Examining the Two Party Preferred (2PP) results for the House of Representatives from the previous four Australian Federal elections can indicate that most seats have a similar fraction of the overall state 2PP votes over time.

Using election data from the 1996, 1998, 2001 and 2004 federal elections enables comparisons between seat 2PP results and the overall results in that particular state. The percentage results for each seat have been averaged and a standard deviation has been computed.

Applying this test to a given 2PP result (say, from the September 2007 quarterly Newspoll, which gave 2PP scores for each state) then gives a range of possible 2PP results at the November 24, 2007 federal election. This range is calculated to be the computed mean fraction of the statewide 2PP, plus or minus the computed standard deviation of the fraction of the statewide 2PP.

Examples

To illustrate this approach, let’s look at some seats in New South Wales, and let’s take the ALP 2PP numbers as our benchmark: 

	1996	%	1998	%	2001	%	2004	%	Mean	S.D.
NSW ALP 2PP	47.3		51.9		49.5		47.7
Bennelong	39.9	84.3	44.0	84.8	42.3	85.4	45.7	95.9	87.6	4.8
Eden-Monaro	45.2	95.5	49.8	96.0	48.3	97.5	47.9	100.5	97.4	1.9
Macarthur	39.3	83.1	44.4	85.6	43.3	87.4	40.5	85.0	85.3	1.6
North Sydney	34.4	72.7	37.8	72.9	36.8	74.3	40.0	83.9	76.0	4.6

 

 

In the examples above, whatever the NSW ALP 2PP statewide result is, Bennelong will be between 82.8 and 92.4% of that, while Eden-Monaro will be between 95.5 and 99.3%. Macarthur will be between 83.7 and 86.9% of the statewide ALP 2PP, and North Sydney will be between 71.4 and 80.6% . From these numbers alone, it is easy to see why Eden-Monaro is considered the “bellwether” of the federal election, given that it closely follows the national 2PP and doesn’t vary much.

 

Applying the method
Using the above method, we can construct a series of percentage ranges for the each seat versus the ALP 2PP in the states. Putting the statewide 2PP results in from the most recent Newspoll then gives a range of 2PP results for each seat. From this range, if the high and low bounds are both greater than 50%, we can say that the seat will be an ALP gain (or retain). If the low bound is below 50%, and the high bound is above 50%, but by a further margin (say 48.2 to 56.7%, for instance), we can say that the seat is a Likely ALP gain (or retain). If the low bound is below 50% by a wider margin than the upper bound (say, 44.5 to 51.2) we can say that the seat is a Likely Coalition gain (or retain) and finally, if both low and high bounds are below 50%, we can say that the seat is a Coalition gain (or retain).

Applying these standards to the September Newspoll 2PP for the ALP in New South Wales gives the following results:
 

	Low bound	High bound	Predicted result
NSW ALP 2PP: 58%
Bennelong	46.0	55.6	Likely ALP gain
Eden-Monaro	54.4	58.4	ALP gain
Macarthur	47.9	51.0	Likely Coalition retain
North Sydney	39.4	48.7	Coalition retain

 NOTES:

1. Redistributions in seats have NOT been taken into account at this time; however refer to the next note (#2).
2. The seats of Bonner and Wakefield have been included based on their 2004 result only. Bonner was a new seat in 2004. Wakefield had significant redistribution in 2004 due to the demise of Bonython, so much so that the 1996-2001 data for this seat had to be discounted. Other seats suffering less extravagant redistribution, such as Bennelong, Macquarie and Parramatta have been left as is.
3. The new seat of Flynn is not considered since there is no previous election data for this seat.

 

Sources:

1. The Australian Electoral Commission. http://www.aec.gov.au
2. Newspoll, via the Australian Newspaper. http://www.theaustralian.news.com.au/files/cumulative-21-sep.pdf