| Survey Weight [message #31381] |
Sun, 07 September 2025 08:22  |
Nahid
Messages: 10
Registered: November 2020
|
Member |
|
|
I am leading a survey similar to DHS. The sample was selected using a two-stage design: 600 clusters were chosen in the first stage, and in the second stage 20 households were sampled within each cluster.
Although 600 clusters were selected, data collection was completed in only 530 clusters.
Could you please advise me on the appropriate weight calculation in this situation? Is any adjustment for the population or non-response required, and if so, how should I implement it?
Thank you in advance for your kind guidance.
|
|
|
|
| Re: Survey Weight [message #31546 is a reply to message #31381] |
Tue, 28 October 2025 01:00  |
OpalTromp
Messages: 1
Registered: October 2025
|
Member |
|
|
|
In your two-stage survey, weights should be calculated based on the probability of selecting clusters and households (basic weights). Since there are 70 clusters that did not complete the survey, the weights should be adjusted for cluster-level non-response rates. If there are households in the responding cluster that did not respond, an adjustment for household-level non-response rates should be added. Finally, if population-wide data are available, population-based adjustments (post-stratification) should be performed to ensure a representative sample. This helps the weights reflect correct sample design and reduce non-response bias.
|
|
|
|
Members
Search
Help
Register
Login
Home
