Shifting demographics coupled with a decade of budget and management woes have put the 2010 census on shaky ground.
It's easy to see why the Census Bureau's latest promotional video is getting rave reviews on Capitol Hill.
The three-minute presentation, A New Portrait of America, is "beautifully photographed, beautifully choreographed," gushed Sen. Thomas Carper, D-Del., at a recent Senate hearing. With soaring music and stirring images of Americans of all ethnic backgrounds at work in schools, hospitals and fields, the video declares that the 2010 census will determine the course of more than $300 billion in federal spending, and will be "the best-planned and most well-researched census ever."
But there's a disconnect between the real-life census and the video version. Not all Americans will want to be counted, Carper said at the May 15 nomination hearing for incoming Census Bureau Director Robert M. Groves. And the Census Bureau itself "has faced many operational and management challenges that have jeopardized its success," Carper warned.
The Census Bureau could have a harder time counting the nation's population in 2010 than at any point in its 107-year history. Mandated by the Constitution and conducted every 10 years since 1790, the national head count is at best a daunting operation. But this time around, big external and internal challenges put the census at high risk of failure, according to the Government Accountability Office.
The economic crisis, widespread unemployment and housing fore- closures, Hurricane Katrina, and an influx of immigrants, many illegal, all will make millions of Americans harder to count. Unconventional housing arrangements, from trailers to tent cities to families doubled up, are on the rise. Demographic and generational shifts, from changing household patterns to the growing use of cell phones and e-mail, also make Americans harder to find.
For the Census Bureau, the timing couldn't be worse. Budget shortfalls, leadership shake-ups, staff turnover, and management problems have made for a rocky decade. A costly technology contract that was supposed to modernize outmoded census-taking methods foundered. This cut short the bureau's dress rehearsal, meaning key elements of next year's census are untested. Cost estimates have soared to $14 billion or $15 billion-a record that translates into roughly a $100 per household to be counted.
Partisan census squabbles, in the meantime, are alive and well. Since the census will apportion seats in the House and in state legislatures and help draw district lines, political scrutiny is relentless. The Obama administration's awkward handling of census questions earlier this year fueled lingering disputes over statistical sampling. The controversial method of adjusting for undercounts involves comparing results from the main count and a large sample survey of the population.
Census Bureau officials have assured repeatedly that sampling, which Republicans suspect would boost the numbers for urban populations thought to favor Democrats, will not be used in 2010 or in the foreseeable future. Yet GOP lawmakers remain hyper-vigilant to any sign of Democratic meddling.
Sen. Judd Gregg, R-N.H., Obama's initial nominee to head the Commerce Department, Census' parent agency, withdrew after a presidential press aide suggested that the White House would help directly oversee the census. The Obama administration quickly back- pedaled, but the flap delayed the installation of both Commerce Secretary Gary Locke and Groves, who now steps in too late to make any substantive changes for the 2010 census. In early June, Senate Republicans held up Groves' confirmation.
Not all signs point to a census disaster. An extra $1 billion for the count in the economic stimulus package will fund additional public outreach and partnerships with state and local community groups. By all accounts, these partnerships vastly boosted responses from hard-to-count populations in 2000. Also, the economic pinch is making it easier for the bureau to hire its army of temporary workers.
Not all recent innovations have fallen flat. Success stories include a better mapping program, more multilingual outreach, and the elimination of the long form questionnaire, which asked a sample of households in detail about matters such as occupation and education. The long form now has been replaced by an ongoing American Community Survey, which is more up to date, and residents next year will receive one short census questionnaire only.
Still, problems remain. A troubling May report by the Commerce Department's inspector general warns that temporary workers hired this year to canvass addresses for three months were doing a questionable job. An incomplete address list means an inaccurate census, critics say. Drafts of ads promoting the census, produced under a multimillion-dollar communications contract, were panned in April by an advisory board. Top talent has left, and a wave of retirements is on the horizon.
One of the biggest hurdles is the Census Bureau has been slow to modernize, despite big investments. Contract woes in 2008 have forced the bureau to abandon plans to use a costly new handheld computer device to follow up with households that fail to mail their census questionnaires back. The handhelds were used this year for address canvassing only.
The upshot is the bureau is using essentially the same methods-forms mailed out and back, pencil-and-paper follow up-as it has since 1970.
