| Child's dietary history [message #9775] |
Mon, 16 May 2016 15:36  |
WoodyWoodruff
Messages: 5
Registered: September 2013
Location: Victoria, British Columbi...
|
Member |
|
|
I am trying to derive some of the standard WHO/UNICEF IYCF indicators from data from the 1999, 2005, and 2011 DHS in Guinea. The 2005 DHS questionnaire asked about the frequency of consumption of several liquids and food groups in the past 24 hours. I would like to use these responses to derive something like adequate dietary diversity in children 6-23 months of age and calculate the prevalence of exclusive breastfeeding in children <6 months of age. However, for the pertinent variables in the child recode dataset (variables V460A - V469Y), there are many missing values (see tables in attached file).
The report from the 2005 DHS gives analyses based on these variables, but I don't know how this was done given the very high proportion of records with missing values.
My question is why are there so many missing values and how was the analysis of these variables done for the report?
|
|
|
|
|
|
|
|
| Re: Child's dietary history [message #9826 is a reply to message #9824] |
Thu, 26 May 2016 16:11  |
WoodyWoodruff
Messages: 5
Registered: September 2013
Location: Victoria, British Columbi...
|
Member |
|
|
Dear Trevor:
Thanks for the information. I had already selected the appropriate children when I previously posted the frequency distributions of the dietary variables. The attached SPSS output shows these distributions for the specific children you mention - the most recently born living children living with their mothers. In addition, because I'm trying to derive the standard IYCF indicators for complementary feeding adequacy, I've selected only those children 6-23 months of age. As you can readily see, for many of the variables, there are still many missing values. This problem seems to be specific to the 2005 Guinea child recode dataset. The 1999 and 2012 Guinea DHS datasets, as well as the 2000, 2005, and 2011 Ethiopia datasets with which I've worked previously, do not have this problem. Any explanation you can offer would be greatly appreciated.
|
|
|
|
| Re: Child's dietary history [message #9946 is a reply to message #9826] |
Tue, 07 June 2016 18:23 |
Trevor-DHS
Messages: 805
Registered: January 2013
|
Senior Member |
|
|
All of the missing data that you are seeing is related to the structure of the questionnaire used at the time of the Guinea 2005 survey. in this survey women are asked if they gave children each type of food in the past 7 days, and if they gave the food in the past 7 days then whether they gave the child the food in the past 24 hours. If they did not give the child the food in the past 7 days or the response was missing, then there is no valid response in the question on the last 24 hours.
You may want to look at variables M37A-X (last 24 hours) and M40A-X (last 7 days). For example, if you compare M37C with V469C for the selected cases you will see that they are identical. If you then crosstab M37C (or V469C) with M40C you will see that the cases that are codes 0 or 9 on M40C are the one that are not valid on M37C and V649C.
Also note that your selection criteria is not quite correct. You need to select the youngest child living with the mother. You have always selected the last born child (bidx=1), but this child is not the youngest child living with the mother if the last born child does not live with the mother or is dead. You need to first sort the data according to the ID fields v001, v002, v003, bidx (the file may be sorted already) and then use the following condition:
COMPUTE filter_$=(HW1>=6 AND HW1<24 AND b5=1 and b9=0 and lag(caseid)<>caseid).
Using the lag function ensures that only the last child living with the mother is selected.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
