In Wave 4 a split-ballot experiment was used to collect information about the preferred mode of administration for the survey. The ordering of questions about mode preference was randomly varied across two different experimental groups: (a) first, ratings of how likely the survey participant is to respond in specific modes, followed by questions about which mode is preferred and least preferred (generalized preference) vs. (b) a generalized mode preference questions followed by ratings of specific modes. The experiment was repeated at Waves 5 and 6 with the same allocation.
The Wave 4 controlling variable for this split-ballot design is d_ff_group2 on record d_hhsamp_ip and is coded:
Group 1 Ratings of specific modes then generalized preference.
Group 2 Generalized preference then ratings of specific modes.
The relevant variables affected by this experiment are d_mpint_a to d_mpnot_a, and d_mpint_b to d_mpweb_b on record d_indresp_ip.
The controlling variable for this design in Wave 5 is e_ff_group2 on record e_hhsamp_ip, with relevant variables affected being e_mpint_a to e_mpnot_a, and e_mpint_b to e_mpweb_b on record e_indresp_ip.
The controlling variable for this design in Wave 6 is f_ff_group2 on record f_hhsamp_ip, with relevant variables affected being f_mpint_a to f_mpnot_a, and f_mpint_b to f_mpweb_b on record f_indresp_ip.
