The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Bangladesh » Developing wealth-specific all women factors
Developing wealth-specific all women factors [message #9672] Tue, 03 May 2016 12:51 Go to next message
sarahn
Messages: 3
Registered: October 2014
Location: UK
Member
Hi
I think I am going to need to develop AWFs for wealth quintiles. I would be very grateful for any assistance in how to go about this.
Many thanks
Re: Developing wealth-specific all women factors [message #9685 is a reply to message #9672] Fri, 06 May 2016 11:20 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:


DHS surveys that are restricted to ever-married women include awfactt, awfactu, awfactr, awfacte, and awfactw. (They may also include other versions of awfact among the country-specific variables.) The final letter tells you what covariate the factor is designed for: t for total, u for urban/rural, r for region, e for education, and w for wealth. What you need, awfactw, should already be in the data file.
Re: Developing wealth-specific all women factors [message #9686 is a reply to message #9685] Fri, 06 May 2016 11:23 Go to previous messageGo to next message
sarahn
Messages: 3
Registered: October 2014
Location: UK
Member
Thanks for this. However, I'm hoping to look at trend data, and some of the earlier surveys do not seem to have an AWF for wealth.
Re: Developing wealth-specific all women factors [message #9689 is a reply to message #9686] Fri, 06 May 2016 11:54 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Another response from Tom Pullum:

Some of the earlier surveys do not include the wealth index, and such surveys would not include awfactw. But I guess it's possible that there are surveys with wealth but not awfactw. Could you identify at least one such survey?

Please search for awfactw with another name among the country-specific variables, using "lookfor factor"....
Re: Developing wealth-specific all women factors [message #11178 is a reply to message #9685] Tue, 15 November 2016 21:24 Go to previous messageGo to next message
charvey91 is currently offline  charvey91
Messages: 10
Registered: February 2016
Member
Hi Tom,

Thanks for your reply. I am also working with Sarah on using these awfact for earlier surveys to create trend analysis (she wishes to pass on her apologies for not replying sooner to your previous responses). We are still experiencing some problems for earlier surveys where there is no wealth index and therefore no awfactw. Do you know if there is any way to create the wealth quintiles and the corresponding awfactw? The specific survey we are looking at is Indonesia 1997. Many thanks! Chloe
Re: Developing wealth-specific all women factors [message #11179 is a reply to message #9685] Tue, 15 November 2016 22:53 Go to previous messageGo to next message
charvey91 is currently offline  charvey91
Messages: 10
Registered: February 2016
Member
Sorry, I also have a further question. Do you know if it is possible to create confidence intervals for cross-tabulations that are using the awfact in stata? Thank you
Re: Developing wealth-specific all women factors [message #11185 is a reply to message #11179] Wed, 16 November 2016 09:50 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:


Quote:
There is no question that awfact (a generic term for awfactt, etc.) is a random variable, and when the weight is multiplied by awfact/100, the product has a substantial random component, especially for women under age 24. The all-woman factor is calculated for a single year of age, such as age 15, as the number of women (girls?) in the household population who are age 15 divided by the number at that age who are ever-married. Thus awfact is a random variable and it has an associated sampling error. However, I recommend that you simply proceed with using the net weight as if it had no random component.

But what do you mean by "confidence intervals for cross tabulations"? A cross-tab is a table of frequencies from the sample, for example the number of cases in the sample in each combination of current age group and current contraceptive method. You should use weights for those frequencies. Then you can calculate percentages, measures of association, etc., all of which could have a confidence interval. Please clarify what you mean by that term.

Re: Developing wealth-specific all women factors [message #11192 is a reply to message #11185] Wed, 16 November 2016 22:08 Go to previous messageGo to next message
charvey91 is currently offline  charvey91
Messages: 10
Registered: February 2016
Member
Thanks! Sorry, I didn't explain very well: we are calculating the proportion of women who have given birth by a certain age and would like to produce the confidence intervals too. When using the awfact, I have been applying it manually (not in stata), so have not been able to produce the confidence intervals. Do you have an example stata code for looking at a proportion, with the survey weight and awfact applied? Thank you
Re: Developing wealth-specific all women factors [message #11203 is a reply to message #11192] Thu, 17 November 2016 12:57 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Another response from Tom Pullum:

Quote:
Here is an example using the latest Egypt survey, with was an EMW survey.

set more off
use e:\DHS\DHS_data\IR_files\EGIR61FL.dta, clear

* what is the proportion of women age 20+ who had a birth before age 20?
* age at first birth is v212
* must restrict to women age 20+; say we restrict to current age 20-24

keep if v013==2
tab v212

gen teen_birth=0
replace teen_birth=1 if v212<20

gen wtt=v005*awfactt/100
svyset v021 [pweight=wtt], strata(v022)
svy: prop teen_birth 

* repeat for wealth categories; must use awfactw rather than awfactt
gen wtw=v005*awfactw/100
svyset v021 [pweight=wtw], strata(v022)
svy: prop teen_birth, over(v190) 

[Updated on: Thu, 17 November 2016 12:58]

Report message to a moderator

Re: Developing wealth-specific all women factors [message #11289 is a reply to message #11203] Tue, 29 November 2016 10:21 Go to previous message
charvey91 is currently offline  charvey91
Messages: 10
Registered: February 2016
Member
Thank you for this. It's very helpful! Chloe
Previous Topic: Bangladesh 2014 - administrative area code (identification)
Next Topic: CASEID & BIDX
Goto Forum:
  


Current Time: Thu Mar 28 09:57:23 Coordinated Universal Time 2024