Mixed modes and measurement error: using cognitive interviewing to explore the results of a mixed modes experiment
Mixed-mode survey designs: small differences in protocol can make a big difference to panel attrition