Web Survey Bibliography
Online polls have become more widely used in election campaigns, but their reliability in accurately predicting the outcome of an election remains variable. This article reports on the results of an ambitious online polling experiment conducted during the 2004 Australian federal election. In contrast to a more modest online poll conducted at the previous election in 2001, the 2004 poll turned out to be the least accurate of all of the polls conducted during the campaign. A series of hypotheses are tested to explain the inaccuracy of the poll. The main conclusion is that the sampling methodology used required the steady recruitment of new respondents, but that became much reduced as the campaign progressed due to lack of publicity from the online poll’s main media partner. The 2004 Australian experience has important lessons for other online polls which use a similar methodology.
Homepage (Abstract)