While Congress bickers about whether to make statistical adjustments to the census, whose old-fashioned nose-counting method famously undercounts poor people and minorities, yet another problem with unadjusted enumeration tends to get overlooked: It almost certainly overcounts rich people.
Perhaps you're familiar with the statistic that the 1990 census missed 4 million (disproportionately low-income) people. This is actually a slight oversimplification. The 1990 census missed a net 4 million people, a figure arrived at by combining the 8.4 million people the census failed to count and the 4.4 million the census never should have counted in the first place. According to a study by the General Accounting Office, about 4 million of those 4.4 million overcounted people were "double counts"--people who'd already been counted once in the census and somehow got counted a second time. (The remaining 400,000 were so-called "fictitious" people: nonexistent people counted by dishonest or incompetent enumerators; people born after the census took place, hence ineligible for inclusion; people who died before the census took place; and so on.) Not a lot is known about the characteristics of people who get counted twice, as compared to the detailed knowledge we have of low-income people who don't get counted at all. But according to Nye Stevens, who oversaw the GAO report, one can make an "informed guess" that the 4 million double-counted Americans are disproportionately wealthy.
Think about it. What are the circumstances likely to get you counted twice? As Robert J. Shapiro, the Commerce undersecretary who oversees the Census Bureau, has pointed out, there are three obvious categories. One is people who own vacation homes. Mom fills out a census form sent to the Vail, Colo., condo; she forgets to tell Dad; a few months later, Dad fills out a census form back home in Beverly Farms, Mass. A second category is people who live one place in the winter and another place in the spring, summer, and fall. (Demographers call them "snowbirds.") Grandma fills out a census form sent to the apartment in Miami Beach, Fla.; she forgets to tell Grandpa; a few months later ... you get the idea. The third category is college students who don't live at home, fill out their own census forms at school, and are also tallied by their parents.
Nobody appears to know how many people fall into each of these categories. But the possibilities for such double-counting are plentiful. For example, according to the Census Bureau, there are 1,953,558 students currently living in college dorms. That suggests a potential student overcount of up to about 2 million--not counting the however-many college students who live in off-campus housing.
The Census Bureau does have some sketchy information about how the overcount affects census enumeration in a few interesting subcategories. Apparently there is an overcount--for some reason, the Bureau doesn't know how much of one--for white homeowners living in the northeast. There is also an estimated 1 percent overcount for white homeowners in mid-sized urban areas (with a standard error of 0.5 percent); and an estimated 2.1 percent overcount for white homeowners in large urban areas (with a standard error of 1.1 percent). Needless to say, as a group, white homeowners tend to be wealthier than the population at large.
The National Committee for an Effective Congress, a liberal political action group, did a fascinating study last fall calculating how undercounts and overcounts in the 1990 census affected representation within individual congressional districts. It's no surprise that places like Harlem ended up getting less representation than a strict mathematical accounting would seem to warrant. But the study also identified 22 congressional districts that, by virtue of a census overcount, ended up with more representation than they appear to deserve. The most over-represented congressional district in the United States is Oyster Bay in Long Island, N.Y., a WASP enclave that includes Teddy Roosevelt's Sagamore Hill estate. The district, served by Republican Peter King, represents 7,448 phantom residents in addition to its 572,889 actual residents.
Overall, the 22 over-represented districts are evenly divided between Democrats and Republicans (but the top half of the list is dominated by Republicans). Pennsylvania, New Jersey, and suburban New York seem especially blessed with no-show inhabitants. Here are the rankings:
[Correction 4/13: Several readers wrote in to point out that because this study was done before Election Day, three of the members Chatterbox originally cited in this graph were no longer in Congress. (Obviously, that did not affect the rankings of the over-represented districts.) Slate has now updated the chart, and all representatives cited here are now indeed sitting members of Congress.]