I am trying to use the CPSIDP variable to match individual observations across months but, for some reason, I am unable to do so across certain months in 1995. For example, I sorted my data by CPSIDP, Year and Month and created a variable that is 1 at month t if two observations have the same CPSIDP in month t and t+1. But, for example, between May 1995 and June 1995, I obtain no links (this indicator variable is equal to 0 for all observations in May 1995). This is weird because Drew et al says that you should get 95,000-100,000 matches across months back to at least 1994. Any help is appreciated!
The Census Bureau updated their sub-state geographic identifiers in September 1995, excluding them completely from the June-August 1995 period. As a result, they intentionally prevented the matching of households from samples with old geographic identifiers to samples with the new geographic identifiers. Specifically, it should be possible to link within the following three time periods, but not across:
-
Prior to June 1995
-
June-August 1995
-
After August 1995
Hope this helps.
Hi,
I am trying to link June 1995 records to July and August 1995 using cpsidp, but it seems I can only get ~4% of records to link to either month; and of the linked records, age, sex, and/or race in June and July/August disagree ~92% of the time. Any help would be much appreciated, thanks!
I was able to replicate what you describe, but unfortunately I don’t have an explanation for you. The changes mentioned by Tim are described in the October 1995 CPS documentation (see page 50): https://cps.ipums.org/cps/resources/codebooks/cpsoct95.pdf
Why no (or very few) linkages are possible within June-August 1995 is unclear, but the IPUMS CPS team will look into this further. Thanks for bringing this to our attention!
Dear Matthew,
Has this issue been resolved in any way? To the best of your knowledge, is there any way to link respondents from August 1995 to those in September?
Thank you!
To the best of my knowledge there is no way to link between August and September of 1995 because of changes to the linking keys in the public use files released for the original source data. I have included an excerpt from the October 1995 codebook below and emphasized the most relevant text. I will note that I have elevated the request to look into the June to August 1995 links with the IPUMS CPS team, but don’t have a definitive answer about what might be happening (and that doesn’t seem relevant to your immediate question about linking from August to September).
Effective September 1995, a number of revisions were made to the CPS public use files. Most were related to the recent phase-in of a new sample based on the results of the 1990 Census. This phase-in was completed in June 1995. Part of this changeover was the use of new metropolitan area definitions based on the results of the 1990 Decennial Census in selecting the new sample. As such, beginning in September 1995, metropolitan area definitions effective June 30, 1993 will be identified subject to confidentiality restrictions on the CPS public use files. The new variables and their locations are given below. Concurrent with this revision, several other changes were made to the CPS public use files. The most important of these was the creation of a new set of household identification numbers for the September forward files. Bureau of the Census confidentiality restrictions require that we preclude the possibility of matching any households from data files before and after the September 1995 date. In conjunction with this, the Bureau revised its sample household numbering scheme. These two considerations resulted in the creation of a 15 character household identification number. The location of this number is now in characters 1-15 (previously 1-12) of all non-March files and characters 344-358 (previously 320-331) on the March files.