Remove this Banner Ad

LIVE Federal Election Coverage 2016

  • Thread starter Thread starter Jiska
  • Start date Start date
  • Tagged users Tagged users None

🥰 Love BigFooty? Join now for free.

Preference flows have diverged a lot from traditional norms over the last few years, I feel like his projections still give too much weight to historical data. In the most recent NSW and Qld elections it sort of bit him in the arse, as a result he's become a lot more cautious with his predictions.

I'm no professional psephologist, but I've been a bit disappointed by how slow he's been to adapt to the changing political landscape. You could say that elections are simply harder to predict these days, I suppose. I kind of feel like that's a cop-out though.
Interesting thoughts.
 
At the moment (roughly 8PM eastern), the ABC have declared the following:

LNP - 70 seats
ALP - 67 seats
Undetermined - 8 seats
Total - 150 seats
Required for majority - 76 seats


Similarly, ABC has predicted that of the above 8 seats that are too close to call, the following ditribution will likely occur:

LNP - 4 seats
ALP - 5 seats

Which in total, after the assumption that the ABC's prediction is correct, we've got:

LNP - 74 seats
ALP - 72 seats

The ABC have turned their predictions off even though the website still uses the word predict.
 
I have a lot of respect for Antony Green and everything he's done over the years, but his preference forecasting at the last couple of elections has been decidedly dodgy.

I think he needs to revisit some of the underlying assumptions behind that aspect of his methodology.
I agree. His algorithms, especially this year have been way off as they were waaay too trigger happy to call individual seats for either party, especially the libs.
He needs to revisit calculating in preference flows from the previous election given how volatile Australian politics is of late and just wait til he has actual preference flows from the AEC

Edit: Just read your follow up post. Glad to see we agree
 
The ABC have turned their predictions off even though the website still uses the word predict.

they haven't, the seats they've called do not reflect the raw count, hence the discrepancy.
for example they have given the liberals 70 seats so far, yet the current raw count only has the liberals leading in 67 seats.

the ABC is mostly likely accurate in their prediction, but its still a prediction.
 

Log in to remove this Banner Ad

I agree. His algorithms, especially this year have been way off as they were waaay too trigger happy to call individual seats for either party, especially the libs.
He needs to revisit calculating in preference flows from the previous election given how volatile Australian politics is of late and just wait til he has actual preference flows from the AEC
You should revisit your avatar choice
 
Reckon Libs get to 76 here.
Hopefully as accurate as your many predictions of a comfortable LNP win. :)

Sent from my HTC_PN071 using Tapatalk
 
By my count, there were three seats called at some point in the night, which have ended up going the other way. Far too generous.

Similarly, ABC has predicted that of the above 8 seats that are too close to call, the following ditribution will likely occur:

LNP - 3 seats
ALP - 5 seats

Which in total, after the assumption that the ABC's prediction is correct, we've got:

LNP - 73 seats
ALP - 72 seats

Typo highlighted
 
To be fair to Green, it's not like he doesn't know this. It's not exactly rare for the computer to give away a seat on the night and for him to respond by saying something like 'I'd wait a bit longer on that one, even though the computer is giving it away'. Remember, he is just as much providing entertainment as he is providing a service, at least on the night of the election. He can't afford to have it all taking hours of the forecast before he's willing to give any results.
 
they haven't, the seats they've called do not reflect the raw count, hence the discrepancy.
for example they have given the liberals 70 seats so far, yet the current raw count only has the liberals leading in 67 seats.

the ABC is mostly likely accurate in their prediction, but its still a prediction.

The ABC predictions I'm looking are an exact match for the figures on the AEC tally room website, so the ones I'm looking at are not predictions at all.
 
I have a lot of respect for Antony Green and everything he's done over the years, but his preference forecasting at the last couple of elections has been decidedly dodgy.

I think he needs to revisit some of the underlying assumptions behind that aspect of his methodology.
There is a God I tell ya
 
I still believe in Antony
Yeah, between him and Malcolm Mackerras you pretty much have the invention of scientific election analysis in Australia. All the systems he uses, he built from the ground up.

I will definitely miss him when he's gone.
 
By my count, there were three seats called at some point in the night, which have ended up going the other way. Far too generous.



Typo highlighted
Yep, I completely overlooked that! That's what post-night shift with no sleep does to me, haha - thanks for picking it up!

That makes it even more interesting!
 

Remove this Banner Ad

care to share a link, because this is the link i have for the ABC http://www.abc.net.au/news/federal-election-2016/results/
it does not match the AEC raw count, it is most certainly a prediction.
If you're really keen and you know your way around XML you can get the raw data from ftp://mediafeed.aec.gov.au/20499/
Differences in standard/detailed and verbosity are explained on page 11 here

fyi, if you use the detailed data you will also need the preload for the polling districts and candidates and have the know how to bring it all together

WsyqoBK.gif
 
Last edited:
If you're really keen and you know your way around XML you can get the raw data from here
Differences in standard/detailed and verbosity are explained on page 11 here

fyi, if you use the detailed data you will also need the preload for the polling districts and candidates and have the know how to bring it all together

WsyqoBK.gif

that's too nerdy even for me...
 

🥰 Love BigFooty? Join now for free.

If you're really keen and you know your way around XML you can get the raw data from here
Differences in standard/detailed and verbosity are explained on page 11 here

fyi, if you use the detailed data you will also need the preload for the polling districts and candidates and have the know how to bring it all together

WsyqoBK.gif

If you're gonna be a nerd at least get the link right: ftp://mediafeed.aec.gov.au/20499/
 
Due to the DD, half the senate needs to serve 3 year terms, while the other half serves 6, but there's no official or legal requirement outlined in the constitution for how this occurs. This means half + 1 could potentially bully the other half into serving a half term.

Given the likely senate numbers, it's highly possible that Labor, Greens, and Xenophon (plus Hinch, depending on total Greens numbers) will form an alliance and force the remaining senators (Libs, Nats, Lambie, Nile, Hanson) to serve 3 year terms.

lol
 
I tend to feel that it's not real trust unless there's something at stake.

Antony Green is quite professionally respected, but it's not like anybody's going to die if he gets a prediction wrong.
 
Due to the DD, half the senate needs to serve 3 year terms, while the other half serves 6, but there's no official or legal requirement outlined in the constitution for how this occurs. This means half + 1 could potentially bully the other half into serving a half term.

Given the likely senate numbers, it's highly possible that Labor, Greens, and Xenophon (plus Hinch, depending on total Greens numbers) will form an alliance and force the remaining senators (Libs, Nats, Lambie, Nile, Hanson) to serve 3 year terms.

lol

There most certainly is a method for who gets 3 and 6 year terms, and I'd be pretty surprised if it wasn't set out in legislation.
 

Remove this Banner Ad

Remove this Banner Ad

🥰 Love BigFooty? Join now for free.

Back
Top Bottom