art photo



Why do the counts from SF1 and SF3 differ?


SF1 is based on 100% data and SF3 is based on sample data (16%).

There are not many times you would compare the two files as SF1 does not have any of the social/economic characteristics like income, education, labor force status, place of birth, etc.

However, on the items you can compare, SF1 and SF3 should agree on population counts at the county level or higher. However, the details within a county can/will vary. For instance,comparing sex*age by counties within Michigan shows agreement across all counties for the total population counts, but the detailed age*sex counts differ. However, it is unlikely that any of these differences are significant (e.g., fall outside the confidence interval).

The Census Bureau discusses this issue rather generally here:

More detail can be found in the "Accuracy of the Data" chapter in the Technical Documentation - pages 933-955 in the following pdf.

Annotated Resources:

Direct Links:

Related Question Groups: