I am writing a re-districting application where I want to be able to create US voting districts based on geographic coordinate population data. In a perfect world, I would have the population for each house and the house's geographic coordinates. I am guessing that per house data is not likely to be available. What is the lowest level available? Blocks? Block groups?

I located a US census website which contains population data for block groups. Is this the finest granularity available for 2010 census data?

My end goal is to implement the shortest splitline algorithm for districting with 2010 census data. My intention is to slice up polygons using the shortest splitline according to the best population numbers relating to latitude/longitude coordinates I can find for people in the polygon.

  • By "geographic coordinate population data," do you mean you need to use points? This is important because Blocks and Block Groups are polygons. If points are needed, the TIGER shapefiles could be converted to points using their centroid attributes.
    – dmahr
    Commented May 22, 2012 at 20:37
  • Have you looked at the "at a glance" summary of Census 2010 data? census.gov/population/www/cen2010/glance.
    – whuber
    Commented May 22, 2012 at 20:41

3 Answers 3


For privacy reasons, personally-identifying house-by-house data from the Decennial Census are not available until 72 years after being recorded. Census Blocks are the highest spatial resolution geography for which more recent Decennial Census data are available.

If you only want Census Block populations for a couple of counties, use the Census Factfinder. Choose the table you want (P1 has basic population count) and the desired geography (e.g. All Blocks within Middlesex County, Massachusetts). You'll probably want to download it as a .csv for further processing.

If you want Census Block populations for a much larger swath of the United States, you are in for a world of hurt. You need to jump through a lot of hoops in order to overcome filesize limitations in most database software. The Summary File 1 page has more information.

  • 1
    I recently spent some time mashing these data sources together and found postgresql/postgis to be an excellent way of handling these datasets as there are no filesize limitations.
    – djq
    Commented May 22, 2012 at 21:04

As noted by dmahr, the lowest level of data available from the Census Bureau is block level data. While it might initially sound appealing to have data which is precise to the housing unit level, this actually will only complicate production of the end product.

In the picture I have included below, the blue box demonstrates the area of a typical census block. In urban areas, typically the area of a census block is bounded on all sides by a street, railroad tracks, a water feature or something which is identifiable. There are, like almost every rule, exceptions.

When creating software for the purpose of reapportionment, you will want to rely on the same data States & local government use when conducting their reapportionment. The data used by these entities is referred to as P.L. 94-171 data and for state level can be found here: http://www.census.gov/rdo/data/2010_census_redistricting_data_pl_94-171_summary_files.html

Additional relationship files for that data can be found here: http://www.census.gov/cgi-bin/geo/shapefiles2010/main

As noted by celenius, I as well highly recommend dumping the data into either a PostGreSQL/Postgis or MySQL database.

Example of a Census block


You might check to see if you can legally use the same mapservices that are used by Esri Redistricting Online in an application that you develop.

These mapservices provide data down to the block level.

For more details see this thread.

Your Answer

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

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