I'm new to working with census data and am wondering about best practices for calculating labor force participation and unemployment at the block group level. I've been using the ACS R package to gather table data from these tables here.

My current thinking says for unemployment:

  1. Add all estimates for civilian labor force for Ages 16 and older (not including Armed Services) to get total civilian labor force (TCLF)
  2. Divide the sum of all unemployed by TCLF to get proportion unemployed

My current thinking says for labor force participation:

  1. Add all estimates for those Ages 16 and older (including Armed Services) to in labor force to get total labor force (TLF)
  2. Add all estimates for those Ages 16 and older not in labor force to get total out of labor force (TOLF)
  3. Divide the TLF by total population of block group to get proportion in labor force
  4. Divide the TOLF by total population of block group to get proportion out of labor force

Is this the best way to go about this? Like I said, I'm new to working with census data so any thoughts would be much appreciated. We are mostly interested in looking at associations between block group characteristics (employment, poverty, walkability) and health behaviors.

2 Answers 2


You can check the Census Bureau's documentation here - look at the Subject Definition file, page 63 under Employment Status:


They define the Labor Force as: "All people classified in the civilian labor force plus members of the U.S. Armed Forces". So the labor force participation rate would be the labor force (both employed, unemployed, and in the armed services) divided by the total population that is 16 years and over.

The Census defines unemployment as the number of unemployed people divided by the total civilian labor force (unemployed + employed). The civilian labor force does not include people in the armed services, nor does it include people who are not in the labor force.

Some caveats - the American Community Survey data are estimates with margins of error at a 90% confidence interval. The margins of error for block group-level data are going to be very high and may not be published for this variable at the bg level. You may want to consider using census tracts instead, which are the next summary level up in size. Since they're larger areas with more people the margins of error will be smaller.

The link to the table you referenced is from the Social Explorer - a problem with the SE is that they only provide the estimates without the margin of error. You probably should use a source that provides both values - the Census via the American Factfinder, the Missouri Census Data Center tools, or the NHGIS. Using these sources may also save you some work; in addition to the B and C tables they also include the demographic profiles and various summary tables - the S tables - where the values for unemployment and labor force participation are already computed for you. For example S2301 from the 2009-2013 ACS:



As best I can tell, ACS table B23001 does not have data down to the block group level. It's probably the case that the sample sizes are too small to meet the Census Bureau's data quality standards.

You could use Census Tracts, but even there, margins of error are often problematic. Since you plan to add sub-estimates, the MOEs may settle down a little.

It's also the case that census tracts and block groups are only included in the ACS 5-year release, which seems at odds with employment status, which seems like a somewhat volatile statistic.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.