The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Do I need to De-normalize and Pool Data for Trends Over Time in a Single Country?
Do I need to De-normalize and Pool Data for Trends Over Time in a Single Country? [message #5535] Wed, 03 June 2015 12:42 Go to next message
jd.weaver is currently offline  jd.weaver
Messages: 5
Registered: April 2015
Location: Seattle, WA
Member
Hi All,

I want to look at trends over time within a single country for a single indicator, in this case, modern contraception usage. Since I am producing separate results for each survey/country and then plotting them on a graph, do I need to create a pooled dataset and de-normalize the data?

Thanks!


Re: Do I need to De-normalize and Pool Data for Trends Over Time in a Single Country? [message #5539 is a reply to message #5535] Wed, 03 June 2015 15:11 Go to previous messageGo to next message
Reduced-For(u)m
Messages: 292
Registered: March 2013
Senior Member

You do not need to adjust the weights in any way if you are just calculating individual survey-round means/proportions. You are not using information across surveys, so there is no problem.
Pooling Data for Trends Over Time in a Single Country [message #6817 is a reply to message #5535] Mon, 20 July 2015 08:12 Go to previous messageGo to next message
Abisiku is currently offline  Abisiku
Messages: 8
Registered: June 2014
Location: Lusaka
Member
I want to look at trends & determinants within a single country for child marriage. What steps do i need to follow.

ENSI
Re: Pooling Data for Trends Over Time in a Single Country [message #6978 is a reply to message #6817] Fri, 07 August 2015 10:41 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 2537
Registered: February 2013
Senior Member
Following is a response from Senior DHS Sampling Specialist, Ruilin Ren:

Your question on whether you need to create a pooled dataset or not depends on the analysis you are doing. If you are looking at the trends and determinants for several surveys together, from the same country, you may need to create a pooled data sets. If you look at the trends and determinants for each single survey and then try compare the changes on the trends/determinants, then you do not need to pool data together.

So it is essentially depends on your analysis. However, if you do pool surveys together, you need to de-normalize the weight (V005) before pooling. Please refer to the forum for instructions on pooling data.
Previous Topic: Error occurred when svy executed logistic
Next Topic: Weighting after merge
Goto Forum:
  


Current Time: Sat Jun 25 18:47:17 Coordinated Universal Time 2022