Ask A Cartographer

Mapping population percent differences between years

February 02 2009 | 1 comment

My task is to create a series of maps showing percent differences in county populations between 1960 and 2000 as well as 1960 and 2007.

I organized census.gov data of decenniel populations for all U.S. counties from 1900 to 1990 in a table, and then in another table I organized the estimated annual populations of all U.S. counties from 2000 to 2007.

Now I want to show the percent difference in population for counties from 1960 to 2007, for example. The three step calculation would be 2007Pop-1960Pop/1960Pop. The only way so far I have been able to figure out to do this is create a new table of fields with the difference values for example 2007Pop-1960Pop for each county, join that difference table to the decenniel population table and normalize the 2007Pop-1960Pop field with the 1960Pop field to get the percentage.

Is there a more efficient method to calculate the differences in population between two years without having to create a table with the values of the differences?

Mapping Center Answer:

I would approach this a bit differently.  I would want to have a single feature class that had attributes for population in 1960, 2000, and 2007.  Then I would add attributes for PC60_00, PC60_07, and PC00_07; you could calculate these latter attributes as you've described above. 

The real issue here depends on which counties you need.  The main issue is that county boundaries and names are not consistent from one census to another because we add new counties, rename them, etc.  I would recommend checking out the National Historical Geographic Information System website.   For instance, Alaska is very different in 1960 than for 2000; or La Paz County Arizona, didn't exist in 1960.  So, you need a plan for aggregating or splitting data values to translate from 1960 to 2000.   On the NHGIS website there is a document that describes county data availability:

http://www.nhgis.org/nhgis-files/County%20availability.pdf

That should get you started, and hopefully you're working in an area where no changes have occurred.

In the meantime, the way to get your population fields appended would be to use a join, but not an attribute join because county names are not unique (lots of Adams counties in the U.S.). Instead, convert your 2000 and 2007 data to points using the Feature to Point tool (inside Option = checked) and then use the Spatial Join tool to join these points to your 1960 data (this presumes that there may be more counties today, because some have split, and you will be able to select both modern counties to their 1960 county.) 

Calculation posted by Simon Seamount on Feb 3 2009 4:50AM
I am using FIPS codes to connect the county population tables to the county shapefile.

I would prefer to use a dynamic calculation method rather than have to create more tables loaded with data of differences between the years. This map will be published as a map service. Is there no way to generate these values with a script?

If you would like to post a comment, please login.

Contact Us | Legal | Privacy |