| House hold sample weight [message #10619] |
Thu, 18 August 2016 19:46  |
bakerchowdhury
Messages: 25
Registered: April 2014
|
Member |
|
|
Hello,
I am working on 2014 Kenya dataset. When I run univariate frequency using the Women's individual sample weight variable (V005), I am getting frequencies with decimal points. For example Urban 12690.37, Rural 18388.63 cases.
However, the total number matches with the report. I am a bit confused in reporting the number (univariate frequency) in the manuscript because univariate frequency for individuals can't be a fraction.
Please suggest how can I deal with this issue.
Note: I created a new weight variable dividing the original variable (v005) by 1000000.
Regards,
Muhammad Abdul Baker Chowdhury
|
|
|
|
| Re: House hold sample weight [message #10657 is a reply to message #10619] |
Fri, 26 August 2016 07:44  |
Bridgette-DHS
Messages: 3230
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
DHS weights are calculated so that the average weight is exactly 1, and the total number of weighted cases (after dividing by 1000000) is the same as the total number of UNweighted cases.
Most subtotals of the weighted cases will have digits to the right of the decimal, as you are finding. Normally they are rounded to the nearest integer during the preparation of the reports. For example, your urban and rural numbers add to an integer, which is the sample size, but will not themselves be integers. Because individuals come in whole numbers, as you say, those two numbers would usually be presented as their rounded values, 12,690 and 18,339, respectively.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
