in espanol
mission statment
board of directors
reports to congress
contact us


Field Office Staffing

Workload: The Bureauís goal is to obtain a completed questionnaire for every housing unit identified in Census 2000. The majority of those forms (the Bureau now estimates 61 percent) will be returned voluntarily by mail. The remainder -- about 46 million forms -- will be collected during an operation known as Non-Response Follow-Up (NRFU).7

Most of this operation will be conducted by enumerators -- part-time census employees that go door-to-door to addresses that fail to return a census form. Enumerators make up the majority of the field staff during NRFU. During the 1990 census, about 300,000 employees worked in the field during peak activities. Staff turnover necessitated hiring and training (from census start to finish) more than 550,000 field employees to fill those 300,000 positions.

Due to a number of changes -- in the population to be counted and in the plan to count it -- more enumerators will be needed for Census 2000 than originally planned. There are more people -- 118 million households in 2000 compared to 107 million in 1990. The Bureau anticipates more households will fail to return a census form by mail -- 46 million estimated non-respondents in 2000, compared to 35 million in 1990. Also, in compliance with the January 25, 1999 U.S. Supreme Court ruling, the Bureau dropped its plan to canvas only a sample of non-respondents, and will conduct 100% follow-up -- by sending an enumerator to every household that does not return a form by mail.8

Basic Enumeration: NRFU is scheduled to begin on April 27, 2000, and run for ten weeks. Local Census Office (LCO) managers and crew leaders will assign each enumerator a list of addresses in their area that have not returned a census form by April 16. Each enumerator will be instructed to visit each address as many as three times, and make as many as three follow-up phone calls, in order to obtain a response from the people at each address. If these attempts are unsuccessful, the enumerator is instructed to get "proxy data" -- information about people living at the address from a third party such as a neighbor, a letter carrier, or someone else who might have some idea about the household. If all else fails, the enumerator makes a judgement of whether or not the address is occupied. Bureau headquarters will then "impute" people into the household -- assign the household characteristics similar to those of neighboring households.

Targeting: Some enumerators will have a harder assignment than others. The Bureauís research has identified specific areas that will probably be hard-to-enumerate (HTE) in 2000. Useful criteria to identify these areas have been listed and quantified by Bureau professionals in the Planning Database (PDB). The PDB "provides a systematic way to pre-identify potentially difficult to enumerate areas that should be flagged for special attention in Census 2000."9

The PDB measures about 20 variables that describe a hard-to-enumerate census tract 10, such as a high percentage of renters, multiunit buildings, no telephones, low-income families, and single-parent, crowded or minority households. Non-English-speaking households and linguistically isolated areas are also identified, and the non-response rate for the 1990 Census is considered. These variables are scored and summed. The higher the score, the harder the neighborhood is to count.11

The PDB distinguishes between hard-to-enumerate and easy-to-enumerate neighborhoods. Any neighborhood with a score over 90 and a 1990 non-response rate over 43% is considered "very hard to count." A score of less than 30 and a 1990 non-response rate less than 25% is considered "easy to count."12

According to the PDB, about 54% of the neighborhoods in the country are easy to count, and probably will not need any extra attention. About 5% of neighborhoods -- 2,689 -- are very hard-to-count, and will probably have high undercounts unless something extra is done.13

The PDB successfully predicted the neighborhoods with low response rates during the 1998 Dress Rehearsals. In addition, partnership specialists are expected to use and update the PDB during the summer and autumn of 1999. A working group is converting the PDB into user-friendly format for this purpose.

Career professionals at the Bureau concluded, "The predictive effectiveness of the planning database and hard-to-count(HTC) scores has been proven ... in the 1990 census, 1995 test census, and the Census 2000 Dress Rehearsal ... That is, we can target right now the groups of tracts that will likely have low response rates in the absence of any special attention."14

Special Enumeration: Section IX-F of the Bureauís January 1999 operational plan consists of two pages listing brief descriptions of eight strategies to improve enumeration in HTE areas.

Strategies include the use of the Planning Database (PDB), and the Questionnaire Assistance Center (QAC), Be Counted and language assistance programs discussed elsewhere in this report.

