Do response rates and responses differ between mail and internet answers in a mixed mode survey? : A comparison of qualitative business survey data collected by regular mail and internet questionnaires

For many years, Business Tendency Surveys on the basis of mailing suffer from an erosion of the response rate. To counter this problem, there are traditional methods as limitation of the number of questions, improvement of the design of the questionnaire, intensified recalls by mail or telephone, more attractive outputs and a time-reduction of the delivery of the results. Another way to stop the reduction of the response rate is to offer different modes of surveys. Therefore, the KOF introduced an online survey in addition to the mail survey. Up to now, there are only few research projects which analysed the effects of the new technique of the online survey on the response rate and the quality of the responses. First studies indicate that mail surveys produce higher response rates than internet surveys. Our results show a big difference in the unit non-response rate too. Differentiating according to language regions, branches, size of firms and by the percentage of exports on sales does not change the picture. If there is a different behaviour of firms between the two modes this could be reflected not only on the response rate but also on the selection of the response category. This would have a significant draw-back for the survey results and their interpretation. The results show at least in the case of "Total manufacturing industry" that the differences between the two modes can be neglected. The results of business tendency surveys are rarely used on a micro level. The internationally common aggregation technique is the balance method. We therefore analysed the effect of the differences of the two survey modes on the balance indicators. The examination had to be restricted to the monthly survey in manufacturing industry. Thirteen out of fifteen questions have equal means in the two survey modes. -- survey mode ; unit-non-response ; item-non-response ; ordinal measures of association ; business tendency survey