PMA - Not in universe codes don't match skip patterns?

Hello - I am working with the SDP (family planning) data from PMA and have a question about Not In Universe (NIU) coding. In the SDP questionnaire, it appears that facilities were asked about provision of contraceptive methods (i.e. variables conprov, depoprov, etc.) if the facility reported offering FP services (i.e. variable fpoffered is “yes.”) However, there are facilities that seem to have reported yes to offering FP services but are coded as NIU for contraceptive provision variables. Can you clarify the skip patterns/relevance of these contraceptive provision variables? Thank you!

Hi Claire, could you be more specific about the survey year, country, and variable name(s)?

Hi Devon,
Yes, apologies – here is a crosstab by country and year of facilities that offer fp services (fpoffered==1) but are coded as NIU for provision of male condoms (as an example):

tab country year if conprov==99 & fpoffered==1

pma country | 2013 2014 2015 2016 | Total

Hi Claire,

In the original data files, blank values in variables could be because of skip logic, a non-interview, or data processing at PMA. At IPUMS, we do our best to identify missing cases that are the cause of non-interviews and code them as such, and the remainder we can mostly attribute to skip logic. There are certainly instances where there are blank values for reasons that are not clear to us. We have access to the metadata that determined the skip patterns, so we know the universe is correct. Any cases coded as NIU that do not fit the skip pattern are unexplained blanks from the original data.

Thanks so much for the clarification, Devon.