Journal of Official Statistics, Vol.21, No.3, 2005. pp. 367–387
Mixed Mode Designs: Finding the Balance Between Nonresponse Bias and Mode Effects
Robert J.J. Voogt and Willem E. Saris
Abstract:Mixed mode designs in survey research can be used to lower nonresponse rates and prevent nonresponse bias. The main disadvantage is that the use of different modes of data collection can bias the results. By making a distinction between selection processes and transformation processes, it is possible to model the effects of a mixed mode design on the quality of survey data. Using data gathered on the occasion of the 1998 Dutch national elections, we show that mixed mode surveys and adding the possibility for nonrespondents to answer so called central questions, leads to higher response rates and to less nonresponse bias. Although we do find a higher level of response bias for the face-to-face mode as compared to the telephone and mail modes, the inclusion of the respondents that answered the face-to-face interview still leads to better estimates of the true population values than when these people are left out.
Keywords:Mixed mode data collection, election research, survey research
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group