In the variable description for PEDADTYP (and PEMOMTYP), there is the below paragraph.

Is this still the case, even now after the IPUMS restructuring where you linked kids to the cohabiting spouses of their parents?

Or is it now the case that users should be able to use PEMOMTYP/PEDADTYP as-is?

“These variables are reported as is and may need additional cleaning by researchers. IPUMS-CPS plans to test these variables in the future and will revise them if necessary. In particular, when a child resides with one biological parent and that parent’s cohabiting partner, the child is usually reported to live with only one parent. Researchers can use the information on the mother’s cohabiting partner (PECOHAB) to identify those cohabiting stepfathers who are not included in PELNDAD and PEDADTYP. Married stepfathers are always identified by PELNDAD and PEDADTYP.”

Note that PEMOMTYP, PEDADTYP, and PECOHAB are all self-reported variables of family-interrelationships and are not the IPUMS defined family interrelationship variables. So, the paragraph you’ve noted is still the case.

Ok thanks.

So does that mean that for any case where you/IPUMS assigned a cohabiting partner that is different than what was listed at PECOHAB, you would have changed the values for PEDADTYP and PEMOMTYP to ‘0’?

And also one more (last?) question on this: do you have any sense on how cohabiting families in which you changed the partner linkages would differ from those where the original linkages remained (which therefore have data for PEMOMTYP and PEDADTYP)?

There are obviously a lot of cases of this, but not sure if there is reason to believe they would differ from the unchanged cases.

Thanks again!

It is best to think of the family interrelationship variables generated by IPUMS programming (POPLOC, MOMLOC, SPLOC) as independent from the family interrelationship variables with self-reported information delivered by the Census (PEMOMTYP, PEDADTYP, and PECOHAB). Therefore, while there many be many cases where this information is identical, there may also be many cases where the information is different.

I hope this helps.

Ok thank you!

So is there any way for me to figure out which cases IPUMS may have changed the pointers for cohabiting partners?

it seems the self-reported data is all referencing line-numbers while sploc, momloc,poploc reference pernum, so i’m not sure how to go about it.

(This is in reference to 2017 data, and there doesn’t seem to be a file with old type of interrelationship variables for any year after 2015).

Maybe there is a better way of doing what I want to do:

All I ultimately want to do is to determine if the values for PEMOMTYP and PEDADTYP are accurate, given the values of MOMLOC and POPLOC assigned to kids.

In particular, I’m trying to figure out if the cases where there is a value>0 for POPLOC BUT PEDADTYP=0 mean that:

  1. the kid didnt identify his mom’s partner as a ‘dad’ of any kind, or

  2. if the value was changed to 0 as a result of the dad linkage being changed to a different man in the household with the IPUMS recode?

You are correct about the availablilty of the “old” pointer variables. We have files for pre-2016 only samples here. I’m not sure if it is possible to answer your questions with CPS data. This is because the questionnare may not have been actually filled out by a child. The reference person in the household often will be responding to the survey on behalf of children.

So more generally, beyond any issues related to changing pointers, how should a value of PEDADTYP=0 be interpreted?

(I know your documentation says the partner of a biological mom might be listed as PEDADTYP=0; is it safe to assume all PEDADTYP=0 would fall into this category/)

PEDADTYP==0 indicates that the child has no father present within the household. While it is true that it is often the case that the partner of a biological mother may be listed as PEDADTYP==0, such cases could represent other family interrelationships as well. It is important to remember that the validity of these variables relies on what was reported by respondents. Therefore, there may be seeming inconsistencies based on how respondents understood and responded to the question.

So if there is a value of ‘0’ for PEDADTYP, does that mean it was generated by the respondent, or in some cases did you all change the PEDADTYP value to ‘0’ if you changed the ‘dad’ pointer with the new coding rules for interrelationships?

for example, if a kid in the initial CPS data had a value of poploc=2 and a value of PEDADTYP=1, if IPUMS changed the poploc value to 3 with the new coding rules, what would IPUMS do with the PEDADTYP variable in that case?

(would you leave it as 1 or would you change to 0 or to some other value?)

The PEDADTYP and PEDMOMTYP variables are reported in IPUMS as reported by the respondent. They are independent from the family interrelationshp pointer variables from IPUMS programming. Therefore, we do not make any changes to PEDADTYP or PEDMOMTYP based on POPLOC and MOMLOC values.

Ok that it helpful.

What type of reasons would you change the PEDADTYP values though?

In particular, there seem to be a fair amount where they go from a value to blank (QPEDADTYP=50); or change from one value/DK/BLANK/REFUSED to another value.

The values of QPEDADTYP are also presented as provided by the Census Bureau. It looks, however, that these are logical edits made for observations with clearly contradictory values. For example, it seems that for most cases where QPEDADTYP==50 there is actually no possible father recorded in the household. Therefore, although a value was given PELNDAD==0 and therefore, PEDADTYP==0.

Ah ok. i had thought those were your/IPUMS edits.

thanks yet again for the help!!!