| It’ll be a Labor Landslide |
| Friday, 16 November 2007 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Page 2 of 4 Labor looks set to win the 2007 Federal Election in a landslide. Here’s why, along with a predicted final seat count. Earlier in the campaign I made a series of seat predictions based on the polls as they were available then. That article is here. Since then, I’ve refined the analysis and come up with hopefully a more accurate and concise prediction model. The long and short of it is that the ALP will win the 2007 Federal Election, if the past results and current polls are anything to go by. Anything that happens in the final week is unlikely to make any difference, and Kevin Rudd will be our Prime Minister after November 24th. Opinion polls (well, Newspoll, at least) Any analysis based on past election results – as this one is – requires a rough estimate of what the final 2PP (Two-Party Preferred) vote will be. To do this, we will tabulate and graph the Newspoll results from 1993 to 2007, for the final 5 or 6 weeks in each campaign. The tabular data looks like this:
The values in the “0” row are the actual election 2PP results. All of the other values are taken from the Newspolls for those years. Graphically, let’s take a look at these numbers: ![]() From the above graph, it’s fairly plain that each year has been different. We have no way of knowing what 2007 will do. Taking only the Howard years (leaving 1993 up there for comparison), we can calculate the average gain in the Coalition 2PP vote in the final two weeks of the campaign is 0.75%. Applying this to the aggregated Newspoll published on November 16 (which was 54-46 in the ALP's favor), we can arrive at a final estimate for the national 2PP of:
Note that this is only the most likely result based on past campaigns! Method used To calculate seat predictions, we can use the 2PP votes from the past four elections to identify trends in each seat. We can measure the 2PP results for a given seat and come up with a fraction of the state 2PP result. Using Normal Distribution formulae, we can compute a mean, standard deviation, and a probability that the seat will exceed 50% 2PP for a given statewide 2PP. Then we can apply the most recent state-by-state Newspoll breakdowns (published on November 16, 2007) to provide a 2PP for each state. Here’s a refresher of that data:
Finally, we will factor in the narrowing of 0.75%, weighted for each state. Let’s apply this method to South Australia. Newspoll on 1 November said that the 2PP for SA was 53-47 in favor of the ALP. Apply the narrowing factor and we get 52.25% for the 2PP. This produces the following table:
In the above table, the “% of State” figures have been computed by averaging the 2PP values for each seat and comparing that against the average statewide 2PP value for the ’96-’04 period. The standard deviation of these values allows us to compute a range of 2PP values for a predicted statewide 2PP of 52.3% for the ALP. The “Probability” column is the probability of the ALP winning that seat if the 2PP for the state is 52.3-47.7. From this table, Labor is favored to gain Boothby, Grey, Kingston, Makin, Sturt and Wakefield, with Barker and Mayo less likely. Applying these to the entire nation, and basing the state 2PP values on the Newspoll data from November 16, with the 0.75% narrowing applied, gives the following tables of probability “pickups” for either site.
National seat predictions Here’s the full table of seats where the ALP has a greater than 75% probability of winning the seat. The table includes the seat name, the current margin, and the probability that the seat will fall.
For ALP predicted wins between 50% and 75%, we get the following:
For ALP predicted wins between 25% and 50%, we get the following:
And for ALP seats with only a very slight chance, between 1% and 25%, we get the following:
Note that Kennedy is currently an Independent seat. Finally, we can group all of the potential Coalition gains together, and those seats are:
Final seat count So how do we make sense of these numbers? Obviously there will be some surprises, as some seats will go against the trends and/or the most likely outcome. However, we can reasonably expect those seats to balance out if the final two party preferred estimate of 46.75-53.25 is accurate. To derive a seat count, we can apportion fractions for each seat, based on their probability values. For example, a 100% chance for the ALP (e.g. Hasluck) counts as one (1.0) seat gained, and a 47% chance for the ALP (e.g. Forde) counts as 0.47 seats gained. Similarly, a 38% chance for the Coalition (e.g. Richmond) counts as 0.38 seats gained. Summing all of these fractions (omitting Kennedy from the ALP count) gives the following final seat change predictions:
Thus, there will be a net gain of approximately 33 seats to the ALP for a national 2PP vote of 53.25%. Of interest, the +33.97 gain for the ALP includes +3.11 in National Party seats, and +0.99 for the Country Liberal Party in Solomon. Assuming that the two Independents hold their seats (86% probability in Kennedy, 100% probability in New England), and adding 33 seats to the ALP’s total of 60 currently held, the final prediction for the 2007 Federal Election is:
For Labor, this is two seats less than the number of seats that the Coalition held after the 1996 federal election and will mean that the ALP will have a 36 seat majority. - “Scotty” McCamish
Views: 4086
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Last Updated ( Saturday, 17 November 2007 ) | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Next > |
|---|


Comments (19)



