Pooled Cross-Sectional Analysis
We are aware that some users wish to pool data from multiple waves for the purpose of cross-sectional estimation (for example, to increase the sample size for a small population subgroup, or to obtain an estimate relating to a calendar year).
We have prepared the attached note, explaining when and how this can be done. We hope this helps.
Updated by Peter Lynn almost 4 years ago
- Assignee deleted (
For this kind of data I see no conceptual logic in pooling the waves for cross-sectional analysis. You could use each wave separately for cross-sectional analysis. Alternatively, longitudinal analysis is a possibility, in which case you want to analyse the sample of people who participated in both waves 2 and 4. Multilevel analysis may also be a possibility (with responses clustered within individuals), but unless you restrict analysis to the longitudinal sample as above you may get estimation problems from cluster samples of size 1.
Hope that helps,