The DHS Program User Forum: Other countries

Home » Countries » Other countries » Pakistan (DHS 2006/7)

Show: Today's Messages :: Show Polls :: Message Navigator

Pakistan (DHS 2006/7) [message #488]

Tue, 28 May 2013 12:38

bzarra
Messages: 1
Registered: May 2013

Member

Hello,

I'm working with the Pakistan DHS data set in STATA at the moment, but the problem I'm having is respect to all my data sets. I'm trying to find an accompanying file that provides a legend for a variable in the data set that is coded as strictly a number.

For example, with the Pakistan data, the "district" level data is coded as a number between 1 - 35, but I can't find what district each of these numbers correspond with. I have an interest in working with the Swabi district, but I can't isolate for those respondents without knowing which one of these numbers is representative of the Swabi district.

Thanks for the help!

Report message to a moderator

Re: Pakistan (DHS 2006/7) [message #508 is a reply to message #488]

Mon, 03 June 2013 11:39

Thea-DHS
Messages: 28
Registered: March 2013

Member

Please view the related post( http://userforum.measuredhs.com/index.php?t=msg&th=70&am p;am p;#msg_122) that explains that district names are suppressed from the recode files for confidentiality purposes. Instead, you may use the GPS data for the 2006-07 Pakistan DHS to reclassify the cluster locations according to your district of interest. In order to gain access to the GPS data, which is separate from the regular DHS datasets, please log onto http://www.measuredhs.com/data/dataset_admin/login_main.cfm request the geographic data files, and update your project description with the information about how your project will use the district level information. The GPS datafiles contain a shapefile of the cluster locations for use in a GIS software package. QGIS (http://qgis.org/) is a good free software option if you do not have access to paid GIS software products. GADM (http://www.gadm.org/) is a good resource of administrative shapefiles that can be used in the reclassification process.

Also, please note that to ensure respondent confidentiality we randomly displace the GPS latitude/longitude positions for all surveys, including those that do not have HIV testing. The displacement is randomly carried out so that:

• Urban clusters contain a minimum of 0 and a maximum of 2 kilometers of error.
• Rural clusters contain a minimum of 0 and a maximum of 5 kilometers of positional error with a further 1% of the rural clusters displaced a minimum of 0 and a maximum of 10 kilometers.

The displacement of the Pakistan 2006-07 data was limited to the DHS regions boundaries, which means that it was possible for a cluster location to cross district or lower level boundaries during the displacement process. Also, caution is advised when conducting your analysis at the district level, as the 2006-07 Pakistan survey was not designed to be representative to the district level.

[Updated on: Tue, 24 March 2015 10:10] by Moderator

Report message to a moderator