Condos in the Woods: The Growth of Seasonal and Retirement Homes in Northern Wisconsin

Appendix C: Survey Methodology

Declining response rates to surveys, particularly for surveys of the general population, are generally accepted in the social sciences (Bradburn ; de Leeuw and De Heer ; Dillman et al. ; Smith ; Steeh ).1 Indeed, a review of articles published in Society and Natural Resources from January  to January  (which reviewed Bright, Barro, and Burtz ; Carr and Halvorsen ; Connelly and Knuth ; Lauber, Anthony, and Knuth ; P. Nelson ; Vaske et al. ) finds response rates to mail surveys of the general U.S. population ranging from  percent to  percent, with a mean of  percent. Declining response rates are worrisome due to the potential impacts of nonresponse bias on the data collected, hampering attempts to draw inferences and analyze relationships . Our study illustrates that obtaining a high response rate for a self-administered survey of the general population is possible despite the downward trend. Employing an expanded version of Dillman’s “tailored design method,” we achieved an . percent response rate for a general population survey of landowners. Our study further explores concerns about nonresponse bias by examining differences between early and late respondents in their demographic characteristics and attitudes toward natural resources management. Comparisons were made between respondents from the commonly adopted  Appendix C survey methodology 1. An expanded version of this appendix was published in Society and Natural Resources (Clendenning, Field, and Jensen ). Reprinted with permission.  Appendix C four-wave mailing recommended by Dillman’s (, ) earlier work (early respondents) and those who responded after additional mailings that were employed for this study (late respondents). Despite a  percent response rate after four mailings, significant differences were detected between early and late respondents on a number of measures. Survey Methodology Previous research has shown that a number of factors influence response rates, both positively and negatively. Among the most influential positive factors are the number of contacts, visual presentation of the survey, personalization of correspondence with respondents, and financial incentives. In contrast, length of survey and surveys of the general population negatively affect response rates (see Dillman , ; Heberlein and Baumgartner ). Our study shows that survey design and implementation that pay heed to these factors can achieve high response rates, even for a long survey of the general population. Our survey, eighteen pages in length, included  response items and took approximately twenty to twentyfive minutes to complete. Despite the length of the survey we achieved a response rate of . percent. We feel that this was a result of several key elements of our survey design and methodology, including a higher than usual number of mailings sent to late respondents. Survey Design and Layout Self-administered questionnaires place particular burdens on respondents when compared with other survey interview methods, making design a critical consideration. The general model of the interview process holds that a respondent first comprehends a question, retrieves the relevant information , makes a judgment, and concludes with a response (Jenkins and Dillman ). However, self-administered questionnaires place additional burdens on respondents. First, respondents must perceive the information. Then they must comprehend both the visual elements of the layout and the written elements of the survey. In addition, respondents must comprehend introductory materials, instructions contained in cover letters, and specialized instructions that occur throughout the survey. With this in mind, the visual elements of the survey must help guide respondents through the survey, reducing respondent burden and error, while verbal elements must present information and questions in a well-organized fashion (Jenkins and Dillman ). Poorly organized and presented surveys tend to increase respondent frustration, often leading to inaccurate responses or even refusals (Dillman ; Dillman, Sinclair, and Clark ; Dillman et al. ; Heberlein and Baumgartner ). Our survey paid particular attention to six visual elements: location of and spacing between elements of the survey; the shape of questions and responses; size of fonts and images; brightness and shading in the survey; simplicity and regularity; and consistent figure-ground formatting (Dillman , ). For example, individual questions and responses were enclosed in a box with light gray shading, making each question easily distinguishable from other questions. Questions, printed in a dark, bold font, were easily distinguished from responses. White spaces, easily identifiable against the gray background, were used for respondent answers. In addition , the survey was printed in a large font for ease of reading. While this increased the number of pages in the survey, it eased respondent burden in filling out the survey. Fewer items were included on each page of the survey, reinforcing the respondent’s progress...