Mode effects in reporting and response of moved household members in a household panel survey.

To maintain sample size and representativeness in household panel surveys, it is essential that household reference persons provide contact details of household members who have left the household, and subsequently that these household members continue their participation and introduce new households in the sample. As more household panel surveys transition from interviewer-based modes to web-based modes, a key unanswered question is whether web-based surveys are equally successful as interviewer-based surveys in maintaining cooperation of household members who left the household. Without the assistance of an interviewer — who typically provides motivation and social control — there is a risk of significant underreporting of moved household members in the web mode.

In this paper, we compare the telephone and web modes using data from a mode experiment and a refreshment sample of a probability sample household panel survey. The experiment shows no significant difference between the modes, suggesting no mode effects. While this is reassuring, the refreshment sample, — in which households are assigned to modes based on the availability of a telephone number — reveals a different picture. Although households in the web group contain more mobile sample members, household reference persons in the web group are less likely to report new addresses. This is an issue that should be considered when the shift to web is made. More research is needed to find an optimal balance between ease of mode assignment and adequate reporting of moved household members.