In addition, some areas will be identified for special treatment in the mailout / mailback and non-response follow-up phases of the census. "Areas will be designated for the targeted methods, such as urban update leave (UUL), update/enumerate (UE), and team and blitz enumeration, that will be used in Census 2000."15

The Board has asked for greater detail regarding the Bureauís tactics to improve the count in targeted areas. In response, Bureau officials have cited the "tool kit," a series of special techniques that could be used by Regional Directors to count HTE areas. The Board requested a copy of detailed tool kit procedures, but the tool kit has not been formalized to this degree.

Rather than a collection of formalized procedures, the "tool kit" referred to by the Bureau is the collected knowledge and experience of its current staff Ė in particular its Regional Directors who meet regularly to share information.16

The Board recognizes and appreciates the institutional knowledge of the Bureauís Regional Directors, and their ability to employ ad hoc enumeration techniques that will improve the count in HTE areas. We also recognize and concur with the Bureau's commitment to providing those directors with "maximum flexibility" in enumerating HTE areas in their region.17

However, the Board also believes that the Bureau should formally document special enumeration procedures to be used in targeted HTE areas. The Bureau should prepare such documentation Ė standardized texts, instructions or manuals Ė for national distribution and use during Census 2000. Such documentation will ensure that valuable knowledge acquired during one census is not lost to the next due to staff turnover or retirement.

Staffing Needs: Targeting HTE neighborhoods should include identifying what skills or background an enumerator needs to most effectively canvas the neighborhood. The professionals at the Bureau have recognized that more and better data will be collected when the enumerator who knocks on a door looks and sounds like the person who answers it. This is especially significant in HTE neighborhoods (including neighborhoods where English is not the first language), where fear or distrust of government will be reinforced by an unfamiliar face on the doorstep in the evening. Ideally, enumerators should live in the same neighborhood as the people they enumerate. At the very least, they must share a common language.

The Bureau has made it a priority to hire "indigenous" enumerators: people who live in the area they canvas. The Bureau also reports that hiring multilingual enumerators, including non-citizens legally authorized to work in the U.S., is a priority. These are excellent objectives that will improve the accuracy of the census.

We agree the Bureau should make focused efforts to identify the staffing and language needs in neighborhoods they have already identified as hard-to-enumerate in 2000. Without these efforts, the Bureau cannot recruit, hire and train a workforce best suited to reduce the differential undercount of minority communities during NRFU.

Indigenous Enumerators: The Commerce Departmentís Office of the Inspector General (OIG) reported that efforts to assign enumerators to work in or around their own neighborhoods during the 1998 dress rehearsal were largely unsuccessful. In Sacramento, the OIG found that, of 154 enumerators selected at random, none were assigned to work in their own neighborhood.18 The South Carolina site had similar problems, attributed largely to impractical and inaccurate Bureau maps that could not be used to match an enumeratorís home address to a census tract.19

The Bureau responded with its intention to improve recruiting maps and to identify enumeratorsí home census tracts during recruiting. The OIG agreed the Bureauís proposed improvements should address the problem.

We believe additional efforts should be focused on HTE areas. HTE neighborhoods should be identified and highlighted in the assignment area for each local census office (LCO). Staffing goals (a minimum number of Spanish-fluent enumerators, for instance) should be set for each HTE neighborhood, and incorporated into recruiting efforts in that area.

Temporary Employment of Non-Citizens: As part of its recruiting efforts to ensure the full enumeration of HTE neighborhoods, the Bureau will face special challenges in communities with large concentrations of recent immigrants. These families are the most likely to be linguistically and culturally isolated, and may not be aware of the Bureauís unbroken track record of strict confidentiality. For those newest Americans who were victims of political, religious or ethnic persecution in their countries of origin, the idea of reporting any personal information to the federal government may be very intimidating.

Identifying and hiring field staff who are able to bridge these barriers of language and culture will be vital. However, many of those most qualified to carry out an effective enumeration of these households are likely to be recent immigrants themselves and may not yet be U.S. citizens.

The Bureau recognizes that the hiring of non-citizens who are authorized to work will be needed, but current regulatory and statutory restrictions may hamper effective recruitment of qualified non-citizens.

Current federal regulations allow the Bureau to hire non-citizens when necessary, but only in rare cases when no U.S. citizen is available. 20 As a result of this restriction, the Bureau must qualify its recruiting messages, warning non-citizens they will be considered for employment only after the pool of citizens has been exhausted. This will likely have the effect of discouraging non-citizens from applying at all.

