| Missing data [message #12810] |
Mon, 17 July 2017 09:58  |
Jana
Messages: 5
Registered: September 2016
|
Member |
|
|
Hello!
I am using all available DHS surveys phases 2-6.
Now I wondered if there are certain surveys where the following information was generally not collected as I have a lot of missing variables:
- Birthweight
- Antenatal care variables
And is there a general overview available where I could see whether this data was generally not collected?
Further, I wondered if the anthropometric data as well as the information on the last born child (e.g. antenatal care use or birthweight) was collected referring to previous 3 years or 5 years before the survey?
Thank you for your support!
|
|
|
|
| Re: Missing data [message #12817 is a reply to message #12810] |
Tue, 18 July 2017 10:35  |
Bridgette-DHS
Messages: 3208
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
You are probably referring to cases that are blanks in ASCII or "." in Stata. Those codes should be interpreted as "Not Applicable" or "NA". Usually they arise in two different ways. First, a question and corresponding variable may be dropped but the variable is retained in the data file, and ALL the cases are coded ".". The second possibility is that it was only asked for a subset of women or children, such as the youngest child born in the past five years. (This is the child whose variables in the IR file have "_1" at the end.) Otherwise the code is ".". The NA cases should not be interpreted as "Missing". DHS uses numeric codes such as 99, etc. for refusals, out of range, flagged, etc., that would properly be described as "Missing".
|
|
|
|
|
|
| Re: Missing data [message #12823 is a reply to message #12819] |
Tue, 18 July 2017 14:24 |
Bridgette-DHS
Messages: 3208
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Specialist, Kia Reinis:
You can familiarize yourself with the Model Questionnaires to get an idea of what data are collected in most surveys run in that phase. If the Model Questionnaire changed from collecting data on children in the previous 5 years to collecting the data for children born in the previous 3 years, for example, then surveys in that phase will have done so as well. Or you can use STATcompiler to see which surveys included particular indicators. For example, you could look up the indicator percentage of births below 2.5 KG and decide whether you want to get results for a particular survey, for recent surveys, for all surveys. STATcompiler returns results for births in the last 3 years and births in the last 5 years in two separate columns. Overall, 222 surveys have birth-weight data, of which 22 have data for 3 years, but not 5 years. All of the three-year data are from surveys conducted in the 1990's (with exception of the 2002 Vietnam DHS).
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
