In this study we examine two different mixed mode approaches, each offering the respondents to choose between a paper questionnaire and a Web questionnaire. These two approaches are compared with a traditional mail approach, as well as an e-mail approach, with respect to costs and response rates.

A survey requesting students at Linköping University to provide data regarding their housing conditions was conducted during spring 2004, in collaboration with the statistical office of the local government of Linköping. The principal purpose of the study was to examine if mail invitation to a Web questionnaire, thus making probability sampling possible, differ in response rate and costs from a traditional mail survey. In an experiment embedded in the survey, a random sample from the student population was randomly divided into four groups distinguished by the type of invitation and questionnaire delivery: (1) a traditional mail invitation and paper questionnaire with one mail reminder including a paper questionnaire, (2) a mail invitation to a Web questionnaire including a paper questionnaire (thus offering the respondent a choice between Web and mail) with one mail reminder not including a paper questionnaire, (3) a mail invitation to a Web questionnaire with one mail reminder including a paper questionnaire (thus offering a choice in the reminder), and (4) for reference an e-mail invitation to the Web questionnaire and two e-mail reminders (all students at Linköping University have Internet access and individual e-mail addresses). Cost estimates and response rates per survey group will be presented in the paper. The findings will be compared and contrasted to similar studies reported in the survey literature.