This experiment compares the impact of placement and wording of consent to data linkage to HMRC records. This consent question was originally carried in the mainstage of Understanding Society. This experiment varies the wording to this question to explore the standard version used in the mainstage survey and a new, “easier” to understand wording to ascertain the impact of this has on consent rates. These versions were varied with equal allocation across respondents in both web and face-to-face versions of the survey. For those responding face-to-face, the placement of this consent request also varied, being asked either early or late in the questionnaire. Respondents were provided with additional information in the form of a leaflet with information and one with a flowchart on the linkage process and usage of data.
Random allocation to treatment occurred at the respondent level, and was stratified by response modes. Respondents were allocated to either standard or easy versions within modes to ensure a near 50/50 split within each. Within the face-to-face mode only, respondents were also allocated using a 50/50 split to being asked the consent question either early or late, which was stratified by question difficulty. After the question was allocated as easy or standard (within face-to-face), then the location was allocated for this version, to ensure another close to 50/50 within mode and question difficulty. The controlling variables for this experiment are:
condiffcawi on record k_indresp_ip:
1 Standard HMRC consent question for CAWI
2 Easy HMRC consent question for CAWI
condiffcapi on record k_indresp_ip:
1 Standard HMRC consent question for CAPI
2 Easy HMRC consent question for CAPI
conlocstand on record k_indresp_ip:
1 Ask HMRC standard consent early in CAPI
2 Ask HMRC standard consent late in CAPI
conloceasy on record k_indresp_ip:
1 Ask HMRC easy consent early in CAPI
2 Ask HMRC easy consent late in CAPI
The variables used for this experiment are the following.
On record k_indresp_ip:
consentq1, consentq2, consentq3, consentq4, intcread1, intflread1, intcread2, intflread2
on record k_keystroke_paradata:
k_keystrokes1, k_keystrokes2
At Wave 15, a follow-up experiment was conducted to examine reasons why consent rates are so much lower in web than face-to-face surveys. Respondents were asked a single consent question (tax record linkage) with some follow-up questions about how the respondent made the consent decision. Some of these questions were previously implemented in IP11. This experiment exploited the mixed-mode design of IP15 (Web and CAPI), as well as the self-completion element (CASI) within the CAPI interview. The follow-up questions were all asked at the beginning of the CASI module, immediately after the consent question.
Face-to-face respondents were allocated to one of three groups, varying when the consent question was asked:
- CAPI: the consent question was the last CAPI question before the start of the CASI module,
- CASI: the consent question was the first in the CASI module, and
- partial CASI: the consent question is the first in the CASI module, but the question is printed on a showcard and read by the interviewer and not displayed on screen, the respondent enters the answer in CASI.
Web respondents were allocated to one of two groups, varying how the consent question was asked:
- the consent question included a photo of, and message from, the study director,
- the control group were shown only the text of the consent question.
The variables controlling allocation to treatments, made equally across conditions at the household level, are on record o_hhsamp_ip:
ff_consentcasiw15 (1/3 each, allocation stratified by sampleorig ff_gridmodew15 ff_incentw15)
1 = CAPI
2 = CASI
3 = Partial CASI
ff_consentwebw15 (1/2 each, allocation stratified by ff_consentcasiw15 sampleorig ff_gridmodew15)
1 = Consent question only
2 = Consent question includes photo of, and message from, study director
The variables used for this experiment are the following, in file o_indresp_ip:
hmrcconsentq2, coundstd2b, coundstd2e, coundstd2f, coundstd2g, scac2, scac, scrx_code, scux_code, hmrcconsentq1, hmrcconsentq3, hmrcconsentq4, cdcsn1, cdcsn2, cdcsn3, cdcsn4, cdcsn5, cfactors1, cfactors2, cfactors3, cfactors4, cfactors5, cfactors6, cfactors7, cfactors8, cfactors9, cfactors10, scrf1, scrf2, scrf3, scrf4, scrf5, scrf6, scrf97, scun1, scun2, scun3, scun97
