NEWS ALERT
May 18, 2000
Questions about the information contained in this News Alert may be directed to TerriAnn Lowenthal at 202/484-2270 or, by e-mail at terriann2k@aol.com. For copies of previous News Alerts and other information, use our web site www.census2000.org <http://www.census2000.org>. Please direct all requests to receive News Alerts, and all changes in address/phone/fax/e-mail, to the Census 2000 Initiative at Census2000@ccmc.org or 202/326-8700. Please feel free to circulate this information to colleagues and other interested individuals.
GAO Cites Census Progress, But House Chairman Angered By Memo Suggesting Bureau
Non-Cooperation Plus: Nearly 40 Percent of Household Follow-up Visits Completed; Mail-back
Rate Hits 66 Percent; Voting Rights Act Reviews of State Laws Continue
In testimony before Congress last week, the U.S. General Accounting Office (GAO) said
initial assessments of Census 2000 are "encouraging, with major operations reportedly
proceeding on schedule and generally performing as planned." GAO representatives
cited the initial 65 percent mail-back rate as "particularly noteworthy," noting
that response by mail reduces both costs and the scope of follow-up field activities while
improving data quality, even though it "does not guarantee a successful census."
The legislative watchdog agency presented its evaluation at a May 10th hearing of the
House Subcommittee on the Census.
An analysis of Census Bureau data showed that nearly all households (99.89 percent, or
76.77 million) that responded during the first phase of the census returned their form by
mail; only 0.09 percent (about 66,000) answered over the Internet, while 0.02 percent
(about 17,000) gave their answers over the phone. The GAO also discussed the 12.5 percent
gap in mail response between short form and long form households, twice the projected
response differential of 6.2 percent. The wider gap, the auditors said, was due primarily
to a higher-than-expected short form response rate (66.6% returned v. 62.1% projected)
rather than "a meltdown on the long form" (54.1% returned v. 55.9% projected).
The GAO's testimony included other analyses based on Census Bureau data, of initial
responses to the census during mail out/mail back operations. GAO auditors concluded that
mail response rates by Local Census Office (LCO) ranged from 39 to 80 percent; LCOs in
inner city and urban areas generally had better-than-expected response rates, while many
suburban, small town, and rural LCOs fell short of the Bureau's expectations; and all but
16 of the 511 LCOs had met or exceeded their hiring goals for the start of NRFU.
Interested stakeholders can obtain copies of GAO's testimony (GAO/T-GGD/AIMD-00-164) and
other GAO reports by visiting the agency's web site at <http://www.gao.gov>.
Subcommittee Chairman Dan Miller (R-FL) continued to express concern about the schedule
for completing follow-up visits to unresponsive households, suggesting that enumerators
would be forced to gather information from neighbors and landlords to finish their
caseloads on time. The GAO said the 10-week period for NRFU is "ambitious,"
noting that the bureau "needed more time to follow-up on far fewer housing
units" in 1980 and 1990, but said the larger temporary workforce in 2000 might help
the bureau complete the operation on schedule. They pointed to enumerator productivity and
turnover rates as key factors affecting the pace of the follow-up work. The Census Bureau
hopes to rely on so-called "proxy data" from indirect sources for six percent or
less of unresponsive housing units, according to GAO auditors.
Internal Census Bureau memo sparks new controversy: However, a
significant portion of the hearing focused not on Census 2000 operations but on an
internal Census Bureau memorandum from the Los Angeles regional census office concerning
various reporting requirements during the 'nonresponse follow-up' (NRFU) phase of the
count. At the start of the hearing, Chairman Miller displayed a copy of an e-mail
communication sent by a "mid-level Census Bureau Manager" to lower level
managers that he said was brought to his attention late the previous day. The chairman
highlighted one paragraph of the short memorandum, which read, "I will try to get the
D-333D report to you all on a daily basis. However this report must and cannot be shared
with any GAO representative. This a report that must be shared with any one else except
the management staff." (Note: Apparent typographical and grammatical errors were
contained in the original text.) A D-333D form is used to report the progress of Local
Census Offices in collecting information from unresponsive households during NRFU.
Rep. Miller said the memorandum is "a very serious matter that cuts to the heart of
this census and severely calls into question the Census Bureau's credibility."
"I am appalled. ...I take it personally. This Congress takes it personally," he
continued. The chairman said he would seek "full accountability" for the
memorandum's contents and that he believed a more senior Census Bureau employee had issued
the instruction to withhold the progress reports from the GAO. He concluded his statement
by questioning whether other bureau documents have been or will be withheld from public
scrutiny. "Today, there is reason to be worried," the congressman said.
The panel's senior Democratic member, Rep. Carolyn Maloney (D-NY), said she was not
informed about the existence of the memorandum until shortly before the start of the
hearing. She suggested that the subcommittee should ask Census Bureau Director Prewitt for
an explanation before drawing any conclusions about the meaning of the paragraph cited by
the chairman. Later in the hearing, Rep. Maloney read a statement from the director, who
also learned about the memorandum's existence earlier that morning. Dr. Prewitt told the
subcommittee that his agency does not have a policy directing its employees to withhold
information from the GAO, but that lower and mid-level managers are instructed to consult
with senior managers before responding to any requests from oversight bodies for
information. He noted that GAO staff already have full access to the NRFU workload data
contained in the D-333D forms.
Under questioning from subcommittee members, the GAO's representative called the e-mail
reference to his agency "disturbing," but acknowledged that the Census Bureau
has complied with all GAO requests for information and that investigators have access to
the figures reported on the D-333Ds.
'Nonresponse follow-up' visits continue at steady pace: Census
enumerators have counted 39 percent of the 42.4 million households requiring a follow-up
visit, the Census Bureau announced at press briefing on May 16. The 'nonresponse
follow-up' work started on April 27 and is scheduled to continue through July 7. The
bureau has hired 460,000 census takers to visit households that did not mail back a
questionnaire. The completion rate for follow-up visits includes vacant households and
addresses found to be nonexistent.
As with mail response rates in the first phase of Census 2000, the pace of counting
unresponsive households varied by census region, bureau officials said. The Denver and Los
Angeles regions posted the highest follow-up completion rates so far, at 50 percent, while
only 31 percent of households that enumerators must visit in person have been counted.
Dr. Prewitt said the bureau could not yet predict if NRFU would finish on schedule,
although the deadline could be met if the current pace of counting previously unresponsive
households continues. The director noted that the bureau had not yet encountered problems
with insufficient staffing and employee turnover, two difficulties that plagued the
follow-up counting operation in 1990. In an earlier statement, he praised the
"dedication and achievements of census workers ...for the continued good progress of
Census 2000."
As part of quality control efforts, census enumerators also are visiting households that
mailed back a questionnaire without sufficient information (i.e. forms listing only
occupant names or the number of residents) or with conflicting information (i.e. forms
containing information for fewer people than the reported number of residents), as well as
households reporting six or more residents or a high number of unrelated occupants. Dr.
Prewitt said two to three percent of all households were likely to encounter census
workers more than once because of various quality control operations, including the
300,000 household Accuracy and Coverage Evaluation survey.
Only days earlier, at a May 12 meeting of the Committee on National Statistics (National
Academy of Sciences), Census Bureau Deputy Director William Barron, Jr. announced that
census takers had successfully counted 30 percent of the unresponsive households.
"[T]he census continues to be running very well and all operations are meeting or
exceeding our expectations," the deputy director said. The GAO reported to Congress
on May 10 that 17.4 percent of the field follow-up had been completed by May 8.
Rep. Carolyn Maloney called the progress of the follow-up visits an "amazing
accomplishment," and added that the "good news ...ironically is disappointing
some in Congress."
Final mail response rate revised: Check-in of late census questionnaires
has boosted the Census 2000 mail response rate to 66 percent, the Census Bureau reported
last week. The revised figure exceeds the 1990 mail-back rate of 65 percent, marking the
first time since the Census Bureau started distributing census forms primarily by mail in
1970 that the mail response rate has not declined from the previous decennial count.
Seventeen percent of states and localities met or exceeded their 'Plus 5' goal of
increasing their mail response by at least five percent over their 1990 rate.
