Tired of relying on 2000 Census data but hesitant to use ACS data due to its high margins of error? Thankfully the Census Bureau just released 3-year data for cities and places at least 20,000 persons in size. This data has a significantly lower margin of error then the normal 1-year ACS survey data.

“New data released today by the U.S. Census Bureau give more than 2,500 midsized counties, cities and towns nationwide (those with populations between 20,000 and 64,999) their first statistical “portrait” since the 2000 Census on a wide range of key socioeconomic and housing topics.

These are the first American Community Survey (ACS) estimates that combine three years of survey responses (2005-2007) to produce data. The technique makes it possible to release a new set of multiyear estimates annually for smaller geographic areas. The three-year data can produce
estimates for areas with populations as small as 20,000.”

“The 2005-2007 ACS estimates are based on three years of data collected nationwide from about 250,000 addresses per month. In addition, approximately 20,000 group quarters across the United States are surveyed each year, comprising approximately 200,000 residents annually.

The population threshold for which geographic areas have three-year estimates available is 20,000. For nation, states, counties, cities, and towns, the estimated population used to satisfy that threshold is the July
1, 2007, Census Bureau Population Estimates. For other areas, the 2005–2007 ACS three-year estimates of total population are used. For example, since the Census Bureau’s Population Estimates are not available
for school districts, three-year ACS estimates of total population are used to determine if a school district meets the 20,000 population threshold.

As is the case with all surveys, statistics from sample surveys are subject to sampling and nonsampling error.”

Excerpt quoted from Census Bureau Public News Alert issued on December 9, 2008.