| Constructing SLI variable in NFHS [message #99] |
Thu, 21 February 2013 12:30  |
DHS user
Messages: 111
Registered: February 2013
|
Senior Member |
|
|
I am doing a comparison over time by using the 92-93 and 98-99 Indian surveys. In going over the 92-93 dataset, I could not find any summary measure of standard of living that is comparable to SSLI from the
98-99 dataset. If it hasn't been constructed, could you tell me how it was constructed for the 98-99 survey so that i can compute it for the earlier one as well. Also, I couldn't find any identifier for village in the 92-93 survey. I am planning on doing a multilevel model so I need to decide what the second level will be (I want to keep it below a district).
|
|
|
|
| Re: Constructing SLI variable in NFHS [message #100 is a reply to message #99] |
Thu, 21 February 2013 12:33  |
Bridgette-DHS
Messages: 3035
Registered: February 2013
|
Senior Member |
|
|
Here is a response from one of our DHS experts Fred Arnold, that should answer your question.
No summary measure of the standard of living was constructed for NFHS-1 and added to the data set. Some people have created their own measures for specific analysis, but you would probably be better off to create your own measure. Since the questions that are part of SLI were slightly different in NFHS-1 and NFHS-2, you would not be able to get an exact match unless you recalculated the NFHS-2 SLI with only the components that were available in NFHS-1. The exact definition of the NFHS-2 SLI is available in Section 2.5 of any of the NFHS-2 reports. I should point out that the SLI is a very simple measure that assigns points to different assets and amenities and sums them. You may want to create your own index using principal components analysis or a similar method.
With respect to the village identifier in NFHS-1, you're correct that there is no village identifier as such, but there are still some things you can do. First of all, there is a variable (SHTEHSIL), which is the sub-district level, which you could use for modeling below the district level. Secondly, for rural areas, you can use the PSU number as a unique identifier of enumeration areas. There are two problems with this, however. First, the PSU number is unique within each state, but the same PSU number may be used in more than one state, so you would have to take into account both the state number and the PSU number in rural areas to have a unique identifier at the national level. Second, PSUs sometimes (not often) contain more than one village. This happens because small villages (typically with fewer than 50 households) are linked with larger villages in the sample design. Therefore, a PSU will contain more than one village if the villages in the selected PSU are linked. However, I don't think this would be much of a problem for you because linked villages are always contiguous, so they are always in the same small area.
I hope this helps.
Bridgette-DHS
[Updated on: Mon, 18 March 2013 09:03] Report message to a moderator |
|
|
|
|
|
| Re: Constructing SLI variable in NFHS [message #4081 is a reply to message #1449] |
Thu, 26 March 2015 08:29 |
Bridgette-DHS
Messages: 3035
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Specialist, Noureddine Abderrahim:
In the recode data files of both surveys India 1998-99 and India 2005-06 the household identification is the state code (5 characters), the PSU number (3 characters) and then the household number (4 characters). All the information, state, psu, and household number are survey specific. Any household that is surveyed twice is done by coincidence and not on purpose. The two samples are independent.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