In addition to regulation, ß602 of the Fiscal Year 1999 appropriations act for the Departments of Commerce, Justice, State and the Judiciary generally bars the hiring of non-citizens by the U.S. government with certain limited exceptions. Among those exceptions are emergency appointments, the hiring of temporary translators, and the employment of international broadcasters employed by the U.S. Information Agency (USIA).

Given that an effective strategy to fully count HTE areas will require the hiring of qualified non-citizens, the Board recommends that the Bureau be given an exemption from current barriers to the employment of non-citizens. The exemption should apply only to the hiring of a non-citizen where 1) the individual to be hired is lawfully present in the United States and authorized to work, and 2) the position to be filled is a temporary one necessary to carry out the 2000 Decennial Census.

Pay Strategy: Hiring and retaining an adequate number of field staff for Census 2000 is an ongoing concern. In March 1998, the Bureau estimated that 100% follow-up would require 59,000 enumerators in addition to the 300,000 staff already planned. Also, another 25,000 Ė 30,000 would be needed to verify all vacant housing units (another improvement announced after the Supreme Court ruling).21 Those estimates placed staff levels somewhere below 400,000. However, the Director of the Census Bureau told the Board in March 1999 that, "500,000 is not going to be enough."22 Senior Bureau officials have told the Board that more detailed estimates of staffing levels are being assessed, but were unavailable when this report was being prepared.

A rudimentary projection of the staffing level needed in 2000 can be derived from the ratio of  workers-to-workload in 1990. In 1990, approximately 300,000 field staff were responsible for 35 million non-responding households: just under 120 households per employee. If the Bureau maintains the same staff-to-household ratio in 2000, follow-up for 46 million households will require about 400,000 field employees during peak activities.

1990 Census
2000 Census
Total Households
106 million
118 million
Non-Responding Households
35 million
46 million (estimated 61% mail response)
NRFU Workload (households/field employee)
Field Staff (during peak activities)

*Projected using 1990 household/field employee ratio.
Source: 1990 Procedural History

Fortunately, all three dress rehearsal sites demonstrated the Bureauís improved ability to recruit and retain staff. Staff turnover was significantly lower than expected, and acceptance rates of census jobs were much higher. Where the Bureau anticipated 100 percent turnover in each site, turnover was only 41 percent, 19 percent, and 13 percent in the Wisconsin, California and South Carolina sites, respectively.23

One factor in the improved hiring is the Bureauís relatively new policy of paying competitive pay rates in each LCO area, based on local prevailing wages. The Bureau plans to pay enumerators 75 percent of the prevailing wage rate.

Offering competitive pay, while common practice in the private sector, is a drastic improvement over previous censuses. Bill Hill, a 30-year career census professional and former Director of three regions, testified that, prior to 1990, the Bureau erred when it "paid the same piece rate in Manhattan that we did in Cincinnati, Richmond and Memphis."24 Since retaining employees will save the time and cost of recruiting and training thousands of others, this improvement is an excellent example of how sound management will realize economies of scale in Census 2000.

Competitive Wages for Census 2000

Field Supervisor
$11.25 - $21.50
Crew Leader
$9.75 - $20.00
$8.25 - $18.50
Source: 22 February 1999 briefing from
Associate Director of Field Operations

Presently, the Bureau is further refining its program, to "smooth" wage rates across metropolitan areas and labor markets and ensure that staff receive similar pay for counting similar areas. This is sound management, and should be standard practice in the future.

In addition, the Bureau and Congress are working aggressively to expand the pool of potential census employees. Recent legislation introduced in the U.S. House of Representatives by Chairman of the Subcommittee of the Census Dan Miller (R-FLA) and Representative Carrie Meek (D-FLA) would allow people on public assistance, as well as former members of the uniformed services, to work for Census 2000 without any loss of their federal assistance.25 Similar provisions have been made for federal and military retirees. The Bureau is also working with states to ensure that recipients of Temporary Assistance for Needy Families can work for Census 2000 without risk to benefits.

Training: The Bureauís method of training field staff has remained essentially unchanged over the past few decennials. Instructions are presented to a class of trainees via lecture and discussion, delivered verbatim from a training guide. A "pyramid" system is used, wherein each employee, after a few days or weeks of on-the-job experience, trains the people he or she will supervise. For example, LCO Managers train their Field Operation Supervisors (FOSs), who train Crew Leaders, who, in turn, train enumerators.

