For Small Area Research, Zip Codes Can't Deliver!

Zip Code Fail-Factor #1
Zip codes are fuzzy.  Most Zip codes areas don’t have specific boundaries, and their implied boundaries do not necessarily follow any features you could find on a map.  Zip codes areas are more conceptual than geographic.  When demographers attempt to overlay census data on areas defined by Zip codes, they get cranky.  Very cranky.

Zip Code Fail-Factor #2
Zip codes are wiggly.  The boundaries keep changing to meet post office needs.  Fine for them, but for us—big problem.  The count for each Zip code changes from one census to the next.  As a result, we cannot produce statistically valid time series using Zip code geographies.  No trend data for you Zipsters.

Zip Code Fail-Factor #3
Zip codes were never designed for demographic research.  They simply aren’t relevant to the type of population slicing and dicing demographers do.  We work with geographies defined by the U.S. Census.  Picture those little Russian dolls that fit one inside another, or better yet, a pallet-load of milk.  Milk fills each carton; a dozen cartons fill a crate; a dozen crates fill a pallet.  Each unit fits neatly into a larger unit.

The geographic “milk cartons” used by the U.S. Census are:
  • Census Block (definition)
  • Census Block Group (definition)
  • Census Tract (definition)
  • City/town/village (these can be as thorny as Zip Codes, because of annexations and the like)
  • County
  • State
Each nests into the next, and none overlap.  Data for each unit can be rolled up to higher levels without distortion.  Neat and tidy, not a drop spilled.

U.S. Census Bureau Faces the Fail-Factors
With its 2000 survey the U.S. Census Bureau set out to face the Zip code fail-factors head on.  They developed a new statistical entity (ZCTAsTM) to overcome the difficulties in precisely defining the land area covered by each Zip code.  Each ZCTA is an aggregation of census blocks that have the same predominant Zip code associated with their addresses. 

With this change, whole census blocks have been adjusted to resemble the Postal Service's delivery areas, so that the Census Bureau can tabulate data by ZCTA. 
Not all Zip codes find a home in the Census Bureau’s ZCTAs.  Zip codes assigned to post office boxes and government buildings, for example, are useful for mail delivery but not for population research.

Third Wave Research Gets There First
Long before the Census Bureau introduced its ZCTAs we were doing statistically rigorous small-area population research.  We solved the problem by working with population data at the Census Block Group level.  We built a proprietary transition model that translates Block Group data to Zip code boundaries, to the extent possible when working with those fuzzy wiggly Zip codes! If you should view our Zip code household counts and note a discrepancy with other sources on that Zip code, that’s why.  Starting with our 2003 data we will report ZCTAs so you can continue to find demographic data in the format about 65% of you seem to think you need. 

Until that data’s available, try a Radius Request.  It’ll cost you a little more, but the result is more accurate.  No fuzz.  No wiggles.  Just damn good data.


Resources
U.S. Census Bureau home page — www.census.gov
   Geographic Concepts: www.census.gov/geo/www/index.html
   More information on ZCTAs: www.census.gov/geo/ZCTA/zcta.html

Third Wave Research home page — www.thirdwaveresearch.com