The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Reproductive Health » Variable name and computation of indicator (Non-users and family planning)
Variable name and computation of indicator Mon, 06 December 2021 11:47
 kclakhara Messages: 20Registered: June 2016 Location: New Delhi Member

I have gone through the NFHS-4, 2015-16 and NFHS-5, 2020-21 state/district fact sheet and want to know about computation of indicator with the condition used for the generate indicator "30. Health worker ever talked to female non-users about family planning (%)"

With Regards,

Kailash

Kailash Chandra Lakhara
Re: Variable name and computation of indicator [message #23815 is a reply to message #23813] Mon, 06 December 2021 12:28
 fred.arnold@icf.com Messages: 6Registered: May 2021 Member
This is a very straightforward calculation. The indicator is based on the response to Question 356 on the Woman's Questionnaire ("Were you ever told by a health worker about any methods of family planning that you can use to avoid pregnancy?"). The question is only asked of women who have never used a method of family planning (a NO response on Question 319) and have not had a hysterectomy.
Re: Variable name and computation of indicator [message #23861 is a reply to message #23815] Mon, 20 December 2021 23:30
 kclakhara Messages: 20Registered: June 2016 Location: New Delhi Member
Dear Sir,

Thank you so much for the reply. I have gone trough the mentioned condition using AGRA NFHS-4 data but could not get expected results.

I have followed below motioned condition:

compute rweights=(sv005/1000000).
EXECUTE.
weight by rweights.

use all.
filter off.
EXECUTE.

compute xa=0.
if (v302a=0) xa=1.
if (v215=993 or s250=1) xa=0.
EXECUTE.

filter by xa.
EXECUTE.

fre s351.

Told by a health worker about any methods of family planning

Frequency Percent Valid Percent Cumulative Percent
Valid No 802 87.6 87.6 87.6
Yes 113 12.4 12.4 100.0
Total 915 100.0 100.0

The figure given in the AGRA factsheet is 12.1% whose not matching with my syntax.

With Regards,

Kailash

Kailash Chandra Lakhara
Re: Variable name and computation of indicator [message #23869 is a reply to message #23861] Thu, 23 December 2021 11:09
 Bridgette-DHS Messages: 2488Registered: February 2013 Senior Member
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

I use Stata, not SPSS, and my response will be in Stata. It should be clear even if you do not use Stata.

I can match the Agra Fact Sheet with the following code. I am not removing women who had a hysterectomy. It is possible that such women should have been removed in the coding for the Fact Sheet, but it appears that they were NOT removed, because if they are removed the percentages are too high (as you found). The percentages are the same if you use v005 or sv005 for the weights; your use of sv005 was ok. I do not have time to go through the data processing archives to find the actual CSPro code.

In this kind of a situation, if you are planning to do more in-depth analysis of these responses, you have two options. The first is to use the code that matches the Fact Sheet but is not consistent with the definition Fred Arnold gave you. The second would be to use the definition Fred Arnold gave you, in which case you will not be consistent with the Fact Sheet. Personally, I do not see a good reason why the definition should include any reference to a hysterectomy. There are many women who have not had a hysterectomy but are unable (for other reasons) to have more children. Also, if there had been an adjustment for hysterectomy I would have expected it to be given in a footnote. For these reasons, I would recommend that you NOT include any adjustment based on a hysterectomy.

```* Define "told" to include 0 for women for whom s351 is NA
gen told=0
replace told=1 if s351==1
tab told v025 if sdistri==146 & v302a==0 [iweight=v005/1000000], col
```

• Attachment: told.png
Re: Variable name and computation of indicator [message #23870 is a reply to message #23861] Thu, 23 December 2021 12:04
 kclakhara Messages: 20Registered: June 2016 Location: New Delhi Member

Thank you so much for the support. Problem has been resolved.

Regards,

Kailash

Kailash Chandra Lakhara
 Previous Topic: Matching Variables from IDIR and IDOD for 2002, 2007, 2012, 2017
Goto Forum:

Current Time: Sat Jan 29 09:25:01 Coordinated Universal Time 2022