STRUCTURAL DETERMINANTSOF MAIL SURVEY RESPONSE RATES OVER A 12 YEAR PERIOD, 1988-19991 Don A. Dillman, Lisa R. Carley-Baxter, Washington State University Don A. Dillman, Washington State University, PO Box 644014, Pullman, WA 99164-4014 Key Words: Mail Survey, Response Rate, Questionnaire One of the most nagging and difficult questions to answer about response rates to mail surveys concerns whether response rates have been declining in recent years, as they appear to be for telephone surveys. Another question about which little is known is the extent to which characteristics of particular survey populations influence the level of response that can be achieved when using particular design and implementation procedures. Our purpose in this paper is to provide tentative answers to these questions. Over a 12-year period, from 1988 through 1999, 102 sample surveys of visitors to U.S. National Parks were conducted by the same sponsor using very similar questions about visitor experiences in the park, and a common set of implementation procedures, except for two features" the length of each questionnaire and the number of replacement mailings. Statistically it is possible to estimate the effect of year of study on response rates for these two major variations in study procedures found in the past to be a major determinant of final response rates (Heberlein and Baumgartner, 1978; Dillman et al., 1974; Dillman, 2000). In addition, the visitors to each national park con- stitute a unique survey population, differing with regard to average age, distance traveled, time spent and other structural features that might influence response rate. This paper provides results from regression analyses in which each park becomes a unit of observation. The 102 response rates (one per park) are then regressed on these structural variables, year of study, questionnaire length, and number of replacement questionnaires in order to understand their individual and combined effects on survey response rates. THEORETICAL BACKGROUND Influence of Year of Survey Today's conventional wisdom is that response rates are declining, especially for telephone surveys. It is much less clear whether response to mail surveys is also declining. An article by de Leeuw, Mellenbergh and Hox (1996), suggests that mail response rates may be staying the same or rising slightly. It is difficult to obtain convincing estimates of possible decline, because relatively few surveys are conducted by the same procedure year after year, and at the first indication of decline surveyors often invoke stronger procedures to compensate. The questionnaires and implementation procedures used in the surveys analyzed here were judged sufficiently similar that a response analysis over time could be accomplished. Salience as an Underlying Population Variable Heberlein and Baumgartner (1978) measured salience by examining each questionnaire and making a judgment as to whether the respondents were likely to find it "interesting" to complete. We measure salience in this study by linking park attributes and whether the visitor who is asked to evaluate them is likely to have found the visit to be an interesting or memorable experience. We seek to go beyond the work by Heberlein and Baumgarmer (1978) by being able to specify population characteristics that give rise to salience, i.e., greater or lesser interest in the survey topic. Should it be discovered that measurement of salience in this way has a positive influence on response, this research would provide confirmation of the importance of this variable as an important determinant of response. Conceptually, there exist strong grounds for expecting salience of an experience to influence response rate. Webster's dictionary defines the salience as something that is "strikingly conspicuous; prominent" (Webster's II New Riverside University Dictionary 1984:1032). Social psychological theories of behavior tend to focus the role of salience in knowledge storage, activation, and accessibility. Past research suggests that salient information receives more attention and therefore increases the amount of recall, as well as makes recall easier (Higgins, 1996). It also suggests that salient information receives enough attention for the individual to make a judgement about the situation, and this knowledge is more likely to be activated in the presence of a stimulus. ISupport for this research was provided by the National Park Visitors Project at the University of Idaho under a grant to Washington State University, by the Department of Rural Sociology under Regional Research Project W-183, and by the Social and Economic Sciences Research Center. The authors wish to acknowledge the assistance of Rodney Baxter of the SESRC in formulating and carrying out this analysis, and Marc Evans and Viginia Lesser for advice on specific aspects of the regression analysis. 394