In a previous post we talked about how the results of the 5-year (2005-2009) American Community Survey, released in December 2010, would provide demographics down to the zip code level for the first time.  Unfortunately, the Census Bureau did not provide zip code data with the current data release, and won't be including it for some time, at least until 2012-2013.

Zip codes, as defined by the U.S. Postal Service, are not actually a "geography" included in Census Bureau surveys.  Realizing there is significant interest in zip code demographics, the Census Bureau did provide Census 2000 data by "ZCTA" (Zip Code Tabulation Area), which required grouping data from small census "blocks" in order to approximate an actual zip code.  You can see how ZCTA's and zip codes compare in these example maps of Washington state. (Also see our previous blog post with a more detailed ZCTA description.)

Since zip codes frequently change in definition and area, the Census Bureau must now update ten-year old ZCTA definitions.  This is a complex task that was not completed in time for this first 5-year data release.  In the meantime, we at CDX Technologies are evaluating other options for providing ACS data in smaller geographies.  (We currently provide state, county, CBSA (Core-Based Statistical Area) and city data in our CDXZipStream ACS data feed.)  We'll keep you posted as our work proceeds.

We've had a number of questions from our clients as to why the CDXZipStream demographic feed from the American Community Survey doesn't include data down to the zip code level.  It does include data for larger geographies like state, county, and "places" (the Census Bureau's version of city, town, borough, etc.) but zip code data seems to be a glaring omission.

To understand this, we need to talk a little bit about how the American Community Survey (ACS) works. The ACS is a survey that has been performed by the U.S. Census Bureau every year since 2005; it has replaced the old "long form" that used to be part of the decennial (10-year) census.  Starting in 2010, the decennial census now only consists of the "short form" that covers basic questions about age, gender, race, and household size.  The more interesting demographic data such as income, educational attainment, and other socioeconomic factors are now being obtained exclusively through the ACS.

Each year the ACS surveys about 3 million U.S. households, and in order to get statistically valid demographics for smaller geographies like zip codes, the data must be combined over a five year period – there just isn't enough data for a single year to get an accurate picture for a zip code.  In fact, one year data is applicable only to geographies with a population of 65,000 or more, and the three year combined data included in CDXZipStream is applicable to geographies of 20,000 or more.   Since the ACS formally started in 2005, five year data (covering the period 2005 through 2009) will be released for the first time in early December 2010; it will be available in CDXZipStream within two weeks after its release.  

So what's the big deal?  The real advantage of the annual American Community Survey is that it will show a true moving picture of American demographics that could never be provided by the 10 year "snapshot" of the decennial census.  Zip code demographics from here on out will be updated annually using data from the last five surveys, and trends can be measured using real data instead of relying on error-prone models based on extrapolations of old data.  So stay tuned ... we expect demographics by zip code will be available in CDXZipStream by the end of 2010.