Small changes in the distribution of votes in each country would have seen very different outcomes. bq. In the United States, the President won 51% of the popular vote but the outcome was determined by a majority of Electoral College that was due, finally, to the results in Ohio. There, the Bushmargin was 130,000 votes; in other words, had 65,000 of the voters who made up their mind on the day of the poll voted for Kerry, he would now be President. What does this tell us about the sentiments of the U.S. electorate? Nothing. bq. In Australia, the gap between the primary votes of the Liberal-National Coalition and the Labor party was much larger than in the Bush/Kerry race. But the “outcome”:http://www.abc.net.au/elections/federal/2004/results/ was still less an endorsement of the Howard government than the U.S. result had been for Bush. The vote for the government was smaller than the vote against the government by a margin of 3.2 percent of the electorate: the conservative government secured 46.8 percent of the popular vote, Labor 37.7 percent and the rest went to Greens, the Democrats and independents. Due to the distribution of votes, however, this minority of the popular vote won the incumbents 87 lower-house seats while the majority non-government vote won only 63 lower-house seats. In neither case can these close and ambiguous results be extrapolated to any political or social conclusion. As the mathematician John Allen Paulos “argues”:http://education.guardian.co.uk/print/0,3858,5065109–110865,00.html, the idea that significant events such as the return of Bush to the White House must have significant causes is only a “charming supersitition.” Far from suggesting a profound swing in electoral moods or values, statistically defensible models suggest a high degree of latency in the electorate’s preferences that may also hold in Australia. Paulos points to Yale economist Ray Fair’s six explanatory factors: bq. The first is incumbency, which has been a distinct advantage historically. The second is party (Republicans have a slight historical edge), and the third is “party fatigue” (two or more terms out of power offers some benefit). The remaining three factors concern the economy: GDP’s per capita growth rate (higher is better for the incumbent), the number of quarters during the preceding four years in which the growth rate exceeded 3.2% (the more, the better), and the inflation rate (lower is better). On the basis of these six factors, Fair’s model has generated quite accurate vote percentages in presidential elections dating back to 1916.