Correspondence between Puma 1990 and Puma 2000

I am trying to delete the PUMAs affected by Hurricane Ketrina in Louisiana --01801, 01802, and 01905 from my analysis. I would also like to delete the same areas in PUMA 1990. How should I match PUMAs in 1990 5% Census and PUMAs in 2000 5% Census in order to delete the correspondent areas in 1990 5% Census?

For example, Lousiana PUMA 1801, which is coded as 221801. (Statefip+PUMA) in 2000.

How should I match the exact puma in 1990? What would be the puma code in 1990?

For PUMA 1801,1802 (in year 2000, Louisiana):

I checked the puma map in 1990 and 2000. The 1990 map did not specify the areas where are coded as 1901-1904 in Louisiana. I do not know how should I relate it to the 2000 map (Louisiana) where 1801-1804 are divided clearly in 4 parts.

For PUMA 1905 (in year 2000, Louisiana)

In 2000 map, it is PLAQUEMINES. In 1990 map, I found 02300 might be similar to the area. But there are two other areas also coded as 02300.

I looked at the 1990 and 2000 5% PUMA shapefiles (available for 1990and 2000) to determine the relationships between 2000 PUMAs 01801, 01802, and 01905. The relationships are straightforward for 01801 and 01802.

2000, 1990

01801, 01901

01802, 01902

01803, 01903

01804, 01904

The relationship for 01905 is substantially more complicated. 01905 contains all of 02300 (1990) and the southern half of 02200. Unfortunately, if you delete 02200 from your data, you can’t quite match 01905.

Based on my inspection of the data, the following sets of PUMAs (one set for 2000 and one set for 1990) must be merged together to make a unit with a consistent footprint in both 1990 and 2000:





I know you’ll be losing more PUMAs that you probably want, but I think this is the safest way to remove all cases in both 1990 and 2000 for the exact same geographic area.

I’ve attached a screenshot of a map to show you the relationship. The red lines are 2000 PUMAs and the blue lines are 1990 PUMAs.