It is important to investigate how respondents ‘stay on board’ as panel members, because of the costs involved in replacement of panel members and because high long-term response rates are important for research purposes (representativity and longitudinal analyses). In this paper, we report a study about motivation to participate in a panel and in panel questionnaires. The panel we studied is an Internet based panel, the socalled “CentERpanel”, which consists of about 2000 households in the Netherlands. Every week, the members of those households complete a questionnaire on the Internet, with panel member without Internet access completing their questionnaires with a so-called set-top box. The main motivation of respondents to participate in the CentERpanel is to help others (researchers), to give their opinion, to learn things, or to help society. Surprisingly, questionnaire characteristics such

as length, layout, and difficulty are considered less important than privacy guarantee and scientific relevance. In line with that, the reasons people gave for quitting the panel have little to do with the difficulty or variation in the questionnaires. Most often, respondent quit for personal reasons such as lack of time or they just do not feel like it anymore. Furthermore, we compared long-term panel members (in the panel for over 437 weeks) with shortterm panel members (in the panel less than 179 weeks). The response pattern of these two groups differed on several points: Long-term panel members more often say they always fill out their questionnaires on the same day and time, whereas short-term panel members do not have a fixed moment for this; they say they fill out questionnaires when they happen to think about it and interrupt questionnaires more often. In addition, the short-term panel members more often motivate their participation on monetary grounds (to earn “CentERpoints”). Summarizing, it appears that motivations to participate are somewhat more intrinsic and habitual for long-term respondents than for short-term respondents.