| Proximate Determinants of Fertility, Nigeria: Stata Syntax [message #25048] |
Tue, 23 August 2022 20:46  |
okunloladavid
Messages: 4
Registered: February 2020
|
Member |
|
|
I will be grateful if I can get Stata syntax for the proximate determinants of fertility in Nigeria (DHS 2018) by wealth, ethnic group and region respectively. I already have the syntax which works but I do not know where to modify the syntax to disaggregate the statistics by wealth, ethnic group and region.
Thank you for your help.
|
|
|
|
| Re: Proximate Determinants of Fertility, Nigeria: Stata Syntax [message #25180 is a reply to message #25048] |
Fri, 09 September 2022 16:04  |
Janet-DHS
Messages: 685
Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member Tom Pullum:
You are using this Stata program for the proximate determinants:
https:// userforum.dhsprogram.com/index.php?t=msg&th=9154&got o=19828&S=27f58c9d16a56e78ba6a70591c8bb2a3
I put that program together in 2019, restructuring work by others (acknowledged at the beginning of the program; John Bongaarts was of course the inventor). DHS final reports do not currently include this decomposition, and the program has only had limited use.
The program has two limitations. The first is that it is not self-contained. It requires the user to provide age-specific fertility rates in a separate excel file, which may, for example, come from a STATcompiler download. Second, it does not include standard errors or confidence intervals. It should be possible to construct them using nlcom, a powerful Stata command, or a jackknife procedure, but I did not build them in.
It would be possible to modify the program to run through all the categories of a categorical variable such as wealth quintiles or region. I believe STATcompiler gives the age-specific fertility rates for those categories for most surveys. Ethnic group would be difficult because it's not on STATcompiler and some categories are very small.
A shortcut approach that does not require advanced Stata procedures would be to set up a separate program that loops through the values of v190 (wealth quintiles), for example. For each value of v190 you would cut the IR file down do that category and then call (i.e. run) the proximate determinants routine. The values from that run would be saved in an output data file. I do this sort of thing all the time, typically looping over surveys, but you can loop over subpopulations. (You may need to include "program drop" at the end of each loop.) It would not be the most efficient approach if you wanted to go through a large number of subpopulations, but it would be ok for what you are doing. But I do worry about not having standard errors and confidence intervals...
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
