Declining response rates in both cross-sectional surveys as well as panel surveys are of great concern to survey researchers (Groves & Peytcheva, 2008). One increasingly common method of increasing response rates is to switch the mode of administration in order to recruit nonrespondents (de Leeuw, 2005). Most often, these sequential mixed-mode designs use a relatively inexpensive mode for the initial contact attempts, then switch the mode at follow-up to a more expensive mode. While the use of

sequential designs and their impacts on response rates are well-documented in cross-sectional surveys, we know little about their utility with respect to response rates and cost in panel surveys. The Relationship Dynamics and Social Life (RDSL) study is a panel survey that uses a design that initiates each wave in web and follows up nonrespondents in CATI at each wave. Web response rates tend to be lower than CATI response rates (Dillman et al., 2008), although the costs of CATI tend to be far greater than web

(Couper, 2005). Thus, RDSL aims to balance cost and response rate concerns by only using CATI for the more difficult respondents. But panel respondents may be exposed to this sequential design, altering future behavior. This study will examine the impact of RDSL’s sequential design on longitudinal response rates across modes, changes in the mode of completion, and changes in cost.