The DHS Program User Forum: India » Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports

Home » Countries » India » Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports

Show: Today's Messages :: Show Polls :: Message Navigator

Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28308]

Mon, 11 December 2023 11:48

asriva
Messages: 5
Registered: December 2023

Member

Hello,

I am trying to reproduce table 11.20 in both NFHS-4 and NFHS-5 reports. I am interested in measuring gendered differences in access to healthcare and am using the variable for Visits to a health facility or camp in the past 3 months. (S368 in Individual Women's Recodes and SM320 in Male Recode)

I am unable to reproduce the estimates shown in column 2 of the table - Percentage who visited a health facility or camp in the past 3 months for Women.

I am able to reproduce approximately the same estimates in column 1 - Percentage of women with any contact with a health worker (using S359 and S361), and column 3 - Percentage who visited a health facility or camp in the past 3 months for Men (SM320).

I am weighting my estimates using the individual's sample weight (V005 and MV005).

Can you help me with the code to reproduce these estimates?

Moreover, are there other variables you would suggest that could capture gendered differences in access to healthcare, and different types of facilities?

Thank you,
Aryan

Report message to a moderator

Re: Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28310 is a reply to message #28308]

Mon, 11 December 2023 15:36

fred.arnold@icf.com is currently offline

fred.arnold@icf.com
Messages: 84
Registered: May 2021

Senior Member

I don't see any reason why you cannot reproduce the estimates in column 2 of the table, but you are able to reproduce the estimates in column 1 and column 3. The questions in the questionnaires for women and men in both NFHS-4 and NFHS-5 are identical. The table title, column headings, and footnote in Table 11.20 are also identical. There is always a possibility that the files you are using to produce Table 11.20 are slightly different from the files that were used at the time that Table 20.11 was produced for the national report tables, be even in that case any difference would be quite small.

You should also check the NFHS-4 and NFHS-5 questionnaires for women since there are some skip instructions.

With respect to additional variables to capture gendered differences in health care, it's best if you check the NFHS-4 and NFHS-5 questionnaires for other variables that might be useful for your analysis.

Report message to a moderator

Re: Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28311 is a reply to message #28310]

Mon, 11 December 2023 15:51

asriva
Messages: 5
Registered: December 2023

Member

Ah sorry, I should have provided more context about why I haven't been able to reproduce those estimates.

I did use the variable that refers to the question about visits to a health facility or camp in the previous 3 months from the women's recode (S368), along with weights, to find the percentage of women per state as per Table 11.20. However, the estimates I am getting from the IR data are much higher than the table. For Rajasthan in NFHS-5, the table reports 11.8% women went to a health facility or camp. But, I get 30.4% running this command -

tab s368 [iw=v005/1000000]

The skip patterns from the questionnaire indicate that Q.368 was asked if the respondent hadn't already indicated visiting a health facility or camp in two of the previous questions. Considering that the case, I replace the missing values in S368 with 1, which raises the percentage even more to 38.62% -

replace s368 = 1 if s368 == .
tab s368 [iw=v005/1000000]

I don't understand how the estimate in the report is so much lower than what I get using the womens recode data. By being able to reproduce the estimates for the first and third, I mean the percentages I get from the data matches the percentages I see in the report table.

Please let me know if I am choosing the incorrect sample somehow, or if there was some post processing done on the variable to achieve the estimates in the table.

I would greatly appreciate it.

Report message to a moderator

Re: Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28322 is a reply to message #28311]

Thu, 14 December 2023 06:59

asriva
Messages: 5
Registered: December 2023

Member

Hi, bumping this thread as this is a bit urgent. Would love to get a response to this.

Thanks,
Aryan

Report message to a moderator

Re: Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28384 is a reply to message #28322]

Fri, 22 December 2023 18:53

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Following is additional response from Senior DHS Specialists, Blake Zachary and Tom Pullum:

Column 2 of table 11.20 was calculated from s360c and s363c (based on questions 360 and 363 in the questionnaire), NOT from s368. The Stata lines are given below. The questions for men did not exactly parallel those for women. It may be somewhat misleading that the percentages for women in column 2 of the table are consistently well below the percentages for men in column 3.

use "...", clear

* Column 1 of table 11.20 
gen     visited=0
replace visited=1 if s366<.
tab v024 visited [iweight=v005/1000000], row 

* Column 2 of table 11.20 
gen camp=0 
replace camp=1 if (s360c>=1 & s360c<=90) | (s363c>=1 & s363c<=90)
tab v024 camp [iweight=v005/1000000], row 

use "...IAMR7EFL.DTA", clear

* Column 3 of table 11.20 
gen camp=0 
replace camp=1 if sm320==1
tab mv024 camp [iweight=mv005/1000000], row

Report message to a moderator

Re: Reproducing Table 11.20 in NFHS-4 and NFHS-5 Reports [message #28448 is a reply to message #28384]

Tue, 09 January 2024 05:07

asriva
Messages: 5
Registered: December 2023

Member

This is really helpful information. You're right that the reports are somewhat misleading. I don't understand why the decision was made to calculate the percentages for men and women from different variables, with no indication of it in the report.

Thank you for your help!
Aryan

Report message to a moderator

Previous Topic:	Merging BR to PR in NFHS-1 and NFHS-2
Next Topic:	Issues with replicating STATcompiler results for India

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Nov 21 14:13:21 Coordinated Universal Time 2024