Rural-Urban sample weights [message #11404] |
Tue, 13 December 2016 05:44 ![Go to next message Go to next message](/theme/default/images/down.png) |
Shek
Messages: 19 Registered: November 2014
|
Member |
|
|
I am trying to do an urban-rural comparison of factors associated with antenatal care utilization in Nigeria. I splitted the Nigeria data set into 2 data sets based on place of residence namely; an urban and a rural data set. I intend to run analysis seperately on the urban and rural data set (chi square and complex sample logistic regression analysis) and subsequently compare what factors are significantly associated with antenatal care utilization in urban and rural Nigeria based on the rural-urban data sets.
My question relates to the weights. Since i am spliting the data set into 2 (urban and rural Nigeria) and running my analysis seperately on each, do i need to compute a new weight for the urban and rural data sets or can i use the weights as given in the survey. In doing the complex sample analysis will there be a need for any modification to the strata and cluster variables.
Thanks.
|
|
|
Re: Rural-Urban sample weights [message #11405 is a reply to message #11404] |
Tue, 13 December 2016 07:52 ![Go to previous message Go to previous message](/theme/default/images/up.png) ![Go to next message Go to next message](/theme/default/images/down.png) |
Bridgette-DHS
Messages: 3139 Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
Quote:You do not need to adjust the weights, and you do not need to make any changes to svyset. The clusters are nested in the strata. The strata are either all urban or all rural, so your urban and rural subsamples have completely different strata and clusters.
Even if you constructed subsamples in a more general way, for example based on age or education, etc., you would not need to modify the weights or svyset. In effect, the adjustments for the survey design (with weights and svyset) will restore the sample to an equivalent simple random sample (srs), and any sub-sample of a simple random sample is another simple random sample, from whatever sub-population you may identify.
|
|
|
|