Mounting experiments on longitudinal surveys adds a further dimension to the value of randomised experiments (designed to maximise internal validity) mounted on probability surveys (to maximise external validity): for example, repeated measurement over time can be used to test effects on inherently longitudinal outcomes, or to test inherently longitudinal treatments. The unique value of experimentation in longitudinal surveys is, however, matched by unique challenges in design and implementation. We summarise key methodological features and challenges based on experiences with the Understanding Society Innovation Panel, a probability-based household panel with annual interviews that exists solely for experimentation and methodological development.
Peter Lynn and Annette Jäckle
(2017) Mounting multiple experiments on longitudinal social surveys: design and implementation considerations, Understanding Society Working Paper 2017-05, Colchester: University of Essex