The Census Bureau's Bad Estimates by Andrew Beveridge, May 01, 2002 TwitterÂ Last year, when the count from the 2000 U.S. Census revealed that there were eight million people living in New York City, many demographers were shocked. The number was almost 600,000 more than they had been led to expect. That is because the United States Census Bureau had been making really bad estimates of the city's population every year for the whole previous decade. The bureau had estimated a growth of about 10,000 in the city each year; the actual growth was closer to 60,000 a year. This was not a trivial error. The estimates were used by federal agencies such as the Department of Housing and Urban Development to determine how much federal funds the city would get. When the actual count from the 2000 census proved that the bureau's estimates had been way too low for way too long, it would have been reasonable to assume that an embarrassed bureau would take steps to make sure that the estimates for the next ten years (until the next actual count) would be more accurate. That is not what has happened, if the first new estimates are any indication. The bureau still does not know how to estimate the population in New York City. WHY AN ESTIMATE In Canada and some other industrialized democracies a census -- an actual count of the nation's population -- occurs every five years. In the United States, it is every ten years. Yet, there is a staggering demand for data on population change that cannot wait a decade to be updated. So the U.S. Census Bureau developed a system that tries to estimate the population change by using a mathematical model or formula. The system the bureau developed uses births and deaths (recorded by state agencies), net domestic migration (estimated from people who change their income tax filing location) and net international migration (estimated from a variety of sources, including data from the Immigration and Naturalization Service on new "Green Card" holders). They also use information on the movement of Federal employees, including the armed forces. Using the official population count from the decennial Census, the bureau applies the formula to every county in the United States -- including the five counties that make up New York City. A FLAWED SYSTEM It is clear that the system has seriously flaws. The bureau has already acknowledged this by putting changes into place for the national estimates. And, the bureau's population division posted "revised estimates" of previous decade's growth in New York City, taking into account all the growth they missed during the 1990s. http://eire.census.gov/popest/data/counties.php But then, as astonishing as this sounds, when they issued the estimates for the growth in the city after the 2000 Census, they went back to their old discredited figure -- 10,000 per year. They even declared that the borough of Queens is losing population. If anyone has recently driven around Queens, they know that this cannot be true. And indeed, recent conversations with Joe Salvo, Director of Population at New York's City Planning Department and Warren Brown, of Cornell, who does demographic analysis for New York State, have convinced me that the census bureau's estimates are simply wrong. Queens did not lose population; New York City probably grew a lot more than the Census Bureau reported. Privately, some demographers at the bureau admit the figures never should have been released. WRONG BUT OFFICIAL Nevertheless, whether or not they are wrong, the new estimates are official, continuing to determine some federal funding, and used in calculating various statistics about the city, such as the rate of disease and the rate of crime. To see just how implausible the estimates are, you can look at the accompanying chart, which presents the new estimates for 2001, the revised estimates for the 1990s, just issued, along with the old ones that came out year to year during that decade. Note how flat the original line was, compared to the line computed using Census count in 2000. Note that after 2000, the curve is once again flat. In short, the Census system is now estimating a growth rate for the 2000s even less than their much-too-low estimates for the 1990s. In the 1990s, New York City and New York State sued because the census count itself was inaccurate. Only Fernando Ferrer is suing now because of an inaccurate estimate, and it is so far to no avail. It appears that with the current estimate system in place, the bureau will under estimate and under count New York City for years to come. This will cost us greatly in both lost pride and lost billions of dollars. The United States Census Bureau stil l seems unwilling to report New York City's real growth. Andrew A. Beveridge has taught sociology at Queens College since 1981, done demographic analyses for the New York Times since 1993, and provides expert testimony on a range of cases, including housing discrimination. The opinions expressed are his alone. Tweet Editor's Choice Everything Dot NYC Council 2.0: Rules Reform Outlines Next Steps in Open Data Overdue & Over Budget, Capital Projects Roll On Comments Comment(s) The comments section is provided as a free service to our readers. Gotham Gazette's editors reserve the right to delete any comments. Some reasons why comments might get deleted: inappropriate or offensive content, off-topic remarks or spam.