Occupation is one of the variables with a high allocation rate (close to 20%).
The ACS Research & Methodology pages describe allocation as: "Allocation, on the other hand, involves using statistical procedures, such as within-household or nearest neighbor matrices populated by donors, to impute for missing values"
This made me wonder:
- what other fields could be used to determine nearest neighbor matrices for occupation?
- if the allocation involves random processes, is the variability of the allocation process included in the MOE and replicate weights or otherwise quantifiable?
hi Jan-- I expect (or hope) that is a generic description of 'allocation' - and not a representation of the allocation process specifically for occupations. Many years back, I started my career in a labor market stats office. US BLS and its partners figured out years ago (!!!) how to receive and ingest text descriptions of type of work, and then narrow down the most likely SOC code. The factors that aid accurate assignment are: the specific industry of employment, education level, maybe wages from work?? I hope the smart folks at Census consult with BLS subject experts when designing coding and allocation rules for occupation.
But like you I'm not seeing this answered in online documentation.
--todd graham