Home » Countries » India » Hosehold Member (Household Composition)
| Hosehold Member [message #18975] |
Sun, 29 March 2020 01:58  |
Isha
Messages: 18
Registered: February 2016
|
Member |
|
|
I am using the Household Member Recode (PR) file, NFHS-4. I am interested in exploring household composition. I had a few questions:
1. Is it methodologically appropriate to use the NFHS to arrive at household composition by state?
2. If so, which weight would you advise to use: household sample weight (v005) or state household weight (SHV005). Currently, I am interested in arriving at a descriptive picture of household composition within one state and later I intend to compare certain HH structures across states.
3. I would like to arrive at household composition using the variable on relation to head of household (HV101). I have created a variable of 'Relation' which includes relationship to head (including head) and sex of each member (eg. female head=1, wife=2, daugther=3, male head=21, husband=22, son-23 etc). I want to analyze households separately by number of members. For instance, in two member households, what percent are wife-husband, what percent are mother-daugther, what percent father-son and so on. If I aggregate (in SPSS) the file by HHID, I am unable to add the 'Relation' variable to the output file since it is for each member (and cannot be aggregated by HHID). Is there any way to work this out?
Alternatively, if I use the split file command, I am able to arrive at what I am looking for, except it shows composition of members by each household (by HHID) and that too for 1000s of households, while I am looking for a count/percent. For instance, attached is a portion of the split file table which shows household composition of two-member households. Each row is for one household and there are around 1200 such rows. Could they be collapsed by common household structure?
Thank you!
|
|
|
|
| Re: Hosehold Member [message #18983 is a reply to message #18975] |
Mon, 30 March 2020 17:36  |
Bridgette-DHS
Messages: 3035
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
You definitely can use the NFHS to describe household composition by state. I would encourage doing that.
For state-level estimates you can use either the national-level or state-level weights. You should get the same results either way, because, within each state, the two sets of weights are proportional to each other. I suggest that you do something simple both ways, and see if there's a difference. I don't think there will be a difference.
I can describe how I would calculate different household types in Stata, and I hope you can translate to SPSS. As an example, lets say you want to identify households that consist of the head and spouse, and no other members. I construct a binary variable for each value of hv101. I won't look up the number of codes for hv101 in the NFHS-4 but let's say there are 16. Then, for example:
* Open the PR file
gen hv101_1=0
gen hv101_2=0
...
gen hv101_16=0
gen nhh=1
replace hv101_1=1 if hv101==1
replace hv101_2=1 if hv101==2
...
replace hv101_16=1 if hv101==16
collapse (sum) nhh hv101_*, by(hv024 hv001 hv002)
rename hv101_1 n_1
rename hv101_2 n_2
...
rename hv101_15 n_16
* You now have one record per household with the number of people in the household of each type
* n_1 is the total number of persons in the household with code 1 (should be exactly 1)
* n_2 is the total number with code 2
* nhh is the total number of people in the household
* Identify households with one head and one spouse and no one else, call this type A
gen hh_typeA=1 if n_1==1 & h_2==1 & n_1+n_2==nhh
* Identify nuclear households, type B
gen hh_typeB=1 if n_1+n_2+n_3==nhh
There are ways to make this more efficient. This should help you get started.
[Updated on: Mon, 30 March 2020 17:37] Report message to a moderator |
|
|
|
Goto Forum:
Current Time: Fri Apr 19 19:26:58 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
