United States General Accounting OflIce /L// 7c-l) Testimony IIIIllII 141405 For Release Progress of ,the 1990 Decennial Census: on Delivery Some Causes for Concern Expected at 1O:OO a.m. EDT Monday May 21, 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 Form 160 (13/W) I. ’ II PROGRESSOF THE 1990 DECENNIAL CENSUS: SOME CAUSES FOR CONCERN SUMMARYSTATEMENT OF L. NYE STEVENS DIRECTOR, GOVERNMENTBUSINESS OPERATIONS ISSUES At the request of the House Subcommittee on Census and Population, GAO has been monitoring the status of important census activities that will significantly affect the quality of the census --follow-up activities and the local review program. As a result of the lower-than-expected mail response rate, the management challenges for completing a high quality follow-up effort have been appreciably heightened. The Bureau may need to hire over 37,000 additional employees, and census costs are projected to increase by over $70 m illion to address the unanticipated follow-up workload. However, the Bureau's performance on two key factors that will affect the timely completion of follow-up operations --achieving full enumerator staffing and meeting productivity goals--are causes for concern. Three of the Bureau's regions--Atlanta, San Francisco, and Dallas --reported that over 70 percent of their district offices were understaffed as of May 9, 1990. For New York State about 40 percent of the Bureau's district offices reported they were understaffed. However, the Bureau's staffing statistics may understate the severity of staffing difficulties because they do not distinguish between full- and part-time employees. For example, in district offices in the New York Region, part-time enumerators ranged from less than 10 percent to 60 percent of the reported work force. W h ile achieving full staffing is important, the single most important indicator of follow-up progress is how many cases are actually completed. The Bureau's management information system was designed to provide current cost and progress information needed to monitor census activities. However, thus far this system has not been able to provide complete data, and census managers have been impeded in their monitoring activities. In addition to GAO's review of the progress of follow-up operations, GAO examined how the Bureau handled New York City's reports of alleged m issed housing units as part of the local review program. For the processes that GAO reviewed, it found that the Bureau's 22 district offices in New York City generally followed planned procedures for handling reports of m issed housing units and proposed that more than 80,000 housing units be added to the City's census blocks. Some of the proposed additions, however, may have already been on the Bureau's address list from other census operations or in other census blocks. Mr. Chairman and Members of the Subcommittee: We welcome this opportunity to discuss the status of the 1990 decennial census, with a particular emphasis on New York. My comments today focus on two important census operations. First, I will discuss the status of the Bureau's nonresponse follow-up operation, which seeks, to obtain questionnaires from households that did not initially respond to the census. Second, I will discuss the preliminary results from our work on the first phase of the local review program in New York City. As you know, this program provides local governments with the opportunity to improve the accuracy and completeness of census counts. My comments on the status of follow-up operations are based on the Census Bureau's management information reports and discussions with Bureau headquarters and regional officials. My comments on the local review program are based on our preliminary work at the Bureau's 22 district offices that are responsible for New York City and discussions with City planning officials and Bureau regional officials. Our local review work did not include observations of the completeness of field reviews done by census enumerators. * I . l NONRESPONSE FOLLOW-UP PROGRESS UNCERTAIN The Bureau had planned that nonresponse follow-up would be the census' most costly and labor intensive field activity. Now that the Bureau has achieved only a 63 percent questionnaire response rate, rather than the budgeted 70 percent rate, the management challenges to completing a high quality follow-up effort have been compounded markedly. The Bureau may need to hire over 37,000 additional employees, and census costs are projected to increase by over $70 million to address the unanticipated follow- up workload. At this point, two key factors affecting the timely completion of follow-up efforts, enumerator staffing and productivity (the number of questionnaires enumerators are completing), are a source of significant concern. Many District Offices Understaffed The Bureau reported on May 9, 1990, that nationally it had about 106 percent of its needed enumerator staff on board. The Bureau's staff needs are based on full-time positions. However, the Bureau's staffing reports reflect the number of enumerators employed and do not account for part-time employees which may represent a sizable segment of the enumerator work force. For example, the New York Region Area Managers we spoke with estimated that part-time enumerators in the district offices they 2 oversee ranged from less than 10 percent to about 60 percent of total staffing. The Bureau's staff shortages become apparent when viewed from the perspective of the district offices, where the nonresponse work actually takes place. For example, figure 1 shows the percentage of district offices in,each region that reported they did not have the necessary number of enumerators. Twenty percent of the district offices in 9 of the Bureau's 13 regions were understaffed. About 40 percent of the offices that were understaffed had less than 80 percent of their needed enumerators. In addition, since the Bureau's staffing statistics do not account for part-time employment, its data overstates the actual staffing situation. Flgun 1: Extant Dhtrkt Offlcm Offl#r Hrvc, Not Mat Mot Enumwator t%Um.mW Staff Nods N..dS loo 100 tbauqpotdlatM~ndf’ dlyr(dlrd lbauqpotdlatM~nathllyr(dlrd 90 c 70 00 RoQbMl colly conlus SourCe:CensusBureaunonresponse follow-upstdtingreportMay 9,1990. 