units as vacant. My comments on the Housing Coverage Check are based on the Census Bureau's management information reports and discussions with Bureau headquarters and regional officials. My comments on Postcensus Local Review are based on discussions this month with Bureau headquarters and regional officials in 7 of the 13 regions, and with local review officials in eight major cities (Atlanta, Baltimore, Chicago, Cleveland, Los Angeles, New York, Phoenix, and Pittsburgh). We also discussed local review with the former chairman of the State Data Center Steering Committee and used a survey we did in June through August of a random sample of 1,047 governmental units that did not respond to the Bureau during Precensus Local Review. HOUSING COVERAGECHECK The Housing Coverage Check was not part of the Census Bureau's plan at the outset of the decennial. According to * the Bureau, the Housing Coverage Check was initiated when 2 media reports and its analysis of such data as calls to the Bureau's telephone assistance numbers suggested that some housing units might have been missed by previous address list development operations. The requirements for the program-- issued in July 1990 --identified a number of sources for targeting areas where additional canvassing and enumeration might be effective, including (1) comparison of independent estimates with census counts, (2) correspondence and phone calls that identified whole areas or buildings that did not receive questionnaires, (3) results of postal checks, and (4) local knowledge of census staffs. AS Of September 19, 1990, (see fig. 1) the Bureau's mahagement information system reported that about 399,000 blocks had been recanvassed in 12 of its 13 regions (these data were not available on Kansas City), which represented more than 99 percent of the recanvassing workload as of that date. Given the wide-ranging sources considered in targeting blocks, the number of blocks recanvassed and the proposed housing unit adds varied widely on a region-by-region basis. For example, some 55,130 blocks were recanvassed in the Atlanta region while only about 5,500 blocks were recanvassed in the Seattle region. As a result of the nationwide program, about 313,000 housing units had been proposed for addition to the Bureau's address list in the 13 w regions. 3 . The ratio of housing units added to blocks recanvassed also varied significantly. On average, according to the Bureau's management information reports, about three-fourths of one housing unit was added for each block recanvassed nationally. In the Chicago region, it was about three-tenths of one housing unit per block recanvassed; in the New York region it was about one-for-one; and in the Boston region about 1.6 housing units were added for each block recanvassed. The scope of recanvassing also varied significantly in the eight cities we contacted, ranging from less than 2 percent of total blocks in Cleveland, to about 12 percent in Atlanta, to about 27 percent of total blocks in New York City. Information about block recanvassing was not available for Los Angeles at the time of our interview. Census officials in the seven regional offices we contacted indicated they experienced no major problem recruiting and retaining personnel for Housing Coverage Check recanvassing. It would appear, Mr. Chairman, that the Bureau was wise to modify its plans and undertake this major recanvassing effort. More than 300,000 additional housing units were identified as a result of the program, at an expenditure of approximately $13 million. The Bureau recognizes that v coverage improvement programs like Housing Coverage Check are 4 . * expensive but believes it cannot ignore the undercount problem that most of the programs were designed to address. POSTCENSUSLOCAL REVIEW In contrast to the Housing Coverage Check, the Postcensus Local Review Program has long been scheduled as a 1990 census coverage improvement program and it is still in progress. On the basis of our preliminary observations, we believe that its potential to improve housing unit and population counts will be reduced by two factors: (1) many local governments will not participate in the program, and-(21 some blocks challenged during Postcensus Local Review will have already been recanvassed during Housing Coverage Check. Since the program is still in progress, however, we are unable to provide summary data on total blocks recanvassed and housing units added. As you know, for the 1990 census, the local review program has two phases: one before Census Day and one after. The first phase, Precensus Local Review, was completed in February 1990. In the second phase, Postcensus maps and housing unit counts were mailed to some 40,000 governmental units. The maps were mailed during the period May through early August 1990: housing unit counts and group quarters pop;lations were sent out in the latter half of August. 