Census 2000 Primer
By Gilbert F. Casellas
April 2000

Few of us can recall exactly where we were ten years ago and what we were doing. I do. I recall 1990 fondly because that year I served as Chairman of the Board of Governors of the Philadelphia Bar Association, I celebrated my tenth year of practice with Montgomery, McCracken, Walker & Rhoads, and I was serving as pro bono counsel to the Philadelphia Commission on Human Relations in connection with a city wide inquiry into the treatment of Philadelphia's Latino community. Interestingly, though, I recall 1990 for another reason: I filled out my first Census form. And why, you may wonder, do I remember this? Because since 1980 (when I don't recall filling out a form), I had gotten married, had a child and bought a house. I felt so "established" and wanted to register these "facts" somewhere "official." Little did I know how my life would change. As I prepare to fill out the form for Census 2000, I reflect on the fact that I left Philadelphia in 1993 for the first of my two appointments with the Clinton Administration in Washington, I moved from Philadelphia to the Washington suburbs and after completing my government service I no longer practice law. These changes in my life are indicative of the challenges facing the federal government in 2000 as it embarks on the largest peacetime mobilization in its history ö the conduct of the decennial census.

For about two years I have served as a presidential member of the U.S. Census Monitoring Board and since last year as Co-Chair. Our role has been one of oversight, reporting to Congress and the President on the current state of planning and preparation for the 2000 Census.

A few weeks after I stepped down as Chairman of the Equal Employment Opportunity Commission, I was asked by the White House to take on this responsibility. They were very clear about two things. First -- that the 2000 Census was a top priority of the President and he wanted my help to make it a success. Second -- I wasn't going to get paid to do it. How could I say no to that?

Since being named to the Board, I have been working and meeting with the professionals at the Census Bureau, the Commerce Department, Members of Congress and staff, local and state government officials, community service organizations, academics and regular folks to get a handle on what this whole Census 2000 thing is about. From my front-row seat, I have gained a deeper understanding of the complex issues facing the Census Bureau in counting 275 million people living in the vast United States: after all, this is the largest, most diverse and most mobile population ever. I have also come to appreciate the Bureau's dedication to seeing that Census 2000 is a success as well as the need for community involvement everywhere.

In this article, I provide background and some insight on key issues relating to the census.

The need for an accurate census

We conduct a census first and foremost because it is required by the U.S. Constitution. We use it to determine the appropriate number of congressional seats allocated to states (apportionment). And within states, census data is used for drawing district lines (redistricting). Moreover, $185 billion in federal money is distributed annually to states based on census data to fund Medicaid and other such programs that provide a "safety net" for those in need.

And it isn't just the distribution of federal money that's affected; it's the allocation of state and local money as well. Literally hundreds of different funding formulas at all level of government use decennial census data as all, or part, of their distribution calculations.

In short, the census helps to define who we are as a nation, and if carried out properly, allows our democracy to function in a way that is fair and equitable to all people.

1990 Census undercount hurt people needing the most help

Regrettably, in 1990, the census missed 8.4 million people, of which half were children. The undercount for whites - - the percentage of whites missed - - was about seven-tenths of one percent. By way of comparison, the undercount for Asians and Pacific Islanders was 2.2 percent. For African Americans, 4.4 percent. For Hispanics, 5 percent. And for Native Americans living on reservations, an incredible 12 percent!

In the eastern region, the overall undercount rates were as high as 3.4 percent in Washington, D.C. and over 1.5 percent in New York, Delaware and Maryland. The undercount in Pennsylvania was just 0.3 percent but a majority of those missed were those living in Philadelphia, giving the city the ominous distinction of having the ninth largest undercount among all major cities in the United States.

Consistent with national patterns, in New York and Massachusetts, over 6 percent of African Americans were missed. In New York, Rhode Island, Maine, Connecticut, Massachusetts, New Hampshire and Vermont, over 5 percent of Hispanics were missed. In Pennsylvania, 4.5 percent of African Americans were missed as were 4.4 percent of Latinos, 2.5 percent of Asians and 1 percent of Native Americans from the state.

Consequently, for the last ten years, states, cities, and minorities (and other groups such as children and those with disabilities) did not receive the funding nor the political representation they deserved. State and local governments and many communities found themselves trying to do more for the poor and disadvantaged with far fewer resources than were needed.

Why undercounts occur

As far back as the 1790s, concerns about the compilation of accurate census data swirled. Our nation's first Census Director, Thomas Jefferson, was so sensitive to this issue that he provided President George Washington with two sets of census figures. The first represented the actual headcount and the second represented his estimate of the actual population. Jefferson knew that, because of geography, mobility of persons and other individual circumstances, headcounts by themselves missed people.

Fast-forward two hundred and ten years. The Bureau confronts the same problems as in Jefferson's day but of a far greater magnitude. A look at certain neighborhoods and residences in inner cities today offers partial insight to why we have undercounts.

In a Philadelphia high-rise, for example, security is very tight, as it should be. When enumerators visit these buildings, they will have to be "buzzed up." If they get in, they'll find many people not at home -- gone to work -- or asleep having just returned from a night shift. They'll find folks who may not speak their language or people who are in the process of moving out.

And they'll find people who are afraid to answer the door, elderly who are afraid of strangers, immigrants afraid of the INS, or people who distrust government.

Census Bureau actions

To its credit, while recognizing the challenges of today, the Bureau is moving forward with the most sophisticated census plan in history to get the most accurate count possible.

