Is conspuma really consistent over the years?

I have constructed the conspuma for 1990 and 2000. However, I have found that the conspuma code is not consistent for both years. For example, Los Angeles county is conspuma 47 in 2000 and conspuma 46 in 1990 and Los Angeles city is conspuma 46 for 2000 and conspuma 47 for 1990.

I am doing an analysis at the conspuma level, taking into account the cities/counties which each conspuma contains.

Is there any reason for this situation?

Thank you,


You can find the CONSPUMA component file for these years here, as well as the detailed composition files with information on which geographic area each component represents. Looking at the component file, CONSPUMA code 46 consistently contains the LA county components and code 47 consistently contains the LA city components. Be careful when reading the detailed composition file for the 2000 PUMAs, as it lists the Los Angeles city components under a Los Angeles county heading.

I also checked the data to see if there appeared to be any inconsistencies in population for these CONSPUMAs. CONSPUMA code 46 has approximately 5,000,000 weighted observations in both 1990 and 2000, while CONSPUMA code 47 remains around 3,000,000 weighted observations in both years.

Hope this helps.