Individual weights of those whose household questionnaires are not available
According to my understanding, the individual weights take into account predicted household response probability. However, I find that the weights for those individuals whose household questionnaires (w_hhresp) are not available are not zero (e.g. people whose household ids are 1297691602 1362862802 1436119202 in wave 2). I am wondering how to model the household response probability when their household questionnaires are not available?
Updated by Olena Kaminska about 1 year ago
I have checked for you and for the households id's that you mentioned we provided b_psnenub_xw b_psnenus_lw b_psnenus_xw weights, the other weights have correctly the value of 0 as these households originate from UKHLS EMB sample.
Could you clarify what you are trying to do? Is your research related to creating weights / looking into nonresponse; or are you studying a substantive topic?
Updated by Louise Luo about 1 year ago
Thank you very much for your reply.
I am creating my own cross-sectional individual weights. I want to adopt the method similar to the one used to create b_indpxus_lw.
Correct me if I am wrong.
To create b_psnenus_lw, a_psnenus_xw is multiplied by an inverse probability. This inverse probability is predicted by logistic regression. The predictors are obtained from the household grid and household questionnaire. The household grid is in b_indall and the household questionnaire is in b_hhresp.
There are some households whose household questionnaires are not available (eg. b_hhresp does not include those households whose household id is 1297691602, 1362862802 and 1436119202). In this case, how to estimate the regression for these individuals whose household questionnaires are unavailable?