Hi,
I’m curious as to why there are discrepancies between the individual and household oversample flags. Both the non-March oversamples and Hispanic oversamples are selected at the HH level, so I would expect ASECOVERH and ASECOVERP to identical.
From 1976-2020, there are 178,252 individuals coded as in the oversample, but who are coded as residing in a HH that is part of the March Basic. There are 19 individuals coded as being part of the March Basic, but in a oversample HH.
Of these, 98% of cases occur between 1976-1981, while the remainder occur through 1988. Most (97%) are children under 14.
However, more than 4500 adults are impacted – most of whom are Heads. They receive CPSIDP = 0, and are not linkable longitudinally even though other members of their HHs are linkable.
For CPSID = 19760300419500, for example, the HH is coded as march basic, as is the spouse. The head is coded as part of the oversample. The children’s codes vary. Both partners are age 34.
For 19810301311700, the HH, spouse and children are coded as march basic, the head as oversample.
Thanks!
Thank you for your question. We are looking into the source of these discrepancies and will get back to you soon.
ASECOVERP and ASECOVERH values are assigned based on the linkability of MARBASECIDP and MARBASECIDH, unique identifiers created by IPUMS CPS for merging person and household records from the ASEC to their corresponding March Basic file. In most cases, the household is designated a linkable MARBASECIDH and all individuals in the household are assigned a linkable MARBASECIDP value. In the period between 1976 and 1988, however, there are some households that contain both persons that link between the March basic monthly file and the ASEC file and those who do not. In these cases, the household is designated a linkable MARBASECIDH and all the linkable individuals are assigned a linkable MARBASECIDP value, whereas nonlinkable individuals are assigned a non-linkable MARBASECIDP.
See this household from the 1985 ASEC, below, as an example of a ‘mixed’ household in which ASECOVERP and ASECOVERH values do not match for all individuals due to the non-linkability of the MARBASECIDP and MARBASECIDH values, respectively:
For more information about the linkability of these identifiers, see this working paper about linking the CPS. Thank you for asking this question. The IPUMS CPS team is working to update the variable descriptions for ASECOVERP and ASECOVERH in order to make this more clear to users.