"They have to update their communications systems, their computers, their databases," said Rep. Carolyn Maloney, D-N.Y., who has authored legislation that would make the Census Bureau a stand-alone agency separate from the Commerce Department. "And they have to adjust to changes in our country-cell phones, transients, diverse populations. They have to adjust to what the 21st century really is like."
'There Is No Do-Over'
It's a tall order for an agency that operates in relative obscurity and in the Commerce Department's shadow most of the time, to be thrust onto the political hot seat once a decade. Considered the nation's largest peacetime mobilization, the census involves hiring some 860,000 temporary workers, opening 500 local census offices nationwide, and counting a population that in 2010 is expected to top 310 million, in 140 million households.
Adding to the challenge are fixed statutory deadlines-Census Day on April 1, delivery of the official population count on Dec. 31-that require specific steps to build in sequence.
During the past decade that sequence was disrupted. Budget stalemates interrupted the year-by-year funding ramp-up that normally precedes the census. Tensions between the Commerce Department and the Census Bureau helped fuel the departures in 2006 of well-respected bureau director Charles Louis Kincannon and his deputy, Hermann Habermann. Kincannon's replacement, Steve H. Murdock, won high marks, but he left after barely a year. Acting Director Thomas L. Mesenbourg has been in charge since then.
Some blame these and other senior-level departures for the problems that plagued the bureau's $600 million contract with Harris Corp. of Melbourne, Fla., to develop handheld computers and provide support services. When that contract ran aground amid technical and scheduling problems, some blamed lack of oversight by top managers.
The bureau's decision to revert to pencil and paper for the nonresponse follow-up portion of the census pushed up costs by $2 billion to $3 billion. It also curtailed the bureau's scheduled dress rehearsal, ringing alarm bells at GAO. The bureau's information technology systems have not been tested end to end, GAO has warned. Neither has its plans to send out a second mailing to certain households, or to fingerprint hundreds of thousands of temporary census workers, both for the first time.
"It's all high risk," says Robert Goldenkoff, director of GAO's strategic issues team. "There are these unresolved issues, unanswered questions, less than a year away." With the operation's fixed deadlines, Goldenkoff adds, "There is no do-over, there is no time out, there is no reset button. And so things are coming down to the wire, and we're concerned by what we're seeing."
The bureau has taken steps to recover and has been responsive to criticism, according to observers. Census has invited state officials to help fill in the blanks with missing addresses, which tend to be hardest to identify in urban areas, and has added training to help workers find unconventional housing units, such as basement and garage apartments and doubled-up housing units.
"Adjustments were made in technology, adjustments were made in management, money is pouring into the bureau," says Andrew Reamer, a fellow in the Metropolitan Policy Program at the Brookings Institution, a Washington think tank. "They are well aware of the risks. No one is hiding that. And they are doing their best to adjust to those risks."
For all the uproar over the handhelds, which employ Global Positioning System technology, early reviews are not bad. While chatter on blog postings suggests there are still kinks-including bugs with a fingerprint scanner- temporary workers interviewed for this story did not report difficulties. One reason for this is that when problems do arise, workers using handhelds have access to a help desk crew-something GAO had recommended.
"Our first operation with the handhelds is to address canvassing, and we are well ahead of schedule," says Arnold A. Jackson, the bureau's associate director for decennial census. "We are almost done with the operation. The handhelds performed superbly."
Racing the Clock
The Commerce IG's report warns that the lightning pace of address canvassing might have proved problematic. Field observations and statements from temporary workers show that "listers are not consistently following certain key procedures," the report says, including knocking on doors and actually walking down rural roads to pinpoint addresses on GPS maps.
"We have received reports from Census field staff that they are under intense pressure to complete their assignments within a limited time frame and to minimize or avoid overtime," the report adds. Address canvassers have voiced similar complaints, both in blog postings and in recent interviews.
"There was great emphasis on getting the address canvassing done rapidly, and the emphasis on getting it done rapidly interfered a bit with getting it done correctly," says canvasser Reynolds Farley, a retired demographer from the University of Michigan who's relied on census data for his research and writings.
Laura Mansnerus, a former New York Times reporter who worked as a temporary crew leader during address canvassing in Philadelphia, agrees: "Everything was focused on speed at the cost of accuracy." Mansnerus complained that her crew of 19 was trained for one week to canvass a region that they were hurried through in two weeks.
