In what ways, if any, do campaigns matter? Specifically, what can we say about election campaigns in Australia? If they shift votes, in what direction do voters move? A comparison of opinion poll data taken at the beginning of each of the last ten national elections with the distribution of the vote at the end of each campaign suggests that the gap in first preferences between government and opposition generally narrowed; certainly, it never widened. The data do not suggest that incumbency (being in government) is an advantage. Nor do the figures conjure visions of voters clambering on bandwagons—as we might expect if there were a 'spiral of silence'. Instead, the data point to an underdog effect, with the party that starts behind making up ground on the party that starts ahead.