A revised mail response rate is not unusual; in 1990, the initial 63 percent response rate
crept up to 65 percent as the Census Bureau accounted for all questionnaires returned by
mail. The 2000 mail response rate represents the percent of all households in the 'mail
back' universe that returned a questionnaire by mail or provided their answers by
telephone or over the Internet. It does not include 'Be Counted' forms, which do not bear
a unique geographic code tying the response to a specific address, or remote households
that were counted by enumerators during 'list/enumerate' or update/enumerate' operations.
Congressional hearings: The House Subcommittee on the Census will hold a
hearing on May 19 on the Accuracy and Coverage Evaluation (ACE) phase of Census 2000.
Census Bureau Director Kenneth Prewitt will discuss the program, which is designed to
measure the accuracy of the population counts derived from the mail out/mail back and
nonresponse follow-up operations and to correct undercounts and overcounts using
statistical methods. Preliminary activities for the ACE program begin this month, with
interviews of the 300,000 households in the quality-check survey starting June 19. The
hearing will start at 9:30 a.m. in room 2247 Rayburn House Office Building. (Note: This
hearing originally was scheduled for May 18.)
Stakeholder activities: The United States Conference of City Human
Services Officials (USCCHSO), an affiliate of the U.S. Conference of Mayors (USCM), held a
press conference on May 9 to highlight the involvement of city leaders in the Census 2000
promotional campaign, How America Knows What America Needs. USCCHSO President Willa Lister
said, "Human service officials have direct contact with traditionally undercounted
populations and can explain the importance of an accurate census count and emphasize the
confidentiality of census data." Ann Azari, a past chairwoman of the Commerce
Secretary's 2000 Census Advisory Committee, joined Ms. Lister in explaining the
"Because You Count" phase of the promotional effort. Ms. Azari, herself a former
mayor of Fort Collins, CO, said the goal and focus of the campaign is "civic
engagement." Laverne Collins, an 18-year veteran of the Census Bureau, said she
wanted to convey four key messages: (1) Census 2000 is not over; (2) people should
cooperate with enumerators when they visit unresponsive households; (3) all census
responses are confidential; and (4) it is important to answer all questions on the census
forms.
State legislative activities update: The New Jersey state legislature is
considering a bill (A. 1682) that would bar the use of statistically corrected census
numbers for redistricting. The Assembly's State Affairs Committee will hold a hearing on
May 22 to review the proposal.
Virginia officials have asked a federal court to approve its new law prohibiting the use
of census data compiled in part through sampling and statistical methods for congressional
and state redistricting. The case, Commonwealth of Virginia v. Reno et al. (Case No.
1:00CV00751), will be heard by a three-judge panel of the U.S. District Court for the
District of Columbia, in accordance with section 5 of the 1965 Voting Rights Act (as
amended). Under the statute, either party can appeal the lower court decision directly to
the U.S. Supreme Court, which must issue a ruling in the case (known as 'mandatory
jurisdiction'). The state filed the suit on April 10, 2000.
In Alaska, also one of 16 states covered by the Voting Rights Act section 5
'pre-clearance' requirements, Attorney General James L. Baldwin responded on May 4 to a
U.S. Department of Justice (USDOJ) request for additional information in support of two
bills enacted last year that (in relevant part) prohibit the use of
statistically-corrected census numbers for redistricting purposes. The state submitted
House Joint Resolution No. 44 and Senate Bill No. 99 to the USDOJ for approval on
September 21, 1999; the USDOJ Civil Rights Division's Voting Section requested more
information on November 18, 1999, "to determine that the proposed procedures do not
have the purpose and will not have the effect of denying or abridging the right to vote on
account of race, color, or membership in a language minority group." The Justice
Department has 60 days to approve or disapprove the bills, which went into law without the
signature of Governor Tony Knowles (D).
Correction: In the May 8th News Alert, we inadvertently included the
National Low Income Housing Coalition (NLIHC) as a participant in an April 25 press
conference on the importance of census long form data. Housing Assistance Council (HAC)
Executive Director Moises Loza participated in the event, along with Betty Weiss,
Executive Director of the National Neighborhood Coalition, and Ricardo Villalobos, Census
Project director for the Center for Community Change. We apologize for the error.