Why so many missing values for population by regency in Indonesia in the 1995 census round?

Dear Bob,

I used the function “Analyze data online” to get population counts for Indonesia by regency. For many regencies, data for 1995 is missing (and thus indicated as “0”). I am aware that “0” does not necessarily mean “missing” all the time: It can also mean that at the respective census round, the regency did not exist yet (as it split off another regency at some point after that census year). However, this is not the answer/solution to the problem for the 1995 census round: This data simply appears to be missing, as data is available for the 1990 census round. I hope I expressed myself clearly, and hope you have the answer.

Dear Paul,

Yes, your question is clear.

Unfortunately for the 1995 sample there are no data for any regencies in the provinces of Kalimantan, Sulawesi, Maluku and Papua, as noted in the comparability discussion for GEO2B_ID (see link below):

"Some provinces and their associated regencies are missing altogether from the data in specific years:

  • Aceh in 2005
  • All provinces of Kalimantan, Sulawesi, Maluku and Papua in 1995
  • East Timor is only included from 1985 to 1995"

We make every effort to obtain the best copies of each dataset. In the case of ID 1995, this is the best that we could do.

Sorry.

Bob.

Hi Bob,

Thanks for the clarification. I have a follow-up question to this thread. I notice that the IPUMS subset of SUPAS 1995 is around 30% smaller than the SUPAS95 data used by other scholars in terms of the number of observations. I am wondering whether the data is missing because IPUMS didn’t get the full version of the original data, or if the original data was missing for these provinces and IPUMS further trim the data for confidentiality concerns. Thanks in advance!

The data that was shared with IPUMS was missing all observations from provinces in Kalimantan, Sulawesi, Maluku, and Papua. This was not the result of any additional trimming on the part of IPUMS. The original survey included 206,484 households, but on account of these missing observations, there are currently only 166,033 households in the sample. The documentation that was shared does not mention these missing observations, but it’s unclear at what stage these observations were lost. I’d really appreciate it if you could point me to the scholars you mention who use microdata from the 1995 SUPAS from these provinces. The IPUMS International team is always looking for opportunities to add data and this may be a good opportunity to update the sample.