While the polls conflict, Australian PM Howard seems to have taken the lead over Labor's Mark Latham. Australians will decide their next leader in 9 October elections; just two weeks from now. Latham has promised to withdraw Australian forces from Iraq. Howard has made Australia a very stong supporter of America's War on Terror. http://www.smh.com.au/articles/2004/09/24/1095961858763.html# The "Sydney Morning Herald" may require registration. Landslide, Cliffhanger or Oblivion - the Polls Can't Decide By Mark Coulton September 25, 2004 The latest Herald Poll showing a dramatic swing to the Government is out of alignment with the other published polls. (but it is apparently the latest poll) Only this poll, conducted by ACNielsen, shows the Coalition being returned in a landslide. If its 54-46 two-party-preferred result was replicated on October 9, the Government would win a bigger majority than it obtained in 1996 and Labor would have its worst result since 1977. Remarkably, it shows the Coalition primary vote at 50 per cent. This compares to a 43 per cent primary vote recorded by Newspoll, and 42.5 per cent by Roy Morgan. In other words, on average, the Government would have won an election held early this week without the need for preferences from any other party. The Australian's latest Newspoll, taken last weekend, showed Labor leading 52.5 to 47.5 per cent after preferences. The Morgan poll, also taken last weekend, showed Labor ahead 53 to 47 per cent on a two-party-preferred basis. The Galaxy poll, taken for News Limited's metropolitan newspapers, shows the Government narrowly ahead, 51 to 49 per cent. To make matters even more confusing, there is also little agreement about the voting trends. The last Galaxy and Newspoll surveys show the Government losing ground, while the Morgan and ACNielsen show them gaining ground. The margin of statistical error in the ACNielsen poll is plus or minus about 2.6 per cent, which means that the Government's support, after preferences, could be as high as 56.6 per cent or as low as 51.4 per cent. Either way, it spells a decisive win to the Coalition. All polls are subject to a margin of error, which decreases as the sample size increases. But these margins of error only express a range of possible results within a 95 per cent probability. In other words, one in 20 results will be outside these margins. If two polls get the same result, then there is only a one in 400 chance they are both outside their margins of error. For pollsters, there is safety in numbers.