I found that there are around 56% of county codes are missed after 1996 due to a confidential reason. Is it possible to get a unrestricted verson of data to reduce the missing rate of county?
The only potential way around the confidentiality restrictions in public use data is through approval and use of restricted use data. Restricted CPS data is available in the Census Bureau Research Data Centers.
If I may follow up on this question: Is the restriction due to confidentiality reasons random or nonrandom? I ask because there are still other cases with the county fips code. Thanks.
The selection is non-random. In general, counties with lower levels of population are not identifiable and counties with larger populations are identifiable.