Using Internet in Stated Preference Surveys: A Review and Comparison of Survey Modes

Abstract

Internet is quickly becoming the survey mode of choice for stated preference (SP) surveys in environmental economics. However, this choice is being made with relatively little consideration of its potential influence on survey results. This paper reviews the theory and emerging evidence of mode effects in the survey methodology and SP literatures, summarizes the findings, and points out implications for Internet SP practice and research. The SP studies that compare Internet with other modes do generally not find substantial difference. The majority of welfare estimates are equal; or somewhat lower for the Internet surveys. Further, there is no clear evidence of substantially lower quality or validity of Internet responses. However, the degree of experimental control is often low in comparative studies across survey modes, and they often confound measurement and sample composition effects. Internet offers a huge potential for experimentation and innovation in SP research, but when used to derive reliable welfare estimates for policy assessment, issues like representation and nonresponse bias for different Internet panels should receive more attention.

Public beliefs about high-voltage powerlines in Norway, Sweden and the United Kingdom: A comparative survey

“…Regarding content, the questionnaire survey was developed by the researchers in the project and comprised questions aiming to grasp key beliefs and responses to high voltage 1 A review of the use of internet surveys including consumer panel populations and respondents in social and economic research points to both strengths and weaknesses with panels, but conclude that such panels become more and more sound as a basis also for research purposes [39]. Table 1.…”

Investigating Internet and Mail Implementation of Stated-Preference Surveys While Controlling for Differences in Sample Frames

“…Yeager et al (2011) reported an effective response rate of 15 % when non-response is included from initial recruitment into the internet panel through to the final survey response. Lindhjem and Navrud (2011b) suggested weighting internet-survey data to account for error introduced by survey non-response. However, Yeager et al (2011) noted that while post-stratification with demographics increases the accuracy, the results are still not comparable to initial random sampling.…”

Public perceptions of planning objectives for regional level management of wild reindeer in Norway

“…Two-thirds of the total number of contacts made (67.8%) did not wish to participate in the survey for different reasons (reported, did not have time, not interested, topic of little relevance, no particular reason). Low response rates in surveys have become a trait of social science research over the past decade or so concurrent with the tremendous increase in internet-based surveys which offer a huge potential for experimentation (Lindhjem and Navrud 2011). Online surveys have also become highly attractive to wildlife research as a cost-effective way to measure constituents' positions on a variety of issues.…”