Understanding Society User Support

Hi there,

I hope this message finds you well. I am trying to link respondents in the COVID-19 Youth postal surveys (cd_youth_p, cf_youth_p, ch_youth_p) to the files on their 'parents' in the adult individual response files (ca_indresp_w, cb_indresp_w ... ci_indresp_w). I basically want to be able to link in information on the parents and households of respondents in the COVID-19 youth surveys. However, I am struggling to match all respondents in the youth files to their parents in the adult files, but may be going about it all wrong.

Taking the wave 4 (cd_youth_p) youth file as an example, as far as I can see, the youth file only contains two identifier variables: pidp_c and cd_aid_dv. However, the adult file (e.g., cd_indresp_w) only contains pidp and cd_aid_dv.

However, I found that the CHILD files (e.g., cd_schild_w) contain a pidp_c. They also contain pidp_m and pidp_f, the identifiers for a child's 'mother' and 'father'. So, using this I created a new pidp variable:

clonevar pidp = pidp_m
    replace pidp = pidp_f if pidp_m==-8

Which hopefully gives me a complete pidp identifier for a child's parent.

Using this new identifier, I have found I am able to perfectly merge the child file (e.g., cd_schild_w) to the adult file (e.g., cd_indresp_w) i.e., all children in the child file are matched to an adult in the adult file, using the pidp identifier.

So, now I have a file that contains both pidp_c for the child and pidp for the adult. My plan was to use this to merge with the YOUTH file (cd_youth_p). Therefore, I would match the pidp_c from the YOUTH file with the pidp_c in the new merged adult/child dataset, to get the pidp for the respondents in the YOUTH file.

However, the problem I'm having is that there appears to be respondents in the YOUTH file (cd_youth_p) that are not present in the child file. So, if I merge the youth file (cd_youth_p) with the child file (cd_schild_w) based on pidp_c, there are n=416 cases in the YOUTH file that are not merged with the child file.

Even if I merge the YOUTH file (cd_youth_p) with all available child files (cx_schild_w, where x=a, d, e, f, g) I am still left with n=237 cases from the youth file that don't appear to have equivalent records in the child files (as a note, I'm focusing on the web surveys - not telephone). As such, there are around n=237 cases in the YOUTH file for whom I cannot get their pidp, and thus cannot link them to the adult files. If I take the pidp_c of one of these unmatched cases and search for it through all the CHILD files I can't find equivalent records.

I may be going about this all wrong. As I mentioned at the start, my primary aim is link records in the YOUTH files to information in the ADULT files on their 'parents' and 'households'.

Any help or guidance would be hugely appreciated.

Thanks so much in advance,

James