Conferences

We report on large-scale randomised experiments carried out to study possible methods of reducing panel attrition. Specifically, the experiments address issues asssociated with a) ability to locate sample members and b) willingness of sample members to continue to participate. The experiments were carried out in 2008 between waves 17 and 18 of the British Household Panel Survey, involving a sample of around 12,500 persons.

The first experiment extends ideas in Couper and Ofstedal (2009) concerning the process of tracking sample members. It involved change-of-address notification cards, with four treatment groups. One group received no card, the other three groups received a card with different types of incentive/reward for returning it. The outcomes at the subsequent survey wave were be systematically recorded: not just response outcome but also the costs of field tracing resulting from un-notified changes of address. We will assess relative costs of alternative procedures as well as impacts on non-response bias. We will consider also the possibility of developing tailored approaches based on previous wave tracing history and other indicators.