5 * . I G o v e r n m e n tal u n i ts h a d 1 5 w o r k i n g d a y s - - c o m p a r e d to 4 5 w o r k i n g d a y s in P recensus --to review th e B u r e a u 's c o u n ts a n d submit challenges to th o s e c o u n ts a t th e block level where th e y believe th e B u r e a u is in error. T h e B u r e a u p l a n s to c o m p l e te recanvassing a n d p r o v i d e fe e d b a c k to local g o v e r n m e n ts o n th e results o f P o s tce n s u s L o c a l R e v i e w b y late O cto b e r 1 9 9 0 . A ccording to th e B u r e a u , th e p r o g r a m is currently on schedule. B u r e a u 's R e c a n v a s s i n g W o rkload W ill V a ry A cross th e N a tio n T h e p o te n tial o f th e P o s tce n s u s L o c a l R e v i e w p r o g r a m to i d e n tify m issed h o u s i n g u n i ts a n d a d d to th e B u r e a u 's p o p u l a tio n c o u n t d e p e n d s in p a r t o n th e e x te n t to w h ich local g o v e r n m e n t u n i ts p a r ticip a te in th e p r o g r a m . A s o f S e p te m b e r 1 7 , th e B u r e a u h a d received r e s p o n s e s fro m a b o u t 7 ,0 0 0 - - o r a b o u t 1 7 p e r c e n t --of th e nearly 4 0 ,0 0 0 local g o v e r n m e n ts in th e c o u n try. A b o u t 1 ,7 0 0 h a d a c c e p te d th e B u r e a u 's p o p u l a tio n e s tim a te s a n d 5 ,4 0 0 h a d file d challenges. T h e e i g h t m a jor citie s w e c o n ta c te d earlier this m o n th w e r e still preparing their challenges to . th e B u r e a u 's preliminary h o u s i n g c o u n ts. N o n e theless, th e s c o p e o f their planned challenges varies w idely, fro m less th a n 4 p e r c e n t o f city cens"us blocks in P h o e n ix to m o r e th a n 5 0 p e r c e n t in C h i c a g o . 6 . , In targeting blocks to challenge, the most common criterion being used by the cities we contacted was blocks where the Bureau's counts were five or more housing units less than local estimates. Chicago, on the other hand, planned to challenge all blocks with differences of any kind, However, many other governments, especially those in small communities, are unlikely to respond, based on participation rates in Precensus Local Review. The Bureau reports that only 16 percent of 21,000 eligible governments participated in the precensus phase of local review. Our random sample survey of 1,047 of the approximately 17,000 governments that did not respond indicates that the actual participation rate was somewhat higher. For example, 17 percent did not identify a significant number of omissions in the Bureau's count and therefore did not respond; 4 percent more did not respond because a higher level government (for example, a county government) responded for them. Overall, it appears that the actual participation.rate in Precensus Local Review was still less than half of eligible governments. About 60 percent of the governments that did not participate said in response to our survey that they also do not plan to participate in the Postcensus Local Review program. The major reasops cited were (1) a lack of funds, expertise, or other resources to carry out the program; and (2) a lack of 7 housing unit data of their own at the block level required to challenge Bureau counts. This was especially true of smaller communities with populations of less than 12,500. The former chairman of the State Data Center Steering Committee told us . that some small governments have a double disadvantage: They lack the expertise to carry out local review themselves, and they lack the funds to hire someone to do it for them. Most --about 80 percent --of the communities that do not plan to participate in Postcensus Local Review are small communities. Bureau Plans for Recanvassinq Questions have also been raised, Mr. Chairman, about the adequacy-of the Bureau's Postcensus Local Review recanvassing budget to handle the challenges submitted by local governments. In that regard, the Bureau has budgeted to recanvass 2 percent --or about 2 million--housing units during Postcensus Local Review, compared to the estimated 15 million housing units covered during Housing Coverage Check. When we discussed this issue with them, local review officials in two of the seven Bureau regional offices we contacted said the budgeted level would be adequate. However, officials in the other five regions thought that it was too early to say whether the 2 percent budgeted for recanvassing was adequate. 8 During Postcensus Local Review, all blocks for which acceptable documentation was provided are eligible for recanvass, provided that they were not already recanvassed during Precensus Local Review or Housing Coverage Check. The Bureau plans to recanvass at least one block and a maximum of 2 percent of the housing units within a governmental unit, starting with the block with the largest positive housing unit difference and continuing in descending order. As needed, budgeted funds for Postcensus recanvassing may be transferred among district offices within a Census region. We asked local review officials in the eight cities whether they thought the 15 working day time frame--mandated by the Bureau for reviewing housing unit counts and submitting challenges-- was adequate. In five cities, officials characterized the 15 working day time frame as very inadequate or generally inadequate. Officials in two cities said that 15 days was generally adequate, while officials in another city indicated that they had no basis to judge. With respect to smaller governments, the former chairman of the State Data Center Steering Committee told us that those governments were "scared" by the short turnaround for submitting Postcensus Local Review challenges. In our " opinion, it is still too early to determine if the recanvassing budget of 2 percent of the total housing units 9 will be adequate. Several factors, however, will serve