The DHS Program User Forum - RDF feed
https://userforum.dhsprogram.com/index.php
Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=17864&th=8255#msg_17864
Can you please clarify which provinces/regions to exclude from the 2013-14 and 2017-18 Pakistan DHS survey rounds for comparability between the two surveys? I'm seeing conflicting information on this.
In the PDHS 2018-18 final report, the footnote on p. 2 reads:
Quote:
The 2017-18 PDHS presents national data totals for Pakistan that exclude Azad Jammu and Kashmir as well as
Gilgit Baltistan. To compare the current data with older data from the 2012-13 PDHS (which already excluded
Azad Jammu and Kashmir [AJK]), the data was rerun to also exclude Gilgit Baltistan [GB].
I read this, and the README on weights included with the 2017-18 data files that helpfully explains that the standard weights are set to 0 for AJK and GB, to mean that users should exclude AJK and GB from the 2017-18 DHS to be able to compare it correctly with the 2013-14 results.
When I look at both survey reports, however, it appears that the 2012-13 survey did include GB, as did the 2017-18 survey. So, it's not clear why GB should be excluded from analysis for comparability, if it was covered in both surveys.
Further, it appears from the report that the 2012-13 survey did not cover FATA, which was included in the 2017-18 survey. Yet I do not see any mention of excluding the FATA region from the 2017-18 survey for comparability with the prior round.
Can you please confirm whether or not
1. Gilgit Baltistan, and
2. FATA
should be excluded from each survey round in order to compare the two surveys? Thank you very much for your help.]]>Sarah B2019-06-28T20:36:16-00:00Re: Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=17876&th=8255#msg_17876
Following is a response from the Pakistan Survey Manager, Anjushree Pradhan:
Due to political dynamics, the country recommended that the 2017-18 PDHS should exclude Azad Jammu and Kashmir (AJK) and Gilgit Baltistan (GB) from the national totals, so we had to comply. We collected data for these two regions, but due to the fact that they are considered separate entities, they were given separate weights.
In order to compare the national totals for Pakistan, we had to exclude these 2 regions from the 2012-13 PDHS. As AJK was not included in the survey in 2012-13 PDHS, we only had to exclude GB to get comparable figures (reference the footnote on pg.2).
Both surveys included Gilgit Baltistan, and the results for GB are presented in both. However, it is not politically correct to include GB in the national totals, which explains why it had to be excluded (from the total) for both surveys.
FATA on the other hand is part of the national total for Pakistan, but due to security issues in 2012-13, this region could not be surveyed. It was successfully surveyed in 2017-18 PDHS. Currently, FATA does not stand alone as a separate region, but is included under KPK.
On a final note, if one wants to compare the national totals for Pakistan, only AJK and GB should be excluded.
I hope this answers your questions.
]]>Bridgette-DHS2019-07-01T17:53:50-00:00Re: Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=17884&th=8255#msg_17884
Thank you very much for this quick and helpful response! I understand now why AJK and GB should be excluded from national estimates when comparing the 2012-13 and 2017-18 PDHSs for political reasons, even though GB was included in both surveys.
I would appreciate further clarification on the effect on comparability of including FATA in national totals in the 2017-18 but NOT the 2012-13 survey. I understand that FATA was excluded from the 2012-13 survey for logistical/safety, not political, reasons. I see that FATA is included as a region and in national totals for the 2017-18 survey.
When comparing national-level estimates, wouldn't including FATA in the 2017-18 survey but not in the 2012-13 survey skew the comparison? We are analyzing national-level trends, and it seems like we should only include areas that were covered in both surveys so the results are as comparable as possible.
Can you please help us understand why the FATA areas were not/should not be excluded from national estimates when comparing the 2012-13 and 2017-18 survey results?
Thanks again.]]>Sarah B2019-07-02T19:58:32-00:00Re: Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=17930&th=8255#msg_17930
As a follow-up to the previous question, which makes the issue very clear, I wanted to ask about a different scenario.
If one wanted to conduct some methodological analysis including all regions included in the PDHS 2017-18, are there any methodological restrictions (other than the political reasons) in combining the weights (for example hv028 and shv028) by replacing the zeros in hv028 by the non-zero values within shv028?
In other words - are the weights for AJK and GB (e.g. shv028) created using the same algorithm as the national weights (e.g. hv028)? If in the future the political situation is resolved, is this how the new national weight (including GB and AJK) would be created?
Thanks,
Diego
]]>DiegoBassani2019-07-18T14:47:37-00:00Re: Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=27710&th=8255#msg_27710
DiegoBassani wrote on Thu, 18 July 2019 19:17
Hi Bridgette and Anjushreen,
As a follow-up to the previous question, which makes the issue very clear, I wanted to ask about a different scenario.
If one wanted to conduct some methodological analysis including all regions included in the PDHS 2017-18, are there any methodological restrictions (other than the political reasons) in combining the weights (for example hv028 and shv028) by replacing the zeros in hv028 by the non-zero values within shv028?
In other words - are the weights for AJK and GB (e.g. shv028) created using the same algorithm as the national weights (e.g. hv028)? If in the future the political situation is resolved, is this how the new national weight (including GB and AJK) would be created?
Thanks,
Diego
Dear DHS, can we get response on this?
]]>owraza2023-09-25T04:00:34-00:00Re: Pakistan 2012-13 and 217-18 regions and provinces excluded for comparability
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=27716&th=8255#msg_27716
If you are asking whether you could construct a pooled weight that would be (in the HR file) shv005 for AJK and GB, and hv005 for the rest of Pakistan, unfortunately the answer is no. The weights would need to be rescaled, such that the total weights of AJK, GB, and the rest of Pakistan in the HR file would be proportional to the total numbers of households at the time of the survey. You would have to find estimates of those totals. You would also have to rescale v005 and mv005 for women and men.