| Antenatal care - at least four visits (%) [message #2231] |
Sat, 31 May 2014 09:10  |
amira.elshal.1@city.ac.uk
Messages: 52
Registered: April 2013
|
Senior Member |
|
|
Dear Sirs,
What is the minimum number of observations required to calculate a maternal health indicator such as Antenatal care - at least four visits (%) etc at the district level?
I need to collapse Egypt DHS individual recode data at the district level, then calculate some maternal health indicators. The problem is that some districts have got 15 observations only.
Thank you.
Sincerely,
Amira
|
|
|
|
| Re: Antenatal care - at least four visits (%) [message #2379 is a reply to message #2231] |
Wed, 11 June 2014 09:40  |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
Dear User,
Here is a response from our expert sampler, Dr. Ruilin Ren:
Hi Amira
Usually the DHS surveys were not designed to allow disaggregation below the survey domains which are the Governorates in Egypt. The indicator you are interested related to last birth (or all births) in last five years which is much smaller than the number of women interviewed (not all women had a birth in the last five years). The DHS's policy is that if an indicator was calculated based on 25 un-weighted cases or less, the DHS will not report the indicator because the sampling errors will be relatively large for small sample size. I did not go to the details of the Egypt data, according to what you reported, some districts have only 15 observations (it is not clear if you got 15 births or 15 women interviewed), we strongly recommend you to not disaggregate to single district level. You may need to group the districts to get relatively larger areas for your analysis. If your study restricts you to single district analysis, you cannot just run your analysis based on the data you have, you may need advanced techniques such as small area smoothing which involves the use of outside-survey information and large amount of calculations.
Hope this is helpful.
Thanks.
|
|
|
|
|
|
| Re: Antenatal care - at least four visits (%) [message #2386 is a reply to message #2383] |
Thu, 12 June 2014 12:31 |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
Response from Dr. Ruilin Ren (sampling expert):
Hi Amira
I do not know what kind of analysis you are doing. The aggregated analysis has two meanings, to get enough sample size, and to aggregate to known geographical or administrative units of the country. If you regroup the districts, you know where the aggregated area is. If you aggregate a few clusters just to get a larger number of cases, but you do not know what is the area or the administrative unit your aggregated clusters represent for, even you can calculate an indicator value, but you do not know where it represents for therefore you cannot interpret it. I think you have to aggregate to known geographical or administrative units level.
Best Regards
Ruilin
[Updated on: Thu, 12 June 2014 12:32] Report message to a moderator |
|
|
|
|
|
| Re: Antenatal care - at least four visits (%) [message #2388 is a reply to message #2387] |
Thu, 12 June 2014 14:50 |
Liz-DHS
Messages: 1516
Registered: February 2013
|
Senior Member |
|
|
Response from Dr. Ruilin Ren(sampling expert):
Dear Amira
If you aggregate to district level it is ok. But do not exclude the districts where there is less than 25 cases, you should group the small ones, or attach a small one to a neighboring district. This makes more sense.
Ruilin
|
|
|
|
|
|
| Re: Antenatal care - at least four visits (%) [message #8799 is a reply to message #2231] |
Sat, 19 December 2015 12:59 |
Gazimbi
Messages: 7
Registered: March 2015
Location: United Kingdom
|
Member |
|
|
Hi, your help will be most useful. I am analysing the determinants of at least 4 antenatal care visits in Zimbabwe. I am doing a multilevel analysis. I want to limit my data to births in the last five years preceding the 2010 survey (v208). I see that there were multiple births among mothers during that period so I want to further restrict my analysis to last born child.I am using SPSS, and using IF condition satisfied v208>0 was my syntax for all births. I have looked at the variable v235, and wondered whether I can use this to restrict my data to the last birth- the trouble is I don't understand because the variable is an index coded 0 for no prior child. Now I dont understand the scores that follows after which range from 2-5.As a guesswork I tried if condition is satisfied v235>0 to get the last birth, is this correct please?
MG
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
