Declining response rates force survey researchers to implement surveying strategies that would motivate target persons to cooperate. Incentives have often proved to be a viable approach to address this issue. They have been increasingly used also in web surveys, which are relatively cheaper, compared to other modes but at the same time suer from generally lower response rates. Since the problems like noncoverage and lack of sampling frames still make web surveys largely inappropriate for surveying general populations, the possibility of their integration into mixed-mode designs has become one of the central questions of survey methodology. This introduced a new area also for methodological research on incentives.

While incentives can substantially increase response rates, they can also signi cantly heighten the research costs. Furthermore, the impact on survey responses often remains uncontrolled and is largely unclear in mixed-mode designs. In this paper, we study the problem of using incentives in mixed-mode surveys of the general population by comparing response rates, errors and costs among dierent experimental mixed-mode designs.

The experiment was conducted on a sample of the general Slovene population. Participants were randomly assigned into one of experimental groups, manipulating two dierent combinations of modes (web/mail and web/telephone) and three dierent types of incentives (no incentive, wallet as a gift and 5 Euros in cash). We first present the impact of incentives on response rates in each of the groups and compare errors using the Mean Square Error approach. The results show that the response rates are not a good predictor of the data quality, since the monetary incentives signi cantly boosted cooperation but at the same time produced a high bias. This finding is then elaborated by a detailed analysis of the response patterns that might have contributed to the bias. Possible adjustments of survey design, that may provide a higher data quality, are discussed as well. Finally, the cost aspects are taken into account and evaluated by a simulation of dierent initial sample sizes and their effect on the costs-errors optimization. The ndings of the study suggest that the use of incentives in mixed-mode surveys should be methodologically well elaborated, not only from the aspect of response rates, but also with a careful consideration of survey errors and costs.