A long-standing notion within survey research is that most respondents have a preference for a particular survey mode. Shortly after the development of telephone data collection methods, Groves and Kahn (1979) found that respondents tend to prefer one data collection mode over another. Because respondents may prefer one mode to another, many survey researchers have assumed that response rates, and general goodwill and attitudes about the questionnaire, can be improved by catering to the preferences of potential respondents, oftentimes by providing a choice between different modes (see Shih and Fan 2007; Dillman, West, and Clark 1994; Diment and Garrett-Jones 2007; de Leeuw, Hox, and Dillman 2008).  In recent years, the Internet has become an increasingly more enticing medium for survey research. Switching mail or telephone surveys to the web has many benefits; web questionnaires are significantly less costly for researchers and could potentially reduce the burden put on respondents. More surveys are now conducted using the web, with varying results. However, web surveys of the general public have not yet effectively achieved response rates that are equivalent to those of mail questionnaires or other forms of surveying (Manfreda et al. 2008; Shih and Fan 2007) and coverage is limited. When given a choice, it appears that most respondents still prefer other modes of response to the Internet (Shih and Fan 2007; Diment and Garrett-Jones 2007). Thus, in order to achieve the best response rates, some prior research suggests surveyors should continue to cater to people’s preference for mail questionnaires.  The analyses presented in this report address the issue of mode preference and its implications for survey response rates. This research utilized a mixed-mode web and mail survey experiment. One purpose of this study was to develop methods for improving response rates for web surveys. The experiment results imply that it is possible to achieve relatively high web response rates using several techniques. In this report we discuss these methods and the results of  the experimental treatments included in this study. Additionally, our data reveal some surprising effects of survey mode on respondents’ mode preference, which suggest that relying on people’s stated preferences may not be a necessary or important method for deciding which survey mode to utilize. In light of these findings, we believe the web is a viable option for conducting certain surveys of the general public. To further understand the usefulness of web surveys, we also examine factors that may influence the likelihood of respondents choosing a web questionnaire over another option. We conclude by summarizing what this research tells us about effectively implementing web surveys.