The Latest Attack on the Census is an Attack on All of Us

The latest fight over the Decennial Census is part of a 30 years' war over efforts to count everyone in America, including immigrants, minorities and poor people.  It's become an ongoing war, because the results carry such large consequences.  The Constitution mandates a census every decade, because the founders saw a regular, state-by-state population count as the best way to determine how many seats each state gets in the House of Representatives.  Beyond that, as Washington's role has expanded, the Census provides the data used to distribute a growing slice of federal spending among the states and their cities and counties - nearly $400 billion worth these days - and to build the baselines and updates used to evaluate the effectiveness of hundreds of federal programs. 

Until this year, the fights over the Decennial Census have focused not on who should be counted - the answer has always been everyone - but on how hard the government should try to find the one to two percent of us who are often overlooked.   It matters, because people who don't return their Census forms and then avoid Census workers trying to follow up - the Bureau calls them the "undercounted" - are predominantly poor minorities, recent immigrants, and American Indians.  So, they're not distributed randomly across the states but rather concentrated in certain places - and this undercount costs those places part of their fair share of federal funds for roads, schools, medical care, parks and other things based on population.  If the count were accurate, a number of big cities and states might be a little less financially strapped. 

I follow all of this pretty closely, because as Under Secretary of Commerce in the late 1990s, I oversaw the Census Bureau planning and operations for the 2000 Census.  The fight that time was over our plans to use a huge sample - some 1 million households - to find out exactly where and to what extent the undercounts happen, and then use that information to adjust the head count and make the final results more accurate.  That was just what the National Academy of Sciences had recommended for the 1990 Census, which the first Bush White House rejected.  When the undercount grew worse that time, the Academy came back with the same recommendation for 2000.  This time, the Clinton administration approved, and the Census Bureau did it.  But between the counting and the reporting, George W. Bush took office.  The sample was discarded, and the undercount grew even worse. 

The opposition to sampling in 2000 certainly seemed to be motivated by purely political concerns that counting all minorities would cost them federal funding or even seats in Congress.  But that opposition wasn't completely shameless: They balanced their attacks on sampling with support for spending as much as we asked for to assemble the largest workforce of census counters in history and mount major advertising and civic outreach campaigns targeted to communities with large undercounts. 

It will be worse this time.  The 2010 Census planned by the Bush administration has no sample, and it's too late now for the Obama team to design and carry out one in time.  It's also almost certain that the undercount will be even larger: The numbers of recent immigrants are way up, and the advertising campaigns aimed at minorities and the outreach to civic groups have been scaled back. 

But now it's getting truly ugly.  Senator Bob Bennett, backed by the Wall Street Journal and right-wing cable TV and radio, has proposed to use the census to identify undocumented people, who would then be deliberately excluded from the count.   In more than two centuries of the U.S. Census, it has always counted whoever is physically here - "inhabitants" in the term used in the first census of 1790 - regardless of their citizenship or other legal status.  One reason is that everyone is protected by the law, so everyone should be counted in determining how many seats a state gets to write those laws.  And whether or not someone has citizenship or residency papers, they still put claims on public services which the funding for those services should reflect.    

The political and social implications of Bennett's radical idea are enormous.  California, for example, may have as many as 4 or 5 million undocumented inhabitants - exclude them and the state could lose perhaps a half-dozen seats in Congress and tens of billions of dollars in federal funds.  Texas and other states with large Hispanic populations would lose seats and funding as well.

This change also could destroy the Census process, with incalculable costs for everyone.   The Census doesn't collect any information beyond people's demographic characteristics - no names or data about their legal circumstances - and it's so fastidious about people's confidentiality that it won't share any specific data with police, the FBI, or anybody.  In 2000, for example, a form came back with a threat against the president scrawled across it - and the Census Bureau refused Secret Service demands to share the address.  (It turned out to not matter, since the respondent was safely tucked away in a state prison.)  The Bureau also knows that if the census process goes beyond demographics, tens of millions of people may assume that their information might be turned over to other parts of government, and the undercount would skyrocket.  People would begin to worry that the IRS might compare the number of people counted in a household against the number of dependents claimed,  or that child welfare services might discover that somebody is taking care of a cousin's child and disapprove of it, or, most obviously, that the Immigration and Naturalization Service would come knocking. 

The Census is the world's largest scientific exercise and provides the basis not only to allocate federal funds and seats in Congress, but also to evaluate the effectiveness of countless federal, state and local programs.  All of that would be at risk if the latest expression of anti-immigrant bias were ever to take hold of the decennial Census.