Home » Data » Sampling » Stratification
Re: Stratification [message #1464 is a reply to message #1454] |
Fri, 28 February 2014 11:40 |
Trevor-DHS
Messages: 803 Registered: January 2013
|
Senior Member |
|
|
DHS used to use an approach of producing implicit strata, based on combining clusters into pairs or small groups of 3 clusters. These implicit strata were used in the calculation of sampling errors. This approach was used in the World Fertility Surveys and in earlier DHS surveys were there was an implicit stratification based on an ordering of the clusters. The DHS sampling experts no longer recommend this approach for the stratification. v022 in the Zambia DHS IV survey contains the variable that has this pairing or grouping, as was used for the sampling error calculations reported in final report.
You will find this true in many DHS surveys, particularly the older ones.
In cases like this, check also v023 to see if this contains the strata. In many surveys this actually contains the regions rather than the strata. If the strata are not actually provided directly in the data set, then check appendix A of the final reports to see what strata were actually used. In many cases the strata will be urban and rural areas within region (in the case of the Zambia survey, urban and rural areas within province). However, in some surveys there is no stratification beyond the region, while in other surveys, there are 3 or 4 strata within each region. Appendix A of the final report should provide the information for this.
I hope this helps.
|
|
|
Goto Forum:
Current Time: Wed Nov 27 11:56:03 Coordinated Universal Time 2024
|