PREPARED STATEMENT OF

KENNETH PREWITT

DIRECTOR, U.S. BUREAU OF THE CENSUS

Before the Subcommittee on the Census

Committee on Government Reform

U.S. House of Representatives

September 29, 1999

Mr. Chairman, Mrs. Maloney, and members of the Committee:

It is a pleasure to be here today to talk about issues related to the Local Update of Census Addresses program known as LUCA. I am accompanied by John Thompson, the Associate Director for Decennial Census, and Jay Waite, the Assistant Director for Decennial Census. LUCA is one of the major operations in compiling the address list for Census 2000 and one of the major partnership efforts between the Census Bureau and local and tribal governments. It is important to note that LUCA is part of a larger, multi-part process of interlocking and often overlapping programs designed to obtain knowledge from many different sources that will enable us to have the most complete address list possible for Census 2000. An address can be added to our list from any one of these sources. So, in essence, to put LUCA in its proper context, I must talk about this entire process.

The importance of having a complete and accurate address list cannot be overstated. We will mail or deliver questionnaires to every housing unit on the list, check in returned questionnaires, and follow up on those valid housing units for which a questionnaire is not returned. Without such a control list, we cannot have an accurate census. That is why we have expended so much effort over several years to reach where we are today.

We have called Fiscal Year 1999 the Year of the Address List, and for good reason. While efforts to compile the address list for Census 2000 began in the early part of this decade and won't be completed until next year, FY 1999 saw the completion of key steps in the process of building an address list. In August, we reached one of the very essential milestones of Census 2000. We completed preparation of addresses that we will include in the questionnaire address label tape, which we will soon deliver to questionnaire printing contractors. This file now contains over 119 million addresses, not including about 1.5 million now being prepared for Puerto Rico. Future steps in address list development will change this number. While we will deliver questionnaires to all of these addresses, followup activities will determine the final number. The work of our local and tribal partners in LUCA contributed to this success. But remember that LUCA is just one piece of the puzzle. We also used United States Postal Service (USPS) information and the results of our own field review and address listing operation.

The combined LUCA efforts involved a tremendous effort on the part of several divisions at Census Bureau headquarters, our 12 regional census centers, our National Processing Center in Jeffersonville, Indiana, and more than 44 percent of the 39,000 governmental units in the United States covering 85 percent of the addresses. Because of these efforts, I can say today that I am very confident that we will meet our goal--the goal we are striving for with LUCA and all our other programs--to have as complete an address list as possible for Census 2000.

This has been a massive and complex job. It has involved developing and running nearly 1,300 individual computer programs involving more than 530,000 lines of code. These programs were designed to create nearly 440,000 large-format multicolor map sheets and nearly 10 million pages of address listings for review by local and tribal governments, and more than 6.1 million smaller black and white map sheets and 13.1 million pages of address lists for use by our field staff.

There is still work to do before Census Day to ensure that our final list of census addresses is complete and current. But the work that we completed in FY 1999 has been significant, and we are in a great position to complete Census 2000 work.

ADDRESS LIST DEVELOPMENT OPERATIONS

I will now take a few moments to describe the major address listing operations that we have completed or nearly completed thus far.

Mailout/Mailback Areas

First, I will describe the three major address list operations for mailout/mailback areas--our partnership with the USPS, the LUCA program, and Block Canvassing. Mailout/mailback areas include nearly 96 million housing units that have city-style addresses used for mail delivery. Examples of city-style addresses would be "101 Main Street" or "310 Oak Street, Apt. A." These are mostly in major urban centers but also in many small and mid-sized towns and some rural areas. In these areas, we compile the address list and the USPS delivers questionnaires to every housing unit on the list at the time of Census 2000.

Since early in this decade, the Census Bureau has been working to find a better way to improve the address list and to correct the deficiencies in the 1990 address list development process. "Early" is the key word. At this point in the last census, we had minimal input from the USPS and from local and tribal governments in compiling our address list. One improvment for Census 2000 for mailout/mailback areas has been to start with our 1990 address list and then begin work early with the USPS to use their address information to update our list. Another such improvement is providing local and tribal governments an earlier and more effective opportunity to correct the Census 2000 address list for their areas.

