Online access panels for surveys on public health and epidemiology

In both the private and public sectors research on attitudes and behaviour is often carried out using “online access panels” – panels sometimes consisting of hundreds of thousands of members of the public who have agreed to answer questionnaires on a variety of topics. However, such panels are currently little used in health care research, perhaps a surprise given Van Gelder, Bredveld and Roeleveld’s (2010) comment “...web-based questionnaires could be considered an alternative or complementary mode in the range of epidemiologic methods of data collection”. Thus in this paper we examine what a typical online access panel is and what it might offer to health care researchers. To set online access panels into context, the potential biases associated with them are examined, and these are compared with the other main modes of data collection. It is argued that an evaluation of the validity of modes of data collection should be based on a number of criteria to decide which mode is most likely to reflect accurately the characteristics of a specific population, in particular its demographics, lifestyle, attitudes and behaviour. Validity is not just a function of the sampling design and the extent and nature of any non-response bias but also of the willingness and ability of respondents to give honest and accurate information. In the case of online access panels the merits of reduced socially desirable responding and greater willingness to provide sensitive information are noted. The positive and negative sides of ‘panel learning’ are also reported. Whether there is one single definitive statement about the relative validity of the different modes is considered. Experimenting with key questions across modes is suggested.