The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » India » Child marriage estimates for 1998 India DHS
Child marriage estimates for 1998 India DHS [message #15048] Mon, 28 May 2018 08:17 Go to next message
dgodha
Messages: 44
Registered: November 2016
Location: India
Member
Hello,

I am trying to replicate Table 3.4 on page 56 of the NFHS-II report but nothing is working. As explained in the report, the estimates are based on all states except Tripura. Also, using country-specific variable on age at marriage 's111', as suggested in one of the threads, is giving completely different estimates while those after using 'v511' are closer. Accordingly, I have used the following Stata commands:
gen wt=v005/1000000 
gen g=group(v024+v025)
svyset [pw=wt], psu(v021) strata(g)
gen aw1 = nwfactt/100

g CM18=(v511<18)
svy: ratio CM18/aw1 if v013 >= 2 & v024~=35, over(v013)

I am getting the following results:
	My estimates	From report
20-24	46.2	50.00
25-29	54.7	58.90
30-34	59.9	64.30
35-39	61.4	66.50
40-44	63.1	68.70
45-49	64.8	70.40

I have tried other specifications for survey weighting such as v022 and v023 for strata and using the all India sample (with 'awfactt' instead of 'nwfactt') but finally decided to go with the above as suggested in this forum and as per the sampling description in the report.

I will appreciate your feedback on this.

Many thanks
DG


Deepali
Re: Child marriage estimates for 1998 India DHS [message #15134 is a reply to message #15048] Thu, 07 June 2018 10:10 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:

If you want to construct a stratum variable using the "group" command, the correct syntax is egen strata=group(v024 v025). The command you used does not do this. I am just using v022 to get the strata.

I had to check the original CSPro code to find how the age at first marriage (call it "afm") is calculated. It does not use v511; instead it uses v503 (number of marriages), s111, and s114. Both v511 and s111 are labelled "age at marriage". s114 is labelled "start of (current) marriage". The coding for afm does not seem to make sense but that's because of the poor labels. The construction of s111 and s114 uses the question on cohabitation (q112), not just the question about formal marriage (q111). Note that I include a factor of 100 in afm so that the numbers produced are interpretable as percentages.

I am attaching a Stata program that constructs all the numbers in table 3.4 except for the median age columns (the last two columns) and the two bottom rows within each panel. You could get the latter with a simple modification of the program.
Re: Child marriage estimates for 1998 India DHS [message #15142 is a reply to message #15134] Thu, 07 June 2018 14:36 Go to previous message
dgodha
Messages: 44
Registered: November 2016
Location: India
Member
Many thanks Tom- it was a bit convoluted and I could never have guessed this.

However, when I apply the same survey setting (and also the group command) to get the estimates in Table 3.1 (age group; marital status), my weighted counts do not match. Note that the proportions and unweighted count match exactly. I will request your guidance on this too.

I look forward to your response.
Deepali


Deepali
Previous Topic: NAs in occupation variable of eligible women
Next Topic: Births protection against neonatal tetanus
Goto Forum:
  


Current Time: Thu Mar 28 07:33:22 Coordinated Universal Time 2024