I would like to merge two datasets, one using ACS data (2008-2016) and the other with ASEC CPS data (2008-2016). The ‘serial’ variable in both datasets that is supposed to link the two does not seem to work (only matches about 10% and demographic variables don’t match), even when uniquely merged alongside ‘pernum’ and ‘year’.
How should I merge these two datasets? I really only need a couple of variables from the CPS dataset, but at this point I can’t rely on any analysis involving merged observations.
I’m using STATA for analysis.
Any advice would be appreciated!
While what you are proposing to do would certainly be a wonderful asset, it is not possible to merge ASEC and ACS data on a household or person level. While both IPUMS CPS and IPUMS USA include the SERIAL variable, they are only the same in that it is an identifying number unique to each household in a given survey year (or in some cases in IPUMS CPS, a given survey month and year).
The ASEC and ASC surveys are two separate surveys. While it is possible that in a given year a person does complete both, it is not true that they will have the same SERIAL value in each sample. The matches you have between the two are based on pure coincidence and should not be considered the same households or people.
1 Like