The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Merging data files » Can continuous DHS be merged with standard DHS?
Can continuous DHS be merged with standard DHS? [message #3057] Thu, 09 October 2014 11:07 Go to next message
sarahn
Messages: 3
Registered: October 2014
Location: UK
Member
I am involved in a project that will involve appending datasets from several different countries. Is it possible to append continuous DHS (e.g. Peru and Senegal) with standard surveys, and is there anything particular I would need to know before doing this?
Many thanks
Re: Can continuous DHS be merged with standard DHS? [message #3100 is a reply to message #3057] Wed, 15 October 2014 11:14 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member
Following is a response from Senior DHS Specialist, Shea Rutstein:

The Peru Continuous DHS, and I believe the Senegal Continuous DHS are representative each year (indeed in Peru, each semester). In order to combine surveys, the weights need to be denormalized. Traditionally DHS "normalizes" sampling weights by dividing each weight by the sum of the original sampling weights and multiplying by the total number. either interviewed households or interviewed persons, depending on whether it is household data or individual data. This normalization process
is unnecessary and creates problems in combining surveys. To denormalize the weights is simple, just multiply each weight in a given survey by the inverse of the normalization factor, i.e. by the the sum of the original weights divided by total number of cases. The problem is obtaining the sum of the original cases, which can be gotten by looking at the standard recode programs for that survey or tabulating the original weights.

Another issue is the wealth index, since it is usually based on each survey year. Use the Comparative Wealth Index instead.

One has to beware about changes in the sampling frame as well, such as when a new frame is based on a new census since the population distribution and definitions of rural and urban and states may have changed between the frames.
Previous Topic: v001 v002 v003 doesnot uniquely identify observation
Next Topic: HH head education level merge
Goto Forum:
  


Current Time: Thu Mar 28 16:53:08 Coordinated Universal Time 2024