When i crosstab FERTYR*YNGCH for women ages 15-50, I find about 1900 cases in the 2015 ACS where a woman reports a yngch>1 in the home, but FERTYR NE 2.
I know some small share of these might be cases where the infant is adopted so the coding is OK, but looking through the HH records, that doesn’t seem to be that common.
Can you give me any advice about how I can get this coding to match up appropriately? There are many cases where it seems pretty clear that FERTYR should probably=2, but some cases are less obvious, so I hesitate to implement a global edit.
There are a few things to consider when comparing FERTYR and YNGCH. First, FERTYR represents a direct response to a Questionnaire item while YNGCH is constructed by IPUMS-USA based on Family Interrelationship linking. Since the family interrelationship linking methodology was developed for use with the Decennial Censuses, FERTYR is not considered in the construction of MOMLOC, so it is possible that IPUMS-USA is incorrectly identifying mothers of newborns in some of these households. It is also worth mentioning that the Census Bureau does allocate this variable if it is missing. You can identify allocated values with the QFERTYR.
Unfortunately, I cannot provide a simple solution for when to assume that the FERTYR value or the YNGCH value is more accurate. You may wish to privileged un-allocated FERTYR values since they represent direct responses. But if YNGCH and FERTYR conflict and FERTYR was allocated, it may be worth while to take a closer look at the household and see which case seems more logical.