weighting youth data
I have a quick query regarding the analysis of the UKHLS data - I have been running analysis using the sample weightings provided in the dataset, but have received a query over whether I also need to include variables to account for clustering, and stratification in the sample. Do the sample weightings provided account for these factors, or do they need to be taken account of separately? (I am running an analysis of the youth self-completion data.
#1 Updated by Redmine Admin over 8 years ago
- Category set to Data analysis
- Target version set to M1
- % Done changed from 0 to 70
Thank you for your questions. Yes, one needs to use all three (weights, strata and clusters) in any analysis. The best command for this is svyset and 'svy:' for analysis. Stratification, clustering and weighting reflect three different aspects of how panel members were selected (and how they respond).
With youth data you may get into the problem that the command will not run as stratification is too detailed. If this happens please combine the adjoint numbers for stratification. What you shouldn't have is one person per strata, and because youth is a subset of our sample, this may happen.