Online election surveys: Keeping the voters honest?

This study investigates the question of Internet mode effects in online election studies. Specifically, we examine whether Web versions of election studies can produce more accurate or truthful estimates of vote choice and party preference than their more conventional offline counterparts. Existing studies have indicated that a Web environment may lower the social context of the survey, promoting greater openness from respondents in answers on political preference items. We examine this question using data from the 2001 Australian Election Study (AES) in which a Web and mail survey were conducted. Crucially, both online and offline questionnaires relied on self-completion, a standardization lacking in previous studies and that allowed for a more controlled test of mode effects. The results reveal no significant differences in the expression of political preferences across the two surveys after controlling for key demographic and attitudinal factors. Significant differences do emerge, however, in vote choice depending on whether an individual had Internet access. We conclude that while Web mode per se does not have any notable effect on respondents' answers to political choice questions, until the issue of universal access is resolved, its substitution for existing methods would be undesirable as this would exclude an important and politically distinctive subset of the population.