Abstract
Objectives
To describe the sample design and weighting procedures used in the Saudi National Mental Health Survey (SNMHS).
Methods
A multistage clustered area probability design was used to select the SNMHS sample with one male and one female KSA citizen ages 15–65 surveyed in each sample household.
Results
A design representative of the household population was developed and modified iteratively to adjust for unanticipated field complications. These modifications, along with variation in within‐household probabilities of selection and geographic–demographic variation in response rates were accounted for through survey weights. Design‐based estimation methods were used to adjust for the effects of these weights and of geographic clustering. Design effects were estimated and simulations were carried out on bias‐variancetrade‐offs in weight trimming to evaluate the implication of design features for precision of estimates.
Conclusions
The multiple purposes of the survey will require the use of different weights for different types of analyses, including household and person weights as well as weights for proxy reports about household members whose disabilities prevented them from participating in the survey. It will be important to use these different weights appropriately in the diverse analyses that will be undertaken with the SNMHS data.