Friday, 2 August 2013

Summaries of summaries of summaries.

Last week I finally concluded my electoral seat run down. Shortly after that my long awaited and lengthy NSW run down imploded. A reconstruction is currently underway. From scratch. Again.

This is something of a recurring problem, which is why it is convenient that I intend to effectively replicate all of my previous data this week. This is something of a summary to put us within striking distance of an informed, though inevitably flawed prediction at the drop of a hat.

Specifically this hat, dropped from around 43.5 megameters, whilst assuming constant gravity of 1G and air resistance comparable to a NASA wind tunnel.*

Variable-Dependent Transparency Array Results:

What, you thought the map with a silly name was gone for good?

Because the VDTA is somewhat subjective in as much as how one determines the average opacity, I have been a little simplistic with my analysis. I simply had the computer select all of the boxes from the corresponding data-dump which were 50-50 red and blue, within the tolerances of the Gimp select by colour tool.

These seats were:

Eden-Monaro, Hughes and Lindsey from NSW; Solomon from the NT; Bonner, Bowman, Dickson, Flynn, Forde, Leichhardt, Petrie and Wright from Queensland; Grey and Makin from SA; Bass and Franklin from Tasmania; Bruce, Chisholm and McMillan from Victoria; and Canning, Durack, Hasluck and Stirling from WA.

Herbert (QLD) and Robertson (NSW) also lost a noticeable number of pixels, so lets include them as well.

Coloured as in original, trend-based VDTA.

Margins Analysis:

When looking at margins I replicated the traditional post-election pendulum. It is not creative or unique, but it is a steadfast and useful tool. As I demonstrated, however, it is not perfect. The size of the margin is less critical, for example, in seats with traditionally less swing and potentially more critical in seats with a large 'others' vote. I may or may not get the time to develop tools to remedy these limitations before the election.

The pendulum is reproduced here:

Seat Run-Downs:

As discussed last week, the "Category 1 - Critical" seats are as follows:

Adelaide (SA), Bendigo (VIC), Blair (QLD), Braddon (TAS), Bruce (VIC), Canning (WA), Capricornia (QLD), Chisholm (VIC), Dawson (QLD), Denison (TAS), Dickson (QLD), Dunkley (VIC), Forde (QLD), Flinders (VIC), Gellibrand (VIC), Greenway (NSW), Herbert (QLD), Leichhardt (QLD), Lindsay (NSW), Longman (QLD), Macquarie (NSW), Makin (SA), McEwen (VIC), McMillan (VIC), O’Connor (WA), Page (NSW), Parramatta (NSW), Paterson (NSW), Petrie (QLD), Solomon (NT), Swan (WA), Wakefield (SA).

There are also several tossups:

Bass (TAS), Bonner (QLD), Brisbane (QLD), Cowan (WA), Dobell (NSW), Durack (WA), Flynn (QLD), Hasluck (WA), Kingston (SA), La Trobe (VIC), Lilley (QLD), McMahon (NSW), Moreton (QLD), Richmond (NSW), Robertson (NSW), Wright (QLD).


Other factors have been discussed throughout the year -- for example the role of the budget -- but I will not be focusing on these this week, because the above factors will form something of a baseline for the rest of our information to build from. There are also, of course, a great many factors and statistics not discussed here this year.

Seats to Consider:

The map below contains all of the above information. For that reason there is no need for a Data Dump this week.

These seats will receive further analysis next week in an automated post, while I am out of the office. It is important to note that not all of these seats will be key players in this election; many safe seats will have been highlighted under one of these approaches and will be dismissed next week. The really interesting seats will be those that are marked by all three measures, like Flynn in Queensland and Robertson in New South Wales.

As I said earlier, I will not be around to respond to any comments, so talk amongst yourselves for a while. Knowing my luck the election will be announced this week while I am unable to blog.

*By these calculations (based on the dimensions and weights provided here, the drag coefficients offered here, a little approximation of hat radius determined from its height, and the assumption that air resistance would flip the hat top-down and maintain this orientation) this hat should quickly reach terminal velocity and take 33 days - the minimum time between an election announcement and an election - to fall to earth.

Randall Munroe, eat your heart out!

No comments:

Post a Comment