Census 2000 Training: RCC Managers

3 day RCC Management Overview classroom training.
Job-Specific classroom training for selected RCC Managers -- 1 day per training course.
Crisis Communication and media skills training for selected RCC Managers -- 2 to 3 days.
Various --Just-in-Time/Operational Briefings-- for Regional Directors, Assistant Regional Census Managers, and Area Managers ó 1 to 3 days per session.
Source: 22 February 1999 briefing from
Associate Director of Field Operations

Depending on the position, training includes two-to-four days in the classroom, followed by a half-day in the field for enumerators, and various briefings and additional training sessions for others. It should be noted that enumerators for the Accuracy and Coverage Evaluation (ACE) will receive essentially the same training as NRFU enumerators, with some additional training customized to the ACE procedures, such as the use of laptop computers.

The advantage of verbatim training is a consistent message delivered to hundreds of thousands of employees in a manner that controls cost and timing. The disadvantage is that training is delivered primarily by newly-hired employees Ė not career professionals with years of practical knowledge and field experience.

Census 2000 Training: LCO Management Teams

LCO Management Overview classroom training -- 4 days.
Job-Specific classroom training for LCO Managers, Assistant Managers, and LCO Automation Technicians ó 1 day per training course.
Media skills training -- 1 to 2 days for LCO Managers.
Various "Just-in-Time"/Management Operational classroom training sessions for LCO Management Teams ó pre-classroom self-studies for 5 sessions and 1 day of classroom time per session.
Source: 22 February 1999 briefing from
Associate Director of Field Operations

The Bureauís evaluation of training during the 1990 census reported that, "some trainees felt [the training] tended to assume ideal conditions, and did not fully prepare them for emergencies or hard-to-enumerate situations."26

Although it would be impossible to plan for every eventuality, examination of the training manuals used in the 1998 dress rehearsal tends to reinforce this concern. The Commerce Departmentís Office of the Inspector General (OIG) reported:

We found a number of shortcomings in the training materials used during the dress rehearsal. For example, student training manuals contained errors and confusing acronyms, did not always match up with instructor manuals, and did not have indexes for finding topics and answering questions. Furthermore, videos referenced in the manuals were not always used. As a result of these training shortcomings, we found that enumerators were not always prepared and occasionally mishandled questions and problems during enumeration. Census needs to improve training materials to ensure enumerators are better prepared.26

Census 2000 Training: Enumerators

2 1/2 day classroom training on NRFU duties and responsibilities.
1/2 day of interviewing practice in the field.
On-the-job training for marginal performers -- given by Crew Leaders in the field.
Source: 22 February 1999 briefing from
Associate Director of Field Operations  

The Bureau responded to the OIG with plans to add indices and improve materials, including more rigorous review of manuals before their distribution. These are necessary improvements to the basic enumeration training. However, the Board concurs with career professionals who have noted that customized training would further benefit those enumerators assigned to HTE areas. Retired Regional Director Bill Hill testified:

         We must evolve away from the one-size-fits-all mentality we have taken censuses withÖ. This administration, rightly so, trumpets our national diversity. Census
         procedures should deal with that diversity through diverse means.

         Critical to success of this is to get these plans and the philosophy of this in your manager trainingÖ prepare special training for enumerators who will work the hot spots.

Regional and local staff and partners, those geographically closest to the population being counted, will often be the best judges of how and where to apply additional efforts. In 1994, the National Research Councilís Panel to Evaluate Alternative Census Methods reported:

         Clearly, a large-scale program of ethnic enumerators would require a significant change in traditional training methods. More resources would have to be available to train
         enumerators, and the process would take longer. It would be foolhardy to hire people and provide them with insufficient training. But the program would be well worth it if
         the program contributed to reducing the differential undercount Ė particularly if it was tied to an ongoing organizational structure.

Concerns with the Bureauís training have been documented since the 1990 census. In particular, customized training to improve the count in HTE areas, including linguistically isolated areas, would be beneficial to reducing the differential undercount.

Training manuals are being updated subsequent to the dress rehearsal experience, but were not available at the time this report was prepared. The Board will review these materials upon their completion.