| Re: Variables representative at the new province level for DRC (2013/14) [message #11513 is a reply to message #11512] |
Thu, 05 January 2017 15:03   |
Bridgette-DHS
Messages: 3230
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
Representativeness is a property of the sample, not of specific indicators. It was misleading for the report to include "the majority of". All of the indicators are representative of the population and the population and stratum level (the strata are combinations of province and urban/rural). However, and I think this is why "the majority of" was included, some indicators will have wide confidence intervals. These will be indicators that are calculated on relatively few cases--for example, very young children who may or may not be breastfed, children who had symptoms of illness and may or may not have been taken for treatment. Such estimates have a lot of statistical uncertainty--but they are still unbiased.
If you look at the Appendix on sampling errors, usually Appendix B, you can see the width of the confidence interval, and use that as a guide to which indicators are least stable.
|
|
|
|
Members
Search
Help
Register
Login
Home
