Dear Understanding Society Team,
I am looking at change in health status if individuals accross wave 2 to wave 5. My variable of interest comes from the longitudinal self-completion questionaire therefore I believe the right weight to use for my fixed-effect models is indscus_lw. However, when I want to run my fixed effect model with this pweight, stata tells me that the weight must be constant within pidp. To which values should I therefore set my weight to? What does a indscus_lw=0 mean?
Thank you for your help,
#3 Updated by Alita Nandi almost 2 years ago
- Status changed from New to Feedback
- Assignee changed from Olena Kaminska to Camille Portier
- % Done changed from 0 to 80
Take the longitudinal weight from the last wave, Wave 5, in your case and assign that to all previous waves for each person.
Longitudinal weights for any wave are non-zero for OSMs (core sample members) who have been continuously interviewed up until and including that wave. It is ZERO for all TSMs and PSMs (those who joined the survey after Wave 1 and not part of the core sample).
Note that the weight you are using does not include BHPS sample members in Understanding Society, that is, the weight will be zero for these individuals. If you want to include the BHPS individuals then use indscub_lw.
Please see Section 3.9.1 in the main survey user guide for further detailed information about weights: https://www.understandingsociety.ac.uk/sites/default/files/downloads/documentation/mainstage/user-guides/mainstage-waves-1-7-user-guide.pdf
#5 Updated by Alita Nandi almost 2 years ago
The longitudinal weights for people who only appear in waves 3 & 4 (or 2 & 4, or 2& 3 etc) will be zero as longitudinal weights are only available for those who continuously respond until that wave. So, Wave 5 longitudinal weights are non-zero for those who have responded in every wave from 1-5 (from 2-5 for BHPS), Wave 4 longitudinal weights are non-zero for those who have responded in every wave from 1-4 (and 2-4 for the BHPS sample), and so on.