Mixed-Mode Designs and Computer-Assisted Web Interviewing in International and Czech Social Surveys

The cost of interviewer-administered data collection (CAPI, CATI) has been gradually rising, while response rates have been declining over time. This situation has led to the domination of online data collection (CAWI), which tends to suffer from poor population coverage and even lower response rates. This could be overcome by combining this method with another mode of data collection. This paper aims to summarise the main principles of mixed-mode design, its implementation in major international surveys, and the first results of a project testing the feasibility of the push-to-web mixed-mode design in the Czech context. The results illustrate that together, unequal distribution of technological skills and willingness to participate distort the recruitment of random samples for CAWI towards middle-aged and more educated respondents. The mixed-mode design – together with the described components – has the potential to better involve underrepresented groups, while it is essential to follow the rules for achieving comparable answers across modes and devices. Finally, we emphasize the need to disseminate findings to researchers and other users of survey data, as the use of mixed-mode design has been rapidly increasing.The cost of interviewer-administered data collection (CAPI, CATI) has been gradually rising, while response rates have been declining over time. This situation has led to the domination of online data collection (CAWI), which tends to suffer from poor population coverage and even lower response rates. This could be overcome by combining this method with another mode of data collection. This paper aims to summarise the main principles of mixed-mode design, its implementation in major international surveys, and the first results of a project testing the feasibility of the push-to-web mixed-mode design in the Czech context. The results illustrate that together, unequal distribution of technological skills and willingness to participate distort the recruitment of random samples for CAWI towards middle-aged and more educated respondents. The mixed-mode design – together with the described components – has the potential to better involve underrepresented groups, while it is essential to follow the rules for achieving comparable answers across modes and devices. Finally, we emphasize the need to disseminate findings to researchers and other users of survey data, as the use of mixed-mode design has been rapidly increasing