Census Update: Shapefiles, ACS, Estimates
I’m in Boston at the Association of American Geographers (AAG) annual conference this week, and attended a great series that explored what the Census Bureau is currently up to. Here are some hi-lites:
The Bureau is now providing the TIGER line files in shapefile format! Before, it was only possible to get generalized cartographic boundary files directly from the bureau in shapefile format. Now, you can get the boundaries in their original detail from a public domain source. Includes 2000 census geography plus some updates for 2007 for states, counties, metros, places, zips, districts, pumas, and more. Currently, it does not include tracts, block groups, or blocks.
The 2008 release of the American Community Survey (ACS) will include two datasets. There will be the annual numbers for geographies that have over 65,000 people, and for the first time there will be three year averages for geographies that have over 20,000 people. In each succeeding year, this average will be recalculated by adding in the most recent year and dropping the oldest one. Data for geographies with less than 20,000 people will become available in 2011 and will be based on five year averages. The good news is that, from that point forward, data will be available for all areas every single year. The bad news is that the long form (the one in six sample of households taken in the decennial census) is being discontinued and will not be conducted in 2010. Census 2010 will consist solely of the short form questions (the 100% count that covers the basic demographic variables). The ACS will serve as the replacement to the long form, but in most cases the data will not be suitable for making historical comparisons (i.e. comparing 2010 to 2000).
Bureau reps gave an overview of their Population Estimates program. Unlike the ACS which is survey based, estimates are calculated using a cohort component analysis that accounts for births, deaths, and migration each year. Estimates are calculated nationally and at the county level. The county numbers are used to create estimates for each state, which are then adjusted to fit national numbers. Data is available for total population, race, age (broken down by gender for each year at the national level and for five year groups below that) and housing units. Some data is also available for metropolitan areas (which are county based) and county subdivisions (for total population only).
The Bureau gave an overview of Dataferret, which is a tool for data power users. It is available in two versions, as download-able software or as a browser-based JAVA applet, and allows users to gather and process data from several different government sources (unlike the Amerivan Factfinder, which focuses solely on downloading census data).
Finally, things are ramping up for the 2010 Decennial Census. The bureau is updating its master address files and has almost finished recalibrating the TIGER files for each county, so that boundaries are precise within a maximum limit of 70 meters.