art photo



I want to create an annual file with census-type characteristics of counties in California. What is the best source for this? This needs to be current, but I want it updated every year.


The Census Bureau has a relatively new data product called the American Community Survey (ACS). It provides long-form data on an annual basis, assuming the unit of geography is large enough (65,000+). So, in your case, they provide a 3-year average for counties between 20,000 and 64,999) and 5-year averages for counties less than 20,000.

Starting this fall (2010), the Census Bureau will be releasing the first wave of 5-year data, which means from now on you can have annual updates for all counties in California. As a note, the recommendation is that you do not mix and match types of data (1-year; 3-year; and 5-year). So even though Los Angeles county (and in fact, most California counties) is/are large enough for annual data, you should be using 5-year data if you want a consistent series of county data.

One can make comparisons across counties for statistical significance, e.g., Los Angeles vs Fresno on a yearly basis, but with the 5-year data (or 3-year data) one cannot make year to year comparisons for a county. That is because the 2007 year (2006-2008) has a great deal of overlap with 2008 (2007-2009).

One other solution to get annual data for all California counties is to use PUMAs to produce pseudo county data for the non-populous counties. PUMAs are units of geography that add up to 100,000 in population. Using a PUMA to County crosswalk, one can produce values for the small counties. As one would expect all counties in the PUMA end up with the same value. However, this allows you to produce timely, annual data rather than relying on a 5-year average.

The following link will address building county level statistics from Public Use Microdata Areas.

Creating County-level Statistics from Public Use Microdata Areas

Related Question Groups: