| Benin DHS 2017-18 apparent urban/rural sample size discrepancies [message #24576] |
Thu, 02 June 2022 12:36  |
andrew.M4M
Messages: 5
Registered: June 2022
|
Member |
|
|
Hello,
I'm attempting to validate the the asset variable descriptive statistics listed in the Benin DHS 2017-18 Excel file containing the principal components analysis scores used to construct the survey's wealth index against those found in the survey data. While the means, standard deviations, and sample sizes listed in the common tab of the document correspond to those found in the household recode dataset, they do not for the urban and rural tabs. While the Excel document's urban and rural tabs samples total the survey's overall sample size their sub-samples differ from those found in the household recode. Here's what I'm seeing:
Excel document:
Urban - 6,715
Rural - 7,441
Total - 14,156
HR dataset (from variable hv025):
Urban - 6,364
Rural - 7,792
Total - 14,156
It appears as if 351 rural residents were reclassified as urban during the wealth index creation process. Any idea what could have caused this? This also raises the question as to which urban/rural breakdown the wealth index factor scores provided in the data set are based on.
Many thanks,
Andrew
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
