| women's empowerment on child nutrition [message #9373] |
Mon, 21 March 2016 17:38  |
castle
Messages: 2
Registered: March 2016
Location: Colombia
|
Member |
|
|
Hi.
I'm going to estimate a IVPROBIT to calculate the impact of women's empowerment on child nutrition. I am using KR and PR database for Colombia 2010. The dependents variables are HW70, HW71 and hw72. First I made unique strata values by region/urban-rural and I think that I classified the nutritional status in children but now I need to generate the IV. I have thinking use the differences in year of education between women's parents. Would you like to help me to create it? I don't know what variable exactly to use because there are a lot of highest year education and I can't identify which one is related to women's parents.
Other points that are also confusing for me are:
- Do I need to use strata variable into the regression? How can I do it?
- Which variable you recommend me to use for the classifcation of the household in married or living together. ? hv115 or v501 variable?
|
|
|
|
| Re: women's empowerment on child nutrition [message #9434 is a reply to message #9373] |
Sun, 27 March 2016 16:12  |
 |
user-rhs
Messages: 132
Registered: December 2013
|
Senior Member |
|
|
Castle,
I'm not sure there is a best way to determine the woman's parents' educational attainment. The forum admins can verify, but I don't think the DHS asks a question about the woman's parents' education in the core questionnaire. Maybe the Colombia 2010 survey includes a module on woman's parents' education--I don't know, but you can check in the questionnaire at the end of the Colombia 2010 final report to see if the question was asked.
If Colombia 2010 used the core questionnaire, the only way you would get woman's parents' education is if the parents live in the same HH as her, in which case the information would be stored in the HH roster (HR) dataset. I believe the HH roster dataset is stored as wide (one line of observation = 1 household), so you would need to reshape it to long, so that each line of observation represents a household member. Then, you would need to determine whether the woman in question is a HH head or the spouse of the HH head. If she is the HH head, then you would need to make note of the people whose value for hv101 (relationship to HH head) == 6(parent). If she is the spouse of the HH head, then you would need to make note of the people whose value for hv101 == 7 (parent-in-law). You could use the hv104 variable to determine the sex of the parent/parent in law and go from there.
Again, this suggestion will only work if the women's parents live in the same household as she does. Otherwise, you will be missing a lot of information on the woman's parents' education attainment.
Re: Strata variable, if you are using Stata, -ivprobit- allows the -svy- prefix, so I suggest setting the strata, clusters, and pweights using -svyset-.
Re: Married or living together, it's probably easiest to use v501.
rhs
[Updated on: Sun, 27 March 2016 16:48] Report message to a moderator |
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