3 Three of the Bureau's regions--Atlanta, Dallas, and San Francisco--reported that over 70 percent of their district offices were understaffed. According to the Bureau, limited applicant pools and, in the case of the Dallas Region, high employee turnover rates are major factors contributing to the low staffing levels. Unfortunately, in many cases, the district offices' applicant pools may not be sufficient to overcome their enumerator shortages. Nationally, the Bureau has met 61 percent of its 1990 recruiting goal. The Bureau's lack of recruiting success is even greater in those offices that are experiencing the most significant staff shortages --those at less than 80 percent of their needed staff levels. These offices have achieved about 57 percent of their recruiting goals. In other words, the Bureau must recruit additional applicants in precisely those areas where it already has experienced relatively greater recruiting difficulties. Let me now turn to the staffing situation in the State of New York, which is of particular interest to the Subcommittee today. Figures 2 and 3 show enumerator staffing in all of New York and the Bureau's New York Region, which covers only New York City and surrounding counties. Statewide, about 60.5 percent of the district offices reported they were fully staffed or better. On 4 the other hand, as figure 3 shows, half of the district offices in the Bureau's New York Region reported they were below their needed enumerator staff levels. About 21.4 percent of the New York Region's offices reported they have less than 80 percent of the needed number of enumerators.' The Middletown and Lindenhurst offices reported that they had less than half the number of enumerators they need. Flgum 2: En~mcrtor Stdfhg In Now York Stat8 Percent of Offiies with Between 50 and Percent of Offices with Full Enumerator Staffing or Better Percerll of Offlo@ with Between 80 and 99 Percent Needed Enumerator Staff New York Stats has 38 distrbt offices. Source: Census Sureau nonresponsefollow-up staffing report, May 0,199O. 5 . . . . Flguro 3: Enumorrtor Staffing In the Now York Roglonal Cormus Cmtor Percent of 0ffiis with Full Enumerator Staffing or Better Percent of Offices with Between 80 and 99 Percent Needed Enumerator Staff The New Yotk regbnal csnsus center has 28 districtoffii. !3ouroe: Cenws Sumau nonresponse folbw-up staffing report, May 9,199o. Enumerator staffing in New York provides an excellent example of how a large part-time work force can mask potentially serious staff shortages. For example, the South Brooklyn district office reported that it had hired 106 percent of the number of enumerators it needed. However, according to the Area Manager who monitors work in that district office, about 15 to 20 percent of the office's enumerators work part-time. The office therefore may be facing a staff shortage rather than a surplus. The situation in the West Manhattan district office is more severe. The office reports that it has only 52 percent of the necessary enumerator work force. However, the Area Manager who oversees work in that office estimates that about half of the enumerator employees are working part-time. As a result, the office seems to be facing a more significant staff shortage than the Bureau staffing reports indicate. Bureau officials in the New York Region said that a variety of factors have hampered their ability to meet staff needs, including applicants' unwillingness to have contact with a resistant public. They said that the region is continuing its recruiting efforts and, in some cases, moving enumerators from other locations into areas that are experiencing severe staff shortages. Complete Data on Nonresponse Costs and Progress Not Available While hiring a sufficient number of qualified employees for the job clearly is important, the single most important indicator of nonresponse progress is how much of the work load is being completed. The Bureau's automated management information system (MIS) I a new development for the 1990 census, is intended to provide the Bureau with current cost and progress information so it can monitor critical census activities, including census costl3, on a daily or weekly basis. 7 However, thus far during nonresponse follow-up, the Bureau's MIS has not been able to provide complete cost and progress information. As a result, census managers at headquarters and in the field have not had all the data they need to monitor the status of nonresponse follow-up field activities. Bureau officials in New York said that the absence of timely information on the status of field.activities was hampering their ability to manage the census. A number of factors appear to be causing problems with the MIS. For example, not all district offices have reported. About 20 percent of the district offices had not transmitted any cost and progress data to Bureau headquarters as of May 9, the point at which the Bureau had planned to complete one-third of its nonresponse follow-up work in urban offices. The Bureau expects the number of reporting offices to increase as initial backlogs in keying in personnel appointment documents and payroll records diminish. However, even full reporting will not address all of the problems with the MIS. For example, the Bureau's software has experienced problems with formatting data already in the system. Bureau officials are working to address these problems. Accurate and complete cost and progress information is important in identifying areas experiencing problems completing field work on schedule. In a recent report we discussed the central role that expeditious completion of field work has in ensuring a high 8 quality census.1 In 1980, delays in completing fieldwork, combined with the need to limit census costs, caused the Bureau to alter follow-up procedures. These changes may have adversely affected the quality of the census' nonpopulation data. As the Subcommittee is well aware, the lower-than-expected mail response rate has left,'the Bureau with a severe funding shortfall. In addition to the funds needed to address the low mail response rate, we understand that the Bureau's request for supplemental appropriations includes $60 million for an anticipated reduction in productivity. A fully operational MIS is critical for the Bureau to closely monitor the progress of the census, especially in key areas having significant cost and schedule implications, such as enumerator productivity. LOCAL REVIEW PROGRAMIN NEW YORK CITY IDENTIFIES THOUSANDS OF ADDITIONAL HOUSING UNITS Now I would like to discuss another important census activity, the local review program, which provides local governments with the opportunity to improve the accuracy and completeness of the decennial census. To carry out this program, the Bureau provides local officials with housing unit counts at the census block level and asks them to identify or "challenge" blocks where the Bureau's housing counts may be incorrect. The Bureau does field l 1390 Census: Enhanced Oversight Should Strengthen Recruitment Program (GAO/GGD-90-65, Apr. 1990.) 