| Sampling [message #13058] |
Tue, 12 September 2017 23:24 |
Glory
Messages: 17
Registered: April 2017
Location: Newcastle
|
Member |
|
|
Dear DHS,
1. I am aware that the complete listing of all households within selected clusters for the survey is not readily available. However, since the sample clusters were selected from a complete list of all clusters from the most recent population census, is it possible to know: i) the total number of clusters in each country as estimated from the population census? ii) the percentage (range) of clusters randomly selected for inclusion in the survey for each country and if that varies over time.
2. I am also considering the possiblity of accouting for the geographic displacement in a Bayesian model to improve the predictive accuracy at cluster level (small area estimation). Kindly advise.
3. Gelman suggests the application of weight trimming in weight-adjusted models to improve model predictive accuracy. I would like to know if this applies to DHS as well.
4. Kindly advise on DHS survey Bayesian hierarchical modeling, weight considerations and the use of objective priors to improve predictive accuracy of models.
Many thanks
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
