Planning is essentially the process of analyzing people and places, and nothing informs that process more than the decennial Census. This past summer, California's newspapers were filled with the first important round of Census 2000 data – detailed information about population and housing counts, age and race profiles, and household size. This is called the "SF1" data – the Summary File 1 data, derived from the "short form" filled out by Americans in April 2000. We are about to feel the impact of a new flood of data. Already, the Census Bureau's more detailed "long form" survey data is available at the state level, and this fall it will be published for most large cities and counties in California. We will see more data released throughout 2002. This means that within about 12 months, city and county planners will have all Census 2000 data at their disposal. But Census 2000 data is not perfect, and that could affect the ability of local governments in California to analyze people and places in a meaningful way. In particular, the Census appears to contain a series of errors about two categories called "Special Places" and "Group Quarters." These errors could significantly affect basic statistics about cities and counties in California – including population and median income – and the Census Bureau appears to have no plans to correct the errors in the permanent data tables. Special Places is a catch-all category for dormitories, nursing homes, military barracks, and prisons (which are defined and tabulated as Group Quarters) and "difficult to enumerate" mobile home parks and large apartment buildings (which are tabulated with regular housing units). If the Census does not correctly account for Special Places, the mistake can skew the results significantly because they involve big blocs of residents. Relocating ‘Special Places' Most local governments did a good job in working with the Census Bureau to identify Special Places. But somewhere along the line, something went wrong. In many cases, residents of these dormitories, prisons, and other alternative living quarters got plopped in the wrong place by the Census Bureau. For example, on the University of Southern California campus, Census 2000 calculated that the population for the main campus dormitories was 246 persons – a huge drop from 2,750 in the 1990 Census. But the dormitories have not been torn down. They are still occupied and their residents completed the Census 2000 forms. At the same time, however, two adjacent off-campus blocks show large — but unlikely — jumps in population. What happened? In all likelihood, the dorms were "geocoded" incorrectly. At least these statistical errors occurred in the same city, Los Angeles. In other cases, people living in Special Places may be turning up in the wrong city altogether. Take the case of Soledad and Salinas as an example. Both are fairly poor cities in Monterey County's Salinas Valley. Soledad prison is located entirely inside the city limits of Soledad — as the Census Bureau's own TIGER mapping system shows. Yet the Census data shows only 51 people living in Group Quarters in Soledad. Meanwhile, 30 miles north, Salinas – which does not have a prison – lists 8,756 persons in "Institutionalized Quarters." Clearly, the Soledad prison population is erroneously included in the population of Salinas. Apparently, this type of problem is occurring all over the nation, although the Census Bureau is not quick to discuss it. This kind of error is a big deal to some California cities. The state went through a prison-building boom during the 1990s. Many cities deliberately annexed prisons to boost their population totals and, therefore, qualify for more state and federal funds, such as community development block grant money and state vehicle license fee revenue. The Census Bureau's Count Question Resolution Program (CQR) is underway, and if a local government provides evidence of an error, the Census Bureau will issue a "correction letter." That may correct the flow of state and federal funds. But it does not appear as if the Census plans to alter the SF1 tables already prepared and the Summary File 3 tables (SF3) for unincorporated areas, which will be released next year. Indeed, the CQR website ( specifically states that the Census "will not make corrections to the data concerning the characteristics of the population and housing inventory" and "base files for the census will remain unrevised so that none of the standard Census 2000 data products will reflect corrections." If these changes are not made, the Group Quarters location error will permanently affect other statistics, such as per capita income, density, and age pyramids, that will be accessed from the Internet with little or no warning that the statistics include this error. All this means more work for local planners. The only way local governments can find these errors is to compare Census 2000 block counts to accurate, recent local maps and lists of Special Places — a painstaking process. More data on the way The SF1 block-level data already prepared are available from several websites, including the UC Berkeley Statewide Database ( SF1 files are segmented and often so large in urban counties they may not work with desktop software, and the geographic coding, table formats and content take some getting used to. The Census Bureau's main data source, American Factfinder ( is more "conversational" and is a quick source of basic and comparative counts, but only down to the census tract level. Updates on the next round of Census releases — including the long-form sample data for cities and counties of at least 250,000 people — will be available at the C2SS website ( California planners should have access to long-form survey data this fall for 22 counties and the cities of Los Angeles, San Diego, San Jose, San Francisco, Long Beach, Fresno, Sacramento, Oakland, Santa Ana, Anaheim, and Riverside — representing most of the state's population.