The next generation of longitudinal studies will likely make extensive use of linked data to augment survey responses. However this is in the context of declining response rates to surveys and in consent to data linkage (Kreuter, Sakshaug & Tourangeau, 2016). To date there have only been a limited number of studies investigating factors associated with consent to link in longitudinal studies and even fewer that incorporate an experimental design (e.g. Sala, Knies & Burton, 2014). In this paper we propose to understand consent to link to administrative data in the context of a representative online longitudinal panel survey, Life in Australia (http://www.srcentre.com.au/our-research/panel).

We propose two experimental treatments, with the main outcomes of interest being consent to link admin data to survey data, as well understanding of consent forms (based on a series of discrete questions afterwards). Our two treatments are as follows:

• Treatment 1 – Type of consent. In longitudinal surveys that extend for decades selective attrition becomes increasingly problematic. In this study we test whether rates of consent to data linkage are affected if participants are asked to consent to data linkage beyond their own survey participation.

• Treatment 2 – Type of linkage. In survey research sensitive topics increasingly have more non-response than other questions, typically questions about financial matters may be subject to higher levels of survey non-response. In this arm of the experiment we test for differences in asking for consent to link to income support versus medical records versus education records.