The May report "almost jumped off the page at us," Carper told Groves at his nomination hearing. "I urge you to bear down on that right away." Census Bureau officials have sent an e-mail and scheduled a teleconference with regional directors, according to the IG.
Another red flag is the bureau's $212 million communications contract, which includes a much-touted paid advertising campaign. The bureau's Joint Advertising Advisory Review Panel gave early versions of the ads, which were produced by communications contractor DraftFCB, a vote of no confidence in April.
"A smart advertising campaign would try to use the economic crisis as a leveraging opportunity," but the DraftFCB ads failed to do that, says Arturo Vargas, executive director of the National Association of Latino Elected and Appointed Officials and a member of the JAARP. (The paid DraftFCB ad campaign is unrelated to the in-house promotional video that so impressed Carper.)
Vargas concedes that DraftFCB "took our criticisms to heart," but he has yet to see revisions. Census officials have defended the ads, which are still in development.
Groves has no illusions about the problems that await him. In a pre-hearing questionnaire he acknowledged the Census Bureau "is perhaps at one of its lowest ebbs in scientific talent," and 45 percent of its staff is eligible to retire within the year. He has pledged to brainstorm new technologies, recruit a more diverse staff, improve contract oversight and resign if politically pressured.
A noted specialist in survey response methods, Groves has won plaudits from colleagues, policy experts, activists and lawmakers on both sides of the aisle. He most recently directed the University of Michigan Survey Research Center, and previously served as Census' associate director from 1990 to 1992.
"He knows the methodology, he knows the decennial census, he knows the issues, he knows the people both internally and externally," says Edward Spar, executive director of the Council of Professional Associations on Federal Statistics. "He's a wonderful choice."
Even Republicans on Capitol Hill, initially wary because Groves endorsed adjusting the count during his census tenure in the 1990s, have welcomed his reassurances. He has stated point-blank that statistical sampling is not on the table for 2010 or beyond.
The partisan tugs-of-war will continue, of course. Civil rights advocates remain concerned that African-American, Latino and low-income Americans will be undercounted. GOP anxiety over sampling remains high, despite arguments against it. The Supreme Court ruled sampling unconstitutional for apportionment in 2000, and Census has repeatedly rejected its use for enumeration.
Some argue sampling is a red herring that distracts from the real issue: How the Census Bureau will replenish its staff, modernize its operations and keep up with quickly changing demographics-all while containing mushrooming costs. The bureau invests more money and effort each decade to get essentially the same results, yet has been slow to embrace change, according to GAO's Goldenkoff.
The challenges in 2010 reflect society's increasing complexity, he says, but also "reluctance on the part of the bureau to fundamentally reexamine how it takes the census. The bureau has long passed the point of diminishing returns."
By 2020, demographers and statistical professionals note, the handheld computers Census spent so much to develop could be obsolete. The U.S. Postal Service as we know it might no longer exist. Already many Americans have cell phone numbers and e-mail addresses that follow them wherever they go. Yet the census "is still essentially a mail out-mail back survey," notes Spar of the Council of Professional Associations on Federal Statistics.
Groves should direct the 2020 staff to rethink that 50-year-old paradigm, say top demographers, and consider using the Internet to reduce costs. The bureau twice studied Internet use earlier in this decade, but concluded that it failed to boost response rates and raised privacy concerns.
The bureau also should consider using data such as birth, death, welfare and driver's license records to improve its address lists and target households, the National Academy of Sciences has recommended. Some even argue for a national identification system-something that's prevalent in other democracies but remains controversial in the United States.
"Until you get an administrative record system that's tied to a national ID system, you don't know who's being left out based on a certain set of characteristics," says Kenneth Prewitt, a professor at Columbia University's School of International and Public Affairs. Prewitt directed Census from 1998 to 2001 and now is a consultant to the bureau.
For Groves and his colleagues the immediate challenge is getting next year's census back on track. At his nomination hearing, he told senators: "Problems will arise in the 2010 census. I guarantee this. It's too large an endeavor to go completely smoothly." For all the controversy swirling around the 2010 Census, that's one statement few would contest.
Eliza Newlin Carney is a contributing editor for National Journal.