Weights for youth dataset
I'm looking to conduct a longitudinal analysis using the youth data from waves 1-10 from Usoc. I'm looking to explore how social media usage at age 13 relates to mental health outcomes at age 15.
Following your advice here https://iserredex.essex.ac.uk/support/issues/1472 I thought the best weight to use would be the suboptimal weight - longitudinal enumeration weight from the last wave of the analysis(j_psnenus_lw). However, 5,855/13842 young people are missing the weight. So I think should perhaps create my own weight to increase the sample size and I was a little unsure how to do this and was wondering if you could help?
One approach I thought might work would be to include the psnenus_lw weight from the last wave each young person participated in but I’m not sure this is ok to do?
Thanks in advance for your help.
Updated by Olena Kaminska about 2 years ago
Yes, technically you will be pooling young people from different years. As a suboptimal weight try to use psnenus_lw weight or psnenub_lw weight when they are 13 at wave 2 or later. Some of your 0s are because of the boost, but you can use boost only if you have enough years, so you can use ui weight only for those who are 13 or older at wave 6.
You can improve your weight by additional modelling. Email us to request the training material.
Updated by Ruth Plackett about 2 years ago
Thanks for your reply it was really helpful. I've emailed user support to request the training material as you suggest. I had a further question related to the previous query - can I use the cross-sectional weight psnenus_xw for those who 13 at wave 1?
Updated by Ruth Plackett almost 2 years ago
As you've suggested I have used psnenus_lw weight when they are 13 at wave 2 and so on and ui from wave 6 to include the boost. I was just wondering what suboptimal weight could be used for those who were 13 in wave 1 and 15 in a following wave – would it be OK to use their psnenus_lw weight from the wave when they were 15?
Thanks in advance.