Going Online with a Face-to-Face Household Panel: Effects of a Mixed Mode Design on Item and Unit Non-Response

There are considerable cost and timeliness advantages associated with web interviewing, compared to interviewer administration. However, web surveys do not perform well in terms of coverage and participation. To harness the strengths of both modes, existing probability-based interviewer-administered surveys are therefore being pushed to consider a mixed mode approach, including web. We assess the effect of introducing web interviewing as part of a mixed-mode design in the context of an existing longitudinal survey in which sample members have previously been interviewed face-to-face. Using experimental data from a household panel survey in the UK, we find that the mixed mode design resulted in a lower proportion of households fully responding. However, more than one in five households fully responded online. Overall, individual response rates were also lower with the mixed mode design, and we were unable to identify any subgroups where the reverse was true. Also, item nonresponse rates were higher with the mixed mode design.