Understanding data linkage consent in longitudinal surveys
Jul 26, 15:20
As new sources of data emerge, such as data generated by social media or private companies, there is increasing interest in supplementing data collected in probability sample surveys with such data. In many countries survey data can only be linked to administrative or other process generated data if survey respondents give informed consent to the linkage. We build on the review chapter by Calderwood and Lessof in the previous MOLS book, which focused on practical, legal and ethical issues. In this chapter we focus on how respondents in longitudinal surveys make the decision whether or not to consent, and how this decision is influenced by the mode of data collection.
We draw on both quantitative and qualitative analyses. We first analyse consent questions in Understanding Society and the Innovation Panel, asked over time and in different modes. Second we report on results from qualitative in-depth interviews carried out in spring 2017 with 25 members of the Innovation Panel sample.
Initial quantitative analyses show that respondents are up to 30 percentage points less likely to give consent when they complete the survey online than when they complete a face-to-face interview. This appears to be due to the mode affecting willingness to consent, not due to the type of people responding online. In addition, up to 50% of respondents who do not give consent do consent when asked again in a later wave. This suggests that the consent decision is not a strong or fixed decision, but can likely be influenced.
Initial results from the qualitative interviews show a lot of misunderstanding of what the consent question is asking: about how personal data would be used, by whom and for what purposes. Factors influencing the respondent’s decision whether or not to consent include personality traits (such as openness versus suspiciousness), attitudes towards data sharing, assessment of the benefits and risks of granting access to personal data, and trust in the survey organisation. The cognitive challenge posed by the consent question plays an important role, with respondents indicating that they would be less likely to give consent late at night, after a busy day, or after a long interview. The presence of an interviewer is seen as key in establishing trust about data security and providing the opportunity to ask questions.
The chapter will end with a discussion of practical implications of the quantitative and qualitative findings, and a discussion of further research needs on methods to increase informed consent.