Recent decades have seen a shift away from surveys in which all procedures are standardised towards a variety of approaches (tailored, responsive, adaptive) in which different sample members are treated differently. A particular variant involves applying to subgroups targeted design features that are identified in advance of field work and are not subsequently modified. Targeted designs have mainly been implemented on panel surveys and mainly to address non-response. I will provide a framework for targeted designs and will briefly discuss their objectives, review their development to date, and outline possible future developments. Then, I will present a study in which the targeted feature is the data collection protocol (computer-assisted personal interviewing (CAPI) single-mode or web-CAPI sequential mixed-mode). We compare four alternative methods for allocating sample members to one of the two data collection protocols. The methods differ in terms of the quality measure that they aim to optimise: response rate, R-indicator, coefficient of variation of the participation propensities, or effective sample size. We use data from an experiment on the Understanding Society Innovation Panel. For each allocation method we predict outcomes in terms of several quality measures. Although allocating the whole sample to single-mode CAPI results in a slightly higher response rate than allocating the whole sample to the mixed-mode protocol, we find that two of the four targeted allocations achieve a better response rate than single-mode CAPI at a lower cost, as around one-fifth of all responses are obtained online. We also find that all four of the targeted designs out-perform both of the single-protocol designs in terms of representativity and effective sample size.
Presented by:
Peter Lynn, ISER
Date & time:
January 18, 2017 1:00 pm - January 18, 2017 2:00 pm
Venue:
2N2.4.16, Large Seminar Room
Internal seminars home