The DHS Program User Forum - RDF feed
https://userforum.dhsprogram.com/index.php
Representativeness
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=11181&th=5648#msg_11181
could you please clarify a question of representativeness of your surveys? You are stating that the surveys are nationally representative. Does this only apply for women aged 15-49 or does it apply for the entire household? So nationally representative of which survey population?
Thank you for your help,
Jana]]>Jana2016-11-16T09:55:24-00:00Re: Representativeness
https://userforum.dhsprogram.com/index.phpindex.php?t=rview&goto=11182&th=5648#msg_11182
Quote:
The surveys are representative of the population living in households, regardless of age, sex, etc. The only limitation is that the household must include at least one person who is age 15 or above. We do not have IRB approval to obtain information directly from anyone under 15.
All DHS surveys are the same in this respect. All of them are nationally representative of the household population, subject to the restriction that there must be at least one person in the household who is age 15 or above. Phrased another way: All DHS survey samples are selected so that households, and the members of those households (which includes women and men who are interviewed) are representative of the entire country (those living in households, that is). They are also representative of urban and rural areas, and provinces or regions. Please read the Appendix at the end of final reports that describes the survey sampling procedure, or parts of Chapter 1 that also address the issue of how the sample was selected.
There are of course some dwellings for which no respondent can be found, and some refusals. The survey reports provide information about nonresponse. For some surveys, some regions have been skipped, typically for security reasons. For example, most, if not all, of the Mali surveys have some omitted areas, not always the same areas. The reports are very specific about any such omissions. Is that what you are talking about?
The data files always include sampling weights to correct for under-sampling and over-sampling of various strata and to correct for nonresponse. Estimates in reports are always weighted and we recommend that users use the weights.