Sampling weights when using the Youth Survey
Dear Understanding Society.
I'm having trouble working out which survey weights to use for my empirical analysis.
I'm looking at the association between parental non-standard work schedules and adolescent risky health related behaviours.
Due to the above research question, my sample is restricted to adolescents who completed the self-completion questionnaire in waves 2 and 4, which I have merged with the respective 'indresp' datasets I order to gather background family information etc.
I've read the relevant sections in the user guide pdf I found on the website, but still can't work out which weights I should use!
I assume I need to use longitudinal weights (even if my analysis is only two periods), but can't find appropriate one for the child self-completion (as this is the 'lowest' level of analysis)
I searched this forum prior to sending this email, and found Support #563 ('Weights when using the youth panel'), which has a similar problem.
Should I also use the longitudinal full adult response weight from the second wave of analysis? (in my case the 4th wave)
Or is there a more appropriate wave for my research question?
Health Economics PhD Student, Newcastle University
#1 Updated by Victoria Nolan almost 4 years ago
- Category set to Weights
- Assignee set to Olena Kaminska
Many thanks for your enquiry. I have assigned this to my colleague Olena who will look into it and respond.
Best wishes, Victoria
On behalf of the Understanding Society data user support forum
#2 Updated by Olena Kaminska over 3 years ago
Indeed, you are one of our users who use an unusual combination of waves and questionnaires. We don't provide weights for each specific such combination as there would be a huge amount of potential weights. Yet you can use one of our weights as suboptimal. I suggest you use either psnenub_lw weight or indinub_lw weight whichever has the lowest number of nonzero weights. The longitudinal weights should be picked from the most recent wave that you are using.
Hope this helps,