9 reviews of those census blocks where loc'al officials report the most significant errors and adds any missed housing units to its address list or, in some cases, moves housing units to their correct census blocks. For the 1990 census, the local review program has two phases: one before Census Day.and one after. For the first phase, precensus local review, housing unit counts were scheduled to be sent to local governments by November 3, 1989. Local governments were given until January 5, 1990, to submit block challenges to the Bureau. As part of our ongoing examination of the precensus local review program, we reviewed how the Bureau handled New York City's challenges for blocks with alleged missed housing units. For the processes that we reviewed, we found that the Bureau's district offices generally followed planned precensus local review procedures and, in so doing, proposed that thousands of housing units be added to the Bureau's address list. While the scope of our work does not permit us to say how many total housing units should have been added as a result of the City's challenges, our review identified two problem areas that I will discuss later. Overall, the City claimed the Bureau had missed a net of 94,000 housing units. This net consisted of blocks with an alleged 175,000 missed housing units and blocks with alleged overcounts 10 . i of 81,000 housing units. The Bureau's district offices focused their field work in those blocks containing alleged missed housing units. As a result of field work, district offices proposed that about 80,000 housing units be added to challenged blocks. More than half of these proposed additions occurred in Manhattan. In addition, district offices proposed adding about 2,500 housing units to other census blocks. However, all of the proposed additions may not necessarily represent actual additions to the Bureau's address list, and as a result, we could not determine the actual additions resulting from the precensus local review program. Some of these proposed additions may have been on the Bureau's address list from other census operations or in other census blocks. For example, in one district office at least 10 percent of the proposed precensus local review additions were already on the Bureau's address list. Several reasons may account for the difference between the proposed additions and the City's reports of missed housing units. For example, the City's records may have been inaccurate; enumerators may have missed housing units; and some of these housing units may already be included in the Bureau's address list in other census blocks. In addition to reviewing reports of missed housing units, the Bureau's New York Region reviewed a portion of the census blocks 11 where the City claimed the Bureau had overcounted housing units. This analysis was done for blocks containing differences of 500 or more housing units. This operation succeeded in moving about 7,000 housing units to correct census blocks and deleting an additional 7,000 housing units from the Bureau's address list. Although the program resulted in corrections to the Bureau's address list, as I mentioned earlier, we did identify two problem areas. First, New York City did not challenge all of the blocks for which it had identified errors in housing unit counts. The City claimed that the Bureau's counts were incorrect in about 46 percent of the City's blocks. However, the City limited its challenges to the 5.5 percent of the City's blocks that contained differences of at least 50 housing units. This was well above the minimum threshold that the Bureau used as its criteria for doing field work --any block with at least five missed housing units. According to city officials, the City limited its challenges to blocks containing errors of at least 50 housing units for several reasons. First, the City received the Bureau's housing unit counts on November 14, 1989--about 2 weeks late. (However, the Bureau accepted challenges 3 weeks beyond the January 5, 1990, deadline.) Second, given the magnitude of the differences and the City's limited time and staff resources, the City focused on those areas that contained the most significant differences. 12 New York City, as well as 39,000 other local governments, will have the opportunity to challenge housing unit counts in the second phase of the local review program scheduled to begin in August 1990. The City is preparing for the postcensus local review, according to city officials. A second problem area involved a limited number of district office procedural oversights that we identified as a result of our work. If not corrected, these errors would have resulted in missed housing units. First, in five district offices field work was not done in 38 census blocks. After we told Bureau officials about this oversight, field work was done in 28 of those blocks. As a result, more than 2,500 additional housing units were proposed to be added. Field work was not done in the other 10 blocks because district office officials believed it was too late to do so. Second, in 12 district offices the paperwork required to propose adding more than 2,200 housing units was not prepared. In most instances, district officials could not explain why this occurred but said they would prepare the necessary paperwork to add these units to the Bureau's address list. That concludes my prepared statement, Mr. Chairman. My colleagues and I would be pleased to respond to questions. 13
Progress of the 1990 Decennial Census: Some Causes for Concern
Published by the Government Accountability Office on 1990-05-21.
Below is a raw (and likely hideous) rendition of the original report. (PDF)