Thank you for your previous support. Your help really means a lot to me.
I am currently looking at the non-respondent adjustment at individual level. The user guide says (p.75):
"The predictors used in the models include all the predictors used for the household level nonresponse models and individual and household-level variables obtained from the household questionnaire, such as age and gender, marital and employment status, household size and presence of children in the household, as well as household expenditure on food and food outside, consideration of use of environmental energy, among others."
However, I cannot find some of these variables, such as household expenditure on food and food outside, consideration of use of environmental energy, household size in the w_indall.dta file. Does UKHLS provide this non-respondent information for the users to measure the weights themselves, or user can only use the available variables in w_indall.dta?
Thank you very much.
#2 Updated by Olena Kaminska 11 months ago
Thank you for your question. Indeed, the variables in indall are only available for respondents in the current wave. We use these variables from wave t to predict nonresponse in wave t+1. Also, some of the variables listed are available in some waves but not the others. We use these variables where they are available.
Hope this helps,
#6 Updated by Olena Kaminska 11 months ago
There are a number of levels of nonresponse in our data:
1) household nonresponse at wave 1 (where some households never participate in our panel)
2) conditional on that individual nonresponse (i.e. even if household participates not everyone provides an interview). This nonresponse happens for youth questionnaire, and separately for adult questionnaire (we differentiate between proxy + individual response, individual response, and self-completion response)
3) and then conditional on wave 1 response there is response in wave 2 and so on.
So, w_indall includes information that is provided by one person in the household about the whole household. This includes people who don't respond themselves (and therefore not present in indresp.dta) as for these people the household information is shared. If we predict response in indresp.dta conditional on response to the household questionnaire (indall.dta) we use the predictors from the household questionnaire, for example.
Hope this helps,