Working with the U.S. Postal Service. Public Law 103-430, the Census Address List Improvement Act of 1994, directed that the USPS periodically provide a copy of the address information it maintains for mail delivery to the Census Bureau for use in creating and updating the housing unit address list. The Census Bureau used this information in two ways: First, to identify new streets that had been built since the 1990 census and add those new streets, their names, and the associated address information to our geographic data base so the information can be used to assign individual addresses to a census block. Second, to update the address list with new housing unit addresses built since the 1990 census.

These processes, which took place from 1995 through 1998 in the Census Bureau's

12 regional offices/regional census centers and with the assistance of more than

1,000 local and tribal governments, added approximately 7 million addresses to the

1990 census address list and determined which census block each was in.

Address List Review/Local Update of Census Addresses (LUCA). The Census Bureau developed the Address List Review Program, popularly known as the Local Update of Census Addresses, or LUCA, in accordance with the requirements of Public Law 103-430. Participants in the LUCA Program in mailout/mailback areas included some 8,400 local and tribal governments. Participating governments constituted about 50 percent of the more than 16,700 governments entirely or partially in mailout/mailback areas, and contained about 91 percent of the addresses. Most participants conducted their reviews between September 1998 and March 1999, which was mostly during fiscal year 1999. Approximately 5,800 governments returned additions and corrections to the Census 2000 address list.

Block Canvassing Operation. To assure uniform quality and a complete Census 2000 address list regardless of LUCA participation, the Census Bureau determined that it needed to field verify the entire mailout/mailback portion of the address list. This operation took place from January through May 1999 and identified additions, address corrections, and deletions to the address file.

Attachment 1 illustrates the effect of LUCA and Block Canvassing on an address list for mailout/mailback areas that started with over 90 million addresses and that now contains nearly 96 million addresses. The "Original MAF Address" line on this attachment reflects the address list after we updated the 1990 list with information from the USPS. MAF stands for Master Address File, another name we use for the Census 2000 address list. As you can see, these processes added some 11 million addresses and deleted about 5 million addresses and there were an additional 380,000 deletes as a result of conversions from residential to non-residential.

Update/Leave (Mailback) Areas

Now, I will discuss address list operations for update/leave (mailback) areas. Update/leave (mailback) areas include about 24 million housing units that have many different address types, mostly in small towns and rural areas where address systems have less geographic structure. For example, an address might be "Rural Route 1" or it might be a postal box number. Census enumerators will deliver questionnaires to every housing unit on the list. They will also at the same time check for any missing addresses.

Address Listing Operation. In update/leave (mailback) areas, the Census Bureau created the initial Census 2000 address list through a systematic field operation called "Address Listing." During Address Listing, which took place from July 1998 through February 1999, temporary Census Bureau staff visited every housing unit they could find in their assignment areas to obtain (where possible) the occupant's name, mailing address, and telephone number. If the housing unit did not have a clearly posted address, the address lister recorded a location description. In all cases, the address lister also recorded the relevant census geographic codes to document the location of each housing unit, and noted the housing unit's location on their Census Bureau map. The Census Bureau then keyed the address and related information to add it to the Census 2000 address list. These processes listed approximately 24 million addresses that were included in the Census 2000 address list.

Address List Review/Local Update of Census Addresses (LUCA). Just as it did in mailout/mailback areas, the Census Bureau invited officials of the nearly 30,000 local and tribal governments entirely or partially covered by the Address Listing operation to review the relevant portions of the Census 2000 address list. This 1999 phase of the LUCA program involved nearly 10,800 local and tribal governments. This is about 36 percent of covered governments, which included about 67 percent of the addresses. Most participants conducted their reviews between February and May 1999.

Approximately 5,000 governments identified roughly 114,000 blocks in which they thought the Census 2000 address list was not correct. Census Bureau staff have completed rechecking the Census 2000 address list for these blocks and has added corrected information, where appropriate, to the MAF.

Next Steps

Completing the LUCA Process. This fall the Census Bureau will provide documentation to each of the LUCA participants in mailout/mailback areas and in update/leave (mailback) areas showing the Census Bureau's response to each of their LUCA suggestions. This is called the Final Determination. They can use these results to decide whether they wish to appeal any addresses to the Census Address List Review Appeals Office that has been established by the Office of Management and Budget in accordance with the provisions of Public Law 103-430. Participants will have 30 days to review the lists and file an appeal. By law, the appeals are all to be decided before the decennial census date to allow sufficient time to prepare and deliver questionnaires for any addresses the Census Bureau must add to Census 2000. In late 1999, the Census Bureau also will provide an opportunity for participating governments to review the list of special place addresses that will be used for Census 2000; this information was not available when the initial reviews were completed. Local and tribal governments with city-style addresses will also have an opportunity, beginning in January, to identify newly constructed housing units that we will need to visit to determine if they should be enumerated.

