Face-to-face, telephone and paper interviewing are well-established data collection modes in survey research. There is a vast literature on how to estimate population statistics using these modes. In the last decade, also web interviewing has become a regular mode for interviewing, and the literature on this mode is growing rapidly. The rise of web surveys also raised the question at many survey organisations whether the standard face-to-face and telephone surveys can be replaced by cheaper combinations of data collection modes while maintaining the same quality level of data. Combining data collection modes, however, affects all survey errors, from measurement error to unit non-response. The question, therefore, translates to total survey design and is not a simple one to answer; it is almost unfeasible to fully disentangle the various errors that contribute to the total survey error while shifting to another data collection strategy. Over the last three years, Statistics Netherlands conducted four pilots that involved both sequential and concurrent mixes of the modes face-to-face, telephone, paper, and web. The pilots were run parallel to existing household surveys. In the presentation, we will discuss the various components of the survey error for the chosen mixed-mode strategies. Furthermore, we will identify a number of research questions

that need attention in the nearby future.