Home » Countries » Bangladesh » Details about KR file
| Details about KR file [message #14318] |
Thu, 22 March 2018 11:54  |
anikhpg42@gmail.com
Messages: 38
Registered: December 2017
Location: Bangladesh
|
Member |
|
|
Hi,
I am using BDHS-14 KR dataset. I want to learn about the data pattern of KR file.
I have read about KR file in DHS website where it says in about 5 lines. Which didn't clear my concept.
I want to know that,
"If I work for maternal and child (under age 5) malnutrition, can I use this?
I mean, suppose, I got 20% underweight mother and 36% stunted children under age 5.
Now, can I say that, there were 20% mothers were underweight and 36% children under age 5 were stunted, in the same household. ?"
What I also want to know, that, is this dataset indicate a household from which a mother and child under age 5 were interviewed?
Here, the total observation number were 7886. Then is this indicate that, there were 7886 households where from each household
a mother and a child under age 5 were interviewed?
I wrote same thing at different patterns, so that, you may capture my question.
Sorry for elaborating. I will be greatful, if anyone clear me this.
Thank you in advance.
ASIBUL ISLAM ANIK
|
|
|
|
| Re: Details about KR file [message #14323 is a reply to message #14318] |
Thu, 22 March 2018 14:54  |
Bridgette-DHS
Messages: 3210
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
In the KR file, if 20% of mothers are underweight, this means that 20% of the children have mothers who are underweight, because children are the units of analysis. If a woman has three children under five, she will appear as a mother three times in the KR file. If another woman has one child under five, she will appear as a mother only once in the KR file.
If 20% of mothers are underweight and 36% of children are stunted, that alone does not tell you whether the underweight women and stunted children tend to be in the same household. That would be a reasonable research hypothesis.
The survey is conducted in households, but the different data files have different units of analysis or cases. The HR file is the only file in which the household is the unit. In the PR file, all the individual household members are cases. In the IR file, women age 15-49 are cases. In the KR file, the children of those women, born in the past five years, are the cases. I think it would be better for you to use the anthropometry variables in the PR file, rather than the KR file.
|
|
|
|
| Re: Details about KR file [message #14325 is a reply to message #14323] |
Thu, 22 March 2018 15:13 |
anikhpg42@gmail.com
Messages: 38
Registered: December 2017
Location: Bangladesh
|
Member |
|
|
Actually I am working with double burden of malnutrition, i.e., overweight mother coexist with stunted child in the same household.
And, as a M.Sc thesis student, my supervisor told me to use KR file. But, at the time of interpretation, we couldn't match the number
of observation between children under age 5 and mothers. Because, stunted children were found to be 6965 and mother's bmi category
reveals 7056 observations. Now, what can I do?
I mean, I want to see the bmi status of mother and her child's z-score of same household.
ASIBUL ISLAM ANIK
|
|
|
|
| Re: Details about KR file [message #14340 is a reply to message #14325] |
Mon, 26 March 2018 07:22 |
Bridgette-DHS
Messages: 3210
Registered: February 2013
|
Senior Member |
|
|
Following is another response from Senior DHS Stata Specialist, Tom Pullum:
Below I have written some Stata code that construct the BMI categories for the mother and calculate the percentage of children age 0-4 who are stunted for mothers in each of these categories. The log file is attached. For example if the mother is in the lowest category of BMI, 44.5% of children are stunted. If the mother is in the highest category of BMI, only 21.8% of children are stunted. As the BMI of the mother increases, there is a monotonic decline in the percentage of children who are stunted.
set more off
use e:\DHS\DHS_data\KR_files\BDKR72FL.dta, clear
log using e:\DHS\scratch\logtemp.smcl, replace
* Child:
* hw70 int %8.0g HW70 height/age standard deviation (new who)
* hw71 int %8.0g HW71 weight/age standard deviation (new who)
* hw72 int %8.0g HW72 weight/height standard deviation (new who)
* Mother:
* v445 int %8.0g V445 body mass index
* Note that v445 is bmi x 10
gen stunted=.
replace stunted=0 if hw70>-600 & hw70<600
replace stunted=100 if hw70>-600 & hw70<-200
gen bmi_category=1
replace bmi_category=2 if v445>=1700
replace bmi_category=3 if v445>=1850
replace bmi_category=4 if v445>=2300
replace bmi_category=5 if v445>=2500
replace bmi_category=6 if v445>=3000
* recommend omitting implausibly high values but this is optional
replace bmi_category=. if v445>6000
label define bmi_category 1 "<17" 2 ">=17 and <18.5" 3 ">=18.5 & <23" 4 ">=23 & <25" 5 ">=25 & <30" 6 ">=30"
label values bmi_category bmi_category
tab bmi_category stunted [iweight=v005/1000000],m
svyset [pweight=v005]
svy: mean stunted,over(bmi_category)
-
Attachment: logtemp.smcl
(Size: 4.82KB, Downloaded 593 times)
[Updated on: Mon, 26 March 2018 07:23] Report message to a moderator |
|
|
|
| Re: Details about KR file [message #15098 is a reply to message #14340] |
Sat, 02 June 2018 06:10 |
Mayank_Ag
Messages: 22
Registered: May 2018
Location: Hyderabad
|
Member |
|
|
I am trying to generate same statistics for India using DHS 15-16.
I used the same code and also applied filters for pregnancy (V213 = 0) and birth within the last 2 months(V222 > 2) but my estimates are still not matching.
Can you please explain why is this happening ?
TIA.
|
|
|
|
Goto Forum:
Current Time: Fri Dec 13 19:53:15 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
