Allocation to treatment groups was of households within PSUs.
Group 1 Version A: “Is that still the case?”
Group 2 Version B: “Has that changed?”
The variable that controls allocation is c_ff_changew2 on record c_hhsamp_ip. Relevant variables are c_sf1_a, c_sf1_b, c_jbterm1_a, c_jbterm1_b, c_jbhrs_a, c_jbhrs_b, c_jshrs_a, c_jshrs_b on record c_indresp_ip.
At Wave 4 the experiment was repeated with the same allocation to treatment groups. The refreshment sample was not asked dependent interviewing questions, since we did not have any wave 3 responses for this sample. Again, the experiment was audio-recorded.
The variable that controls allocation is d_ff_changew2 on record d_hhsamp_ip. Relevant variables ared_sf1_a, d_sf1_b, d_jbterm1_a, d_jbterm1_b, d_jbhrs_a, d_jbhrs_b, d_jshrs_a, d_jshrs_b on record d_indresp_ip.
In Wave 5 and 7, the experiment contrasts the two question formats with similar questions not answered by “yes” or “no”. Furthermore, this work exploits the mixed mode experiment to study potential response order effects, and whether these differ in CAPI and web, by varying the order of response options in the new format. Respondents were assigned to one of four experimental question variants: (1) “Still the case? Yes/No”, (2) “Has this changed? Yes/No” (3) “Still the case or has this changed?” and (4) “Has this changed or is it still the case?”. The variable that controls allocation is w_ff_diw5 on record w_hhsamp_ip. This variable is coded as:
1 Still the case
2 Has this changed
3 Balanced, still first
4 Balanced, changed first
The variables used for this experiment are – on record hhresp:
hsroomchk_a, hsroomchk_b, hsroomchk_c, hsroomchk_d, hsowndchk_a, hsowndchk_b, hsowndchk_c, hsowndchk_d, xpmg_a, xpmg_b, xpmg_c, xpmg_d, rentchk_a, rentchk_b, rentchk_c, rentchk_d
On record INDRESP:
lkmove_a, lkmove_b, lkmove_c, lkmove_d, edtype_a, edtype_b, edtype_c, edtype_d, jbterm1_a, jbterm1_b, jbterm1_c, jbterm1_d, jbsic07_a, jbsic07_b, jbsic07_c, jbsic07_d, jbsoc00_a, jbsoc00_b, jbsoc00_c, jbsoc00_d, jbsemp_a, jbsemp_b, jbsemp_c, jbsemp_d, jbsizechk_a, jbsizechk_b, jbsizechk_c, jbsizechk_d, jbhrschk_a, jbhrschk_b, jbhrschk_c, jbhrschk_d, paygl_a, paygl_b, paygl_c, paygl_d, paynl_a, paynl_b, paynl_c, paynl_d, paytypchk_a, paytypchk_b, paytypchk_c, paytypchk_d, wktravchk_a, wktravchk_b, wktravchk_c, wktravchk_d, jshrschk_a, jshrschk_b, jshrschk_c, jshrschk_d, jspartchk_a, jspartchk_b, jspartchk_c, jspartchk_d, jstravchk_a, jstravchk_b, jstravchk_c, jstravchk_d
Note that the Wave 5 implementation of this experiment was corrupted. (Further details can be found in the section Known Data Issues.) The experiment was therefore repeated in Wave 7.
