Web Survey Bibliography
With telephone RDD coverage and response rates declining, survey researchers are looking for new ways to survey the general public effectively. Postal Delivery Sequence Files (DSF) of all residential addresses that receive regular postal delivery now make general public surveys by mail possible. The consensus among most surveyors is that the Web is an inadequate mode by itself for surveys of this type. In this paper, we report results of an experiment conducted using the DSF address list in a small metropolitan region of Western United States to evaluate the potential for surveying the general public by web, mail, or a combination of web and mail. Random subsamples of 2,800 addresses were assigned to one of five treatments: mail only, mail preference, web preference, and equal preference (½ with a standard instruction and ½ with an elaborated instruction for accessing the web survey). A significant effort was made to develop similar visual appearance of the mail and web questionnaires and every treatment contained the same 52 questions. Preliminary results for this experiment, which was fielded in the Fall of 2007, show that response rates vary significantly by treatment, but that the combined overall response rates are similar to those traditionally obtained by Dillman s Tailored Design Method, despite the address-only characteristic of the DSF file where individual personalization was not possible. The high variation across treatments in the percent of households responding via the web suggests that a significant proportion of households will respond by that means instead of mail. Both the response rates and success in encouraging response via the web suggest the potential for a new general public survey methodology that combines mail and web and provides detailed instructions for how to access the web survey.
Homepage (abstract)