On OCC2010, for years before 2003, why does OCC2010 show about 344 separate occupations when OCC shows 500?


OCC2010 does not seem especially satisfactory for years before 2003. Go to the case count view. Look at the first occupation OCC2010=0010. In January 2003 there are 842 cases. In Dec 2002 there are all of 12 cases. What happened? This seems fishy when I examine OCC for the 2002 year. The documentation for OCC tells me that in these years OCC is based on the 1992-2002 codes and these codes reveal that 003 is legislators and 004 is chief executives. So these occupations were being enumerated in 2002 but OCC2010 suggests there are almost nil chief executives and legislators before 2003.


Though IPUMS-CPS strives to maximize consistency over time, certain variables (especially occupation and industry coding) are very difficult to harmonize perfectly. What you are seeing is simply a result of coding schemes changing over time. The IPUMS-CPS team used this BLS working paper to guide the harmonization process for these variables and we believe these schemes represent the maximum level of comparability while still preserving the as much detail as possible. To improve the comparability of OCC2010 over this period you can sacrifice some detail and pool similar occupations. The OCC2010 variable description includes a recommended grouping of occupations into broader categories.

I hope this helps.