Often the survey design decisions regarding how to obtain consent for administrative data linkage to social survey data are based on anecdotal accounts and common sense rather than empirical evidence. This set of experiments and additional data collection examined: (a) the reasons for consenting or not consenting; (b) whether survey context matters in asking consent; (c) the stage in the life of a panel in which data linkage should be performed; and (d) the method of re-asking consent when maintaining consent in a panel is at issue.
The experiment has two factors. First, consent to link to benefit and tax credit records held by the Department for Work and Pensions was asked either just after questions on the amounts received in these forms of unearned income, or at the end of the questionnaire (context dimension). Second, respondents were either reminded of their consent or non-consent given previously and asked whether this consent should still apply (dependent interviewing) or asked independent of any previous consent given (independent interviewing). If no information about prior consent was available, then the consent question was asked independently.
Table: Schematic design of Wave 4 consent experiment
| Independent Interviewing | Dependent Interviewing | |
|---|---|---|
| Ask consent in context | Group 1 | Group 2 |
| Consent asked at end of interview | Group 3 | Group 4 |
The controlling variable for this experiment is d_ff_conexpw4 on record d_hhsamp_ip:
Group 1 – Independent question within context
Group 2 – Dependent question within context
Group 3 – Independent question at end of interview
Group 4 – Dependent question at end of interview
The relevant variables affected by this experiment are d_bncn_a1 to d_bcoi_a, b_bncn_b1 to d_bcoi_b, d_bcsig, d_bcrat1 to d_bcrat9, and d_bcchnge on record d_indresp_ip.
