It can be observed that a lot of opinion research is currently carried out through the Internet. Following Couper’s typology (2000), one of the eight online survey types is the volunteer panel of Internet users. This opt-in online access panel can be defined as a ‘previously recruited poll of respondents who have agreed to take part in surveys’ (Stenbjerre & Laugesen, 2005). The commonly known advantages and disadvantages of online surveys over other modes also apply for this type of Internet panel research. In a short fieldwork period and with a very low cost, a large group of people can be accessed, of which the answers can be rapidly collected and analysed because of an extensive automation. However, besides these benefits online research has to cope with some important methodological problems which are mainly associated with coverage (possible bias because a group of people does not have Internet access) and sample selection (no known chance of selection, possible bias from recruiting methods) (Bandilla et al., 2003; Dillman, 2000; Couper 2000). According to Couper (2000) it is most problematic when surveys based on volunteer panels of Internet users claim they are representative of a more general public.

The primary research question of the paper is to what degree data collected by an opt-in online access panel are representative for the general population. To address this question, an empirical comparison is performed between the data obtained by a self-administered online survey with panel members on the one hand and the answers obtained by a face-to-face interview with a sample of the general population 151 on the other hand. The online panel under investigation is composed of respondents who are partly selfselected (e.g. pop-ups) and partly recruited by non-probability based methods (e.g. client databases), both online and offline. Based on personal background and contact information gathered through a registration procedure, a database of potential respondents is established from which panel members are sampled. The online panel survey and face-to-face interview are compared based on questions with identical wordings, asked during a similar time period in the same geographic region, namely Flanders in Belgium. Based on findings of previous research (Bandilla et al., 2003) it can be expected that Internet users who substantially differ from the general population in socio-demographic terms and in their interest levels to participate in surveys, they answer differently on various opinion questions. The degree and direction of these differences are considered.