Additional Updates from the U.S. Postal Service. The Census Bureau plans to supplement the Census 2000 address list for mailout/ mailback areas with several "refreshes" from the USPS information, the last in January 2000. The additional addresses from these refreshes will be delivered questionnaires.

Completing the Census 2000 Address List. The processes described above cover more than 99 percent of the housing units in the United States and Puerto Rico. In very remote areas, and in areas with significant seasonal resident populations, the Census Bureau will employ a "List/Enumerate" method of enumeration. From late-March through mid-May 2000 (late-January through March 2000 for remote areas in Alaska), temporary Census Bureau staff, called enumerators, will collect the same street address (or location description) information as their counterparts did during the Address Listing operation and note the same types of map updates and map spots. Thus, the Census Bureau will create the Census 2000 address list for these areas as part of the enumeration process, rather than before the census. At the same time, the enumerators also will conduct the enumeration (this strategy will ensure that expensive follow-up visits need not be made to the housing units in these areas).

SPECIFIC ISSUES

Next, I will address the three specific issues listed in your letter of invitation.

1.) Illustrate the participation rates of eligible local governments in the LUCA program, and explain the Census Bureau's definition of participation

We define as "participating" those governments that received address materials for review and had not, to our knowledge, officially dropped out of the program. In order to receive materials for review, localities had to express an interest in participating and submit a signed pledge to maintain the confidentiality of the materials. Public Law 103-430 protected the confidentiality of the Census 2000 address list by including it under Title 13 of the U.S. Code. Only those individuals officially designated by their government who sign a Confidentiality Agreement may have access to the Census 2000 address list.

A government jurisdiction could include both mailout/mailback and update/leave (mailback) procedures, and, thus, could be invited to participate in both phases of the LUCA program. As I mentioned earlier, the overall participation rate was about 44 percent and included 85 percent of the addresses; the participation rate for mailout/maiback areas was about 50 percent and included about 91 percent of the addresses; and the participation rate for update/leave (mailback) areas was about 36 percent and included about 67 percent of the addresses.

2.) Discuss the percentage of households in the United States that are covered by the LUCA program, and explain some of the hazards to local governments for not participating

Over 99 percent of the U.S. housing units are in areas covered by the LUCA program and participating governments contain about 85 percent of the housing units.

Participating governments used their knowledge about local situations to help the Census Bureau improve the quality of the address list for their areas. Governments that did not participate in LUCA missed this opportunity to help ensure a complete and accurate Census 2000 by participating in this program, but as I have explained, LUCA is just one of several ways that we compile addresses.

We realize that many local and tribal governments may not have the staff, resources, and expertise needed to participate in the LUCA program and we have taken steps, when necessary, to compensate for that. I have already mentioned the Block Canvassing. We added this to ensure uniform quality and a complete Census 2000 address list regardless of LUCA participation. The Census Bureau determined in the summer of 1997 that we needed to add a 100-percent field verification activity in areas with mostly city-style addresses. We believed this operation was needed to ensure address list completeness and quality. In the update/leave (mailback) areas, census enumerators will conduct a 100-percent field canvass of addresses at the time they deliver questionnaires and we have significantly strengthened quality control procedures in these areas. This effort will also help compensate for those areas that did not participate in the LUCA.

3.) Please address the concerns of local governments regarding delays in returning the corrected address lists to the local governments for LUCA 1998

These are serious, legitimate concerns. The Census Bureau experienced problems and delays with its initial plan. We then put in place a plan that corrected for these earlier problems and that set forth a more realistic schedule. We appreciate that changes to an announced program make it difficult for our partners who must plan ahead in their use of staff and resources. Their frustration is understandable. Below I will explain how the new plan does try to compensate by reducing the operational burdens placed on our partners. Although the Census Bureau paid a price in public relations with our partners, we believe that the design changes have improved the ultimate quality of the address list -- which is the goal we all share.

