Science 15 October 2010: Vol. 330. no. 6002, pp. 310 - 312
by Sam Kean
The 2010 census will cost U.S. taxpayers $13 billion, making it the most expensive census in the world and prompting policymakers to ask if there are cheaper and better ways for the Census Bureau to do its job. For instance, why not tap into the vast amount of digital data on U.S. residents already being collected by various state and federal agencies and sitting in computers? But using digital data is not as straightforward as it would appear, and the Census Bureau must be very careful if it decides to rely on it in the future. For starters, a census can't err on the side of counting someone multiple times when the same person's name appears in different databases. Data from government agencies also contain more mistakes about individual characteristics—age, race, sex, and so on—than a census can tolerate. In addition, few databases come close to delivering the universal coverage the decennial census demands. What's more, there are no obvious technical fixes to these problems, in part because of the dearth of research on the topic.
[Complete article available, for a fee, HERE.]