Increasing costs and declining response rates lead to the utilisation of mixed-mode surveys, especially push-to-web surveys. Our project develops an optimal design of CAWI+CAPI data collection for the Czech settings, which faces the thread of mode effect (especially for sensitive questions). Differences in distributions of several sensitive questions appeared in the data, although the pattern is not consistent. Thus, we recommend cautious use of mixed-mode data, including checks for mode differences, using mode as a control variable, and most importantly - a thorough preparation of each questionnaire and data collection.