The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » India » India NFHS-3 - slum coding
India NFHS-3 - slum coding [message #11870] Wed, 22 February 2017 10:24 Go to next message
soniwe is currently offline  soniwe
Messages: 7
Registered: October 2016
Location: Auckland
Member
Hi there,

I am using the NFHS-3 dataset to calculate equity of coverage of vaccination and other maternal and child health interventions. One of the variables we are using to stratify by is sslumc (slum designation by census). According to the map file and the variable categories in the dataset, 1 is coded as "slum" and 2 as "non slum". When I do cross-tabs of these (after svyset) by variables such as vaccination and skilled birth attendance, coverage seems to be much higher in the "slum" group. This is counter-intuitive, and I wonder whether there may be a mistake in the variable coding here?

Thanks,

Sonia
Re: India NFHS-3 - slum coding [message #11875 is a reply to message #11870] Thu, 23 February 2017 04:14 Go to previous messageGo to next message
soniwe is currently offline  soniwe
Messages: 7
Registered: October 2016
Location: Auckland
Member
Hi again,

I have realised why I was having those counter-intuitive results. I had recoded the missing values and included them as "non-slum". The majority of these were from rural populations, which have much lower coverage of health interventions even than urban slums, and this was skewing the results. I now want to recode into three categories: "Urban non-slum" "Urban slum" "Rural". Is it ok to re-code the missing as "Urban non-slum" if v025 is 1 (urban) and as "Rural" if v025 is 2 (rural) (see code below)? Are those participants categorised as "non slum" in the variable sslumc equivalent to other urban dwellers categorised as 1 in v025 - i.e. does it make sense to combine these two groups?

Sonia


recode sslumc (.=2) if v025==1
recode sslumc (.=3) if v025==2
label define slum 1 "Urban slum" 2 "Urban non slum" 3 "Rural"
Re: India NFHS-3 - slum coding [message #11878 is a reply to message #11875] Thu, 23 February 2017 10:35 Go to previous message
fredarnold is currently offline  fredarnold
Messages: 45
Registered: March 2014
Member
In NFHS-3, information on whether or not enumeration areas are in slums or non-slums is only available for eight cities (Mumbai, Delhi, Kolkata, Hyderabad, Chennai, Indore, Meerut, and Nagpur). For those eight cities you could identify slums, non-slums, and rural households, but you can't do that for the whole country.
Previous Topic: Regarding the new DHS dataset
Next Topic: Maratha & Mahar caste codes (NFHS 1, 1992-93, India)
Goto Forum:
  


Current Time: Thu Mar 28 11:50:50 Coordinated Universal Time 2024