to reduce the level of recanvassing needed: (1) many eligible governments will lack the time, data, or resources to respond to the Bureau; (2) as occurred in the Precensus Local Review, some governments will agree with the Bureau's counts and therefore require no recanvassing; and (3) many of the blocks challenged by local governments will already have been recanvassed under the Housing Coverage Check Program. For these same reasons, Mr. Chairman, we agree with your statement in announcing this hearing that the Bureau's remaining operational programs-- including Postcensus Local Review-- are unlikely to eliminate the gap between the housing units and the individuals counted so far by the Bureau and its independent estimates. This observation does not resolve the question whether the gap is real--that is, whether the estimate was too high, whether there are still substantial numbers of housing units that recanvassing does not find, or whether the gap is a combination of both. Improved Feedback for Postcensus Local Review Since some local governments were dissatisfied with the feedback they received after Precensus Local Review, we examined the Bureau's plans for feedback after the Postcensus phase. To the Bureau's credit, it has decided to provide much'more detailed feedback on the results of Postcensus as 10 . - compared to Precensus Local Review. On a block-by-block basis, for all blocks challenged during the Postcensus phase, the Bureau will provide local governments with the net change in housing unit counts populations. Feedback to governmental units on the results of Precensus Local Review was very limited: Governments were simply advised that their challenges had been accepted for review and informed of the total number of blocks that were recanvassed. No information waq provided on the results of recanvassing. The former chairman of the State Data Center Steering Committee told us that this limited Precensus feedback may discourage governmental units from participating in Postcensus Local Review. While planned Postcensus feedback represents a significant departure from feedback provided during Precensus, there are, nonetheless, limitations. For example, the Bureau does not plan to provide data on net changes on blocks that were not challenged. Further, the source of the net change for challenged blocks --Housing Coverage Check, Postcensus Local Review, or other Census operations--will not be provided. Finally,' the Bureau does not plan to furnish governmental units with information on the net change to the total popu\ation estimate given to them along with Postcensus housing unit counts. 11 . The promise of improved feedback could encourage more communities to participate in POStCenSuS Local Review. However, not all.of the eight major cities we contacted were aware of the Bureau's plans. Two of those cities said they expected to receive the same type of feedback they received after Precensus Local Review, two were not sure of what feedback they would receive after Postcensus, and an official in one city indicated that he did not expect to receive any feedback. The other three cities expected block-by-block feedback as planned by the Bureau. Postcensus Maps Did Not Pose Major Problems The Bureau appears to have improved the accuracy and usefulness of the maps it provides to local governments as part of the local review program. During Precensus Local Review, local governments expressed widespread dissatisfaction with the quality of the census maps. However, for Postcensus Local Review, five of the six major cities we COntacted that submitted map corrections to the Bureau during Precensus Local Review indicated that all/almost all, or most of the corrections they had identified during precensus were reflected in the Postcensus Local Review maps that the Bureau sent to them. Officials in New York City, for example, said that most of the corrections they identified had been made, and that the New York Regional Census Center usually had an acceptable explanation in cases where it did not make suggested changes. Los Angeles was not sure of the extent 12 to which its suggested changes were made. Two cities we contacted had not submitted any map corrections to the Bureau during Precensus Local Review. Bureau regional officials informed us that local governments did not express widespread dissatisfaction with their Postcensus maps. We asked officials in 7 of the Bureau's 13 regional census centers to estimate how many governmental units in their reg'ions reported problems with Postcensus maps. Five of the seven regions estimated that some or few governments had reported problems, while two were not sure of the magnitude of problems reported. For example, the Denver region was one of the regions reporting that some governmental units had reported map problems. The Denver official estimated that between 150 and 200 governmental units had called in with problems: however, there are more than 5,200 governmental units in this region. This concludes my prepared statement, Mr. Chairman. My colleagues and I would be pleased to respond to questions. 13 50 45 40 55 so 25 20 15 10 5 0
Decennial Census: Status of Housing Coverage Check and Postcensus Local Review Programs
Published by the Government Accountability Office on 1990-09-25.
Below is a raw (and likely hideous) rendition of the original report. (PDF)