The latest Nielsen poll, results published in the Fairfax papers on July 31, gives the election to the Coalition by 52% to 48%. It's at the outer edge of where the polling results have been and could be just a statistical fluke. Some commentators say the result represents too large a move and that moves like this don't happen in the real world. So why do these commentators bother with polls if they know what happens in the real world better than the polls?

This is the election where Galaxy may prove or cruel their reputation. They have consistently put Labor lower than the other polling organisations, although still within the polling margin of error.

Morgan shows a closing of the gap between Labor and Coalition, but still gives a healthy margin to the government of 6%. The change is one percent, and therefore not statistically significant. What is really statistically significant is the gender gap.

Newspoll this morning is reporting that the Labor/Coalition split is back to 52% to 48%. This is just a statistically significant result. Their measure of the relative popularity of the leaders is certainly statistically significant with a sharp decline in Gillard's approvals and an increase in her disapprovals.