Share of children received ICDS services in India [message #18684] |
Tue, 28 January 2020 06:07 ![Go to next message Go to next message](/theme/default/images/down.png) |
biswaj41_ssf
Messages: 7 Registered: May 2019
|
Member |
|
|
I have tried to figure out the share of living children under age six years (0-71 months) who have received any service from ICDS center in India. This is a replication of table no 9.18 in India's national report. I have used the BR file and following codes
* Generation of age
gen CMC_0_to_71 = v008 - b3
* Percentage of living children under age 6 years (0-71 months) used any benifits from ICDS Centre
tab v025 s555 [iweight = v005/1000000] if CMC_0_to_71 <= 71 & b5==1, row nofreq
But, this estimation is not exactly matching with the national report. Any help in this regard will be highly appreciated.
|
|
|
|
Re: Share of children received ICDS services in India [message #18721 is a reply to message #18684] |
Wed, 05 February 2020 10:13 ![Go to previous message Go to previous message](/theme/default/images/up.png) ![Go to next message Go to next message](/theme/default/images/down.png) |
Bridgette-DHS
Messages: 3155 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
After some checking, I have determined that column 7 of this table, which has the heading "Number of children under age six years" actually gives the number of children under five, i.e. age 0-59 months. If you open the file and enter "tab s555 if v008-b3<60 [iweight=v005/1000000]" you will get 238,945.45 weighted cases, which matches the 238, 945 at the bottom of column 7 of the table. There is no need to restrict to b5=1, because s555 is NA if the child has died.
It appears that columns 1-6 were correctly based on children under six years, and it's just column 7 that was based on children under five. If you enter "tab s555 if v008-b3<72 [iweight=v005/1000000]" you will get "yes" for 53.62%, which matches the 53.6% at the bottom of column 1.
To get column 2, for food supplements, I would enter these lines:
gen food=.
replace food=0 if s555<.
replace food=1 if s555==1 & s556>0 & s556<.
tab food if v008-b3<72 [iweight=v005/1000000]
However, this gives 48.22% who received food supplements, and table 9.18 has 48.1% at the bottom of column 2.
I have tried some variations on this. I see, for example, that some children have 1 for column 1 but 0 for all of the benefits in columns 2-6. It would have been reasonable to change their s555 code from 1 to 0 during editing or during the construction of the table. However, no matter how I take that into account, I cannot match the 48.1% at the bottom of column 2.
I cannot take the time to check further. I will ask someone else to look at this table and let you know if we can get a match.
|
|
|
|
|
|
|
|
|
Re: Share of children received ICDS services in India [message #19533 is a reply to message #19504] |
Tue, 07 July 2020 15:02 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
Bridgette-DHS
Messages: 3155 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
There has been quite a bit of discussion of this issue. Unfortunately I have no additional information and can't put any more time into resolving the inconsistencies.
This kind of a dead end only happens occasionally but it is not unprecedented. When you cannot reconstruct an estimate in a report table but you want to do further analysis, you just have to proceed with your own estimate. Include a footnote saying that the estimate in the table cannot be replicated.
I will point out that the tables in this report went through a major revision at one point when the sampling weights were corrected. It is possible that some numbers with the old (incorrect) weights were inadvertently carried forward. Also there were several special tables in this report, and I think this is one of them, that required some special programming (I believe at IIPS) and that made them more vulnerable to error than a table that is standard. I suggest that you just go ahead with your own analysis and a footnote.
|
|
|