Support #1204
open
UKHLS zero weight in panel data
100%
Description
Hello. I'd like some advice about weighting variable. I'm using UKHLS waves 1 through 7, individual data.
When I ran svy command in Stata, the number of observations shrink due to zero. I tried iweight but it's not fixed.
svyset psu [pw=my_lw], strata(strata) singleunit(centered)
svyset psu [iw=my_lw], strata(strata) singleunit(centered) // iweight applied for experiment
I am aware that any participant who missed any survey wave, gets zero weight, thus when I copied 2015 weight to the rest of waves, this generated a large proportion of zero weight.
Is there any fix for this, please, and should I? Or should I just report and explain ?
The code below is from user manual.
- my longitudinal weight
tab wave
/*copy the LAST wave's value over to previous waves
gen my_lw = indinus_lw
bysort pidp (year) : replace my_lw = my_lw[_N]
Thank you and I really appreciate your support.
Updated by Alita Nandi over 5 years ago
- Assignee set to Olena Kaminska
- Private changed from Yes to No
Updated by Alita Nandi over 5 years ago
Many thanks for your enquiry. The Understanding Society team is looking into it and we will get back to you as soon as we can.
Best wishes,
Understanding Society User Support Team
Updated by Olena Kaminska over 5 years ago
Hello,
Don't worry about zero values of weights - as long as you use weights and adjust for clustering and stratification your results are representative. You don't need to report about zero weights in your paper either. Just say that you are using weighted analysis.
If you are using waves 1 through 7 your weights should come from wave 7 (unless you are using long format of the data).
Hope this helps,
Olena
Updated by Alita Nandi over 5 years ago
- Assignee changed from Olena Kaminska to Sook Kim
Updated by Stephanie Auty
Updated by Understanding Society User Support Team over 2 years ago
- Category set to Weights
- Status changed from Feedback to Resolved
- % Done changed from 80 to 100