The Bureau has launched its first-ever paid national advertising campaign to build awareness of the census (a recent study that the Census Monitoring Board commissioned indicated that less than half of the population knew last year that a census would be conducted this year) and to boost participation (the same study revealed that many Americans are not inclined to participate in the census because of distrust toward government). Television, radio, print and billboards ads now appear in 17 different languages in targeted media markets; they address important issues such as government funding, political representation, the confidentiality of the census and the call to civic duty.

In addition, the Bureau will hire about 500,000 people on a part-time basis to fill close to 860,000 jobs, making this census the largest peacetime mobilization by the federal government ever. And the Bureau has entered into numerous local partnership agreements on the basis that local jurisdictions know best how to make the census work in their particular area. 520 local census offices have been opened. Multilingual questionnaires are being printed and multilingual enumerators are being recruited. The citizenship requirement for census enumerators is being waived (in part because the Monitoring Board urged the Bureau to do so). And Questionnaire Assistance Centers have been established to assist limited English speakers, the blind, the homeless and others who need help filling out the form.

The Bureau is committed to getting the most accurate count through traditional enumeration means. However, the Bureau, the National Academy of Sciences (Academy) others have concluded that a massive undercount in 2000 is inevitable unless statistical correction methods are used. Without it, the 1998 "dress rehearsal" in Sacramento, California showed a 6.3 percent undercount, and nearly 10 percent were African Americans and Hispanics. Such results would likely be repeated in towns and cities across America. For this reason, in 2000, official statistical data will be used to correct the initial count.

Using statistical methods

Following the 1990 Census, the Bush Administration and Congress agreed that steps were needed to avoid the same poor result in 2000. Bipartisan agreement was reached so that the Academy could work with the Bureau to develop a new method for counting everyone in 2000. Since then, three separate Academy Panels recommended combining the traditional headcount with two different versions of modern statistical methods (sampling).

First, instead of trying to actually find and count everybody who doesn't respond to census questionnaires sent by mail, the Academy recommended taking a large sample of the non-respondents in each census tract in order to account for everyone in that particular tract. And the second recommended use of sampling was to do a nation-wide post enumeration study as was done in the previous census.

The Bureau agreed with the Academy's recommendations and the General Accounting Office (GAO) also endorsed sampling as cost effective. However, based on unfounded fears that these counting methods would somehow serve to benefit Democrats and hurt Republicans, former House Speaker Newt Gingrich cried foul; the issue landed in court.

On January 25, 1999, the U. S. Supreme Court issued its decision in Clinton v. Glavin. The Court agreed with the House Republicans that current law does not allow any use of statistical sampling in order to produce the census count for apportionment. In other words, the Bureau must deliver a census count by the December 31, 2000 for apportionment based solely on traditional means of enumeration. However, the Court's decision strongly implied that the Bureau is required under existing law to use statistical methods for all other purposes, including the distribution of federal funds and for redistricting - - if feasible.

This was a balanced decision, and it is consistent with the original Republican position. Of course, Speaker Gingrich and House Republicans didn't think the Supreme Court would actually differentiate between apportionment and districting; that is exactly what the Court did.

With the Supreme Court decision on the books, Commerce Secretary Daley has determined that "sampling" or "correction" is indeed feasible. As a result, by law, state legislatures and local councils will be able to use more accurate official census data.

The Administration will release state population totals and the congressional apportionment information on December 31, 2000. The Administration will also release official population data to the states for intrastate congressional and state legislative redistricting on or before April 1, 2001.

Political debate

Democrats and Republicans continue to differ on the viability of the use of statistical methods. Most Democrats view the methodology as a means for an accurate count and as an avenue to social justice. Most Republicans maintain that Democrats are seeking to manipulate the census for political advantage. However, the Republicans have pulled away from political attacks in Washington; the statistical sampling issue has moved to the states.

Ignoring objective analysis, legislatures in Arizona, Alaska, Colorado and Kansas have passed laws prohibiting the use of the official adjusted census figures for redistricting, and resolutions opposing the use of official data have been introduced in New Jersey, Indiana and Nevada. More positively, the New York State Assembly has passed a resolution supporting congressional funding for the Bureau's Accuracy and Coverage Evaluation.

Yet, despite the Supreme Court ruling of last year, the census continues to be a subject of political debate; but the truth is that Republican allegations about fairness are unfounded much as the fears of the Motor Voter legislation were a few years ago. Census correction will not result in an avalanche of new Democratic legislative seats. As with all politics, the census is local. To be sure, Republican mapmakers will maintain considerable influence in most major states including those in the eastern region.

What you can do to make Census 2000 a success

Census 2000, regardless of today's inherent challenges and misguided allegations, will provide the best possible snapshot of the American landscape. With the Bureau's aggressive campaign to get a full headcount and with scientific correction, the census is poised to be the most accurate in history and to serve our democracy well.

Clearly the Bureau is playing the lead role Census 2000, but Washington can't make the census a success on its own; people everywhere need to become engaged in the initiative. Get involved. Talk to your friends, neighbors and colleagues about the census as a guarantor of proper funding and civil rights. Fill out your form when it comes in the mail this March. You can make Census 2000 work for you and for the greater common good as well.

Mr. Casellas is Co-Chair of the U.S. Census Monitoring Board and president of The Swarthmore Group, an investment advisory firm based in West Chester. Mr. Casellas has previously served as Chair of the U.S. Equal Employment Opportunity Commission and as Chair of the Board of Governors of the Philadelphia Bar Association.

[back]