Understanding Society User Support

Dear Lydia,

w_livpar is calculated from the relationship grid during the interview, and is then used during the interview to route through or past the appropriate questions. Generally, the grid gets completed at the beginning of the interview sequence, once per household. However, if there is an issue during a later individual interview, the interviewer may go back to the grid and correct the information. If they then do not re-do the whole module, the derived variables w_livpar and w_livsp are not recalculated. An example may be where someone is asked about their child in the household when they do not have a child, or about their partner when they do not have a partner.

For households which have previously responded to the survey, interviewers have the previous wave info preloaded and some of this will have been cleaned so there is less scope for inconsistencies. IEMB respondents who joined the survey in Wave 6 would not have had this information so there were likely to be more enumeration gaps in the Wave 6 household grid. These people's information was entered for the first time and the households have a greater chance of being larger. I also see that the are a high number of 'refused' in this group.

In the w_egoalt file, are you using the variable w_relationship or w_relationship_dv? w_relationship is the data which was collected and w_relationship_dv has been manually corrected in cases where it did not make sense, so the latter is more likely to differ from the data in w_livpar.

Best wishes,
Stephanie Auty - Understanding Society User Support Officer