The Census Bureau has made every effort to inform local and tribal officials as soon as possible when program changes are necessary and to minimize the effect of those changes wherever possible. It is my belief that a significant majority of the local and tribal partners in the LUCA program have found the Census Bureau to be responsive to their concerns, but on this I await with keen interest the GAO presentation of their systematic survey. And I am pleased that, despite the problems we've encountered, our local and tribal partners have been able to participate fully to make LUCA a successful program. We have encountered problems, but because we started early, we have had time to make refinements and correct the problems. Through our extensive outreach efforts, nearly twice as many local and tribal governments are participating in LUCA as participated in the 1990 Post Census Local Review; those governments cover about 85 percent of all addresses. So, thus far, response to the program has been great.

I earlier noted that we tried to reduce burden on our partners compared to the original design. To illustrate this, consider the issue of returning address corrections to the local and tribal officials. Under our original plan, the LUCA program would have had the following seven steps. First, the Census Bureau was to provide participants with addresses for review. Second, the participants were to provide the Census Bureau the results of their review. Third, the Census Bureau was to match the participant results to the results of the Block Canvassing operation and provide the participants detailed feedback on that match. Fourth, the participants were to review the detailed feedback to respond and request reconciliation of addresses they believed the Census Bureau erroneously excluded or included from the list. Fifth, the Census Bureau was then to verify the existence of disputed addresses and process the results. Sixth, the Census Bureau then would issue a final determination that would include the results of this final verification. Seventh, after receiving the final determination materials, participants could appeal any address they still believed the Census Bureau had not included on the Census 2000 address list.

Based on what we learned in the Dress Rehearsal, the Census Bureau simplified and streamlined this plan. This change involved combining the detailed feedback with the final determination, that is combined steps 3 and 6. A letter and flyer announcing the change were sent to every affected local and tribal government in February 1999.

Basically, this change means that we are simplifying the program for participants. We will provide them information after we have verified not just disputed but all addresses provided by the participant that do not match to the results of the Block Canvassing operation. Participants will no longer need to do a separate review of detailed feedback addresses that do not match to the Block Canvassing operation as well as identify specific addresses they believe are still missing or are incorrect in the final determination address list. It saves time in the schedule because we do not need to produce, deliver, and ship detailed feedback materials separately from the final determination materials. This means we were able to start verifying addresses sooner because we will no longer have to wait for the reaction of the local and tribal governments. I should note that we will mail questionnaires even to addresses that did not match to the Block Canvassing operation. We will followup on any addresses that have been verified and for which we do not receive a returned questionnaire.

You may have also heard about delays in our LUCA program for update/leave (mailback) areas. As a result of the delays, we were forced to compress the review time that local and tribal participants had. But it is important to stress that although the review time was reduced, the process was greatly simplified by eliminating the previously required address-by-address review. Instead, the Census Bureau asked that local and tribal governments document only the changes in the housing unit counts at the block level.

Mr. Chairman, in conclusion I want to emphasize three main points: First, despite problems we encountered, the LUCA program has been a success. Second, LUCA is just part of a much broader, multi-part process to develop an address list for Census 2000. And third, I am confident that we will have an accurate and complete address list for Census 2000. We began early in the decade working on our address list for Census 2000. We have already begun looking at ways to improve the process for the next census and we believe the American Community Survey offers an opportunity to keep the address list continually updated throughout the decade. I will be pleased to answer any questions.

ATTACHMENT 1

CURRENT OPERATIONAL STATUS

Address List Development--CITY STYLE

RESULTS

Original Master Address File Addresses 90,304,226

Plus Additions 11,094,789

Included in both LUCA and Block Canvass 3,072,069

Included in Block Canvass Only 5,687,293

Included in LUCA Only* 2,335,427

Minus Deletes 5,034,356

Deleted by both LUCA and Block Canvass 184,577

Deleted by Block Canvass Only* 4,606,730

Deleted by LUCA Only 243,049

Minus Conversions from Residential

to Non-Residential 379,208

Both LUCA and Block Canvass 4,712

Block Canvass Only 356,024

LUCA Only 18,472

TOTAL 95,985,451(numbers as of 9-99)

*Final result will change after including the findings from "field" confirmation which will be completed in November, 1999.