Conventional wisdom suggests that public opinion researchers must accept as fact that response rates to all forms of surveying are in decline. In the case of web surveys the problem is less one of decline than the fact that they have never been high, except for certain specialized populations, and there is little evidence that they are likely to become higher. This sense of reality has led to a shift in concern by many internet surveyors towards simply measuring whether non-response error exists and away from putting emphasis on getting higher response rates. In this paper we propose that the appropriate application of existing theories can provide guidance for improving web survey response rates significantly beyond current expectations. Doing so may help researchers in the reduction of sampling error as well as improving the credibility of surveys and the methods used. It may also help in the reduction of nonresponse error.

Several theoretical explanations of why people do and do not respond to various survey modes exist, including cognitive dissonance, social exchange, reciprocal obligation and leverage-salience. In this paper we draw from existing theories and research to propose means by which web response rates can be improved significantly over most current expectations. We do this by using theory-based arguments to propose methods for collecting survey information over the Internet, and provide evidence to support those arguments from experiments by the authors as well as others. We conclude with specific recommendations for improving response to web surveys and for advancing the testing of the theoretically-based propositions presented here.