babble home
rabble.ca - news for the rest of us
today's active topics


  
FAQ | Forum Home
  next oldest topic   next newest topic
» babble   » archived babble   » election 2006   » predicted vs. actual results

Email this thread to someone!    
Author Topic: predicted vs. actual results
election model 2004
rabble-rouser
Babbler # 6305

posted 01 February 2006 12:58 AM      Profile for election model 2004   Author's Homepage     Send New Private Message      Edit/Delete Post
I forgot to post this here, but if anyone is interested in seeing the DemocraticSPACE actual vs. predicted results, riding-by-riding, you can download the file at:

http://democraticSPACE.com/canada/2005election/analysis-national.pdf
(800 kb PDF)

You can also see a final summary at:
http://democraticSPACE.com/blog/2006/01/the-final-word-predicted-vs-actual-summary/

Cheers,
Greg


From: CA and ON | Registered: Jun 2004  |  IP: Logged
candle
rabble-rouser
Babbler # 3103

posted 01 February 2006 01:51 AM      Profile for candle     Send New Private Message      Edit/Delete Post
Greg - you should send the pdf's to Ipsos-Reid and show them how a real projection model works. Pretty impressive on the % breakdowns for each candidate on the seats - although there were some unexpectedly good results for NDP'ers that threw you off a bit on some

(e.g.) Peter Stoeffer and JUdy W-L getting over 55% of the vote each. Charlie Angus getting 51% of the vote. Alexis MacDonald getting 33% vs 23% predicted (Alexis did as well as Peter Mancini and almost as well as Marilyn Churley - she cut MacKay's plurality down to about 3300 votes or 8%). In all these ridings you had the right results just a little low on the %.


From: Ontario | Registered: Sep 2002  |  IP: Logged
Privateer
rabble-rouser
Babbler # 3446

posted 01 February 2006 02:57 AM      Profile for Privateer     Send New Private Message      Edit/Delete Post
No surprise that the NDP vote was quite underestimated in many seats in NS and NB. I could tell we were gaining strength in unusual places but that kind of thing is hard to guess in real numbers on the ground.

BTW the percents are wrong in the "actual" results for Halifax West. It says 24% CPC and 23% NDP when the results were 24.4% NDP and 23.0% CPC. Small difference but an important one to me.


From: Haligonia | Registered: Dec 2002  |  IP: Logged
Albireo
rabble-rouser
Babbler # 3052

posted 01 February 2006 02:55 PM      Profile for Albireo     Send New Private Message      Edit/Delete Post
Greg, an excellent effort. Your projections were more accurate than any I've seen.

Of course, you have the advantage of incorporating all polls into your model. If you were working for one pollster, like Ipsos, you'd likely be limited to using their polls, and thus would have a higher margin of error. But it would still be better than what they do.

Hopefully you'll extend this effort to other elections.


From: --> . <-- | Registered: Sep 2002  |  IP: Logged
adma
rabble-rouser
Babbler # 11856

posted 02 February 2006 10:14 PM      Profile for adma     Send New Private Message      Edit/Delete Post
If only I had the time to finish, I'd do my own comprehensive number-crunch spin (going off my electionprediction.com record).

Just to pick on one thing, I always felt democraticspace overweighted the Tories in Brampton West; after all, a lot of the 2004 vote was a Tony Clement vote rather than a Tory vote per se...


From: toronto | Registered: Jan 2006  |  IP: Logged
adma
rabble-rouser
Babbler # 11856

posted 02 February 2006 10:20 PM      Profile for adma     Send New Private Message      Edit/Delete Post
Oh, and speaking of electionprediction, 278 out of 308, 90.3% accuracy...
From: toronto | Registered: Jan 2006  |  IP: Logged
Stockholm
rabble-rouser
Babbler # 3138

posted 02 February 2006 10:38 PM      Profile for Stockholm     Send New Private Message      Edit/Delete Post
and how many of those 278 correct calls were safe seats for one party or another than any idiot could have predicted accurately?
From: Toronto | Registered: Sep 2002  |  IP: Logged
peterjcassidy
rabble-rouser
Babbler # 372

posted 02 February 2006 10:40 PM      Profile for peterjcassidy     Send New Private Message      Edit/Delete Post
Much appreciated. Can you give us some sense of the minor shifts that made a big difference?

For example, from some limited evidence, I believe there was a slight shift from the NDP to the Libs in Ontario,particularly in the Hamilton region, maybe a percentage point or two, in the last few days of the election. If that shift had not occured the NDP maybe would have picked up 2 or 3 more seats in Ontario from the Libs, giving us the balance of power. (This is also one reason HESC was closer than it should have been, 500 or so of our supporters probably went to the Libs to "stop Stephen Harper".)

Then what about Quebec? Would a percentage or two staying with the Bloc or Liberals in the Quebec city region have made a big difference for the Cons?

Was the variance in particular ridings where the atual result differed from the predicted generally within the margin or error- e.g. the predicion was party X wins by less than 5%, they lose by less than 5%?


And what your model provides is not just the provincal breakdown, but the regional. So if the model was out one or two seats in a region, does that mean that the region went though a shift in the last few days or that the winning party was stronger locally than projected? Any thoughts would be appreciated.

Thanks again. And what is your next project?


From: Screaming in language no-one understands.. | Registered: Apr 2001  |  IP: Logged
adma
rabble-rouser
Babbler # 11856

posted 03 February 2006 09:29 AM      Profile for adma     Send New Private Message      Edit/Delete Post
quote:
and how many of those 278 correct calls were safe seats for one party or another than any idiot could have predicted accurately?

I know--though checking what EP did get right versus a lot of pre-election polling, I'm still surprised things turned out as well as they did.

Of course, for all I know, Milton Chan might have had inside numbers on the marginals up his sleeve...


From: toronto | Registered: Jan 2006  |  IP: Logged
josh
rabble-rouser
Babbler # 2938

posted 03 February 2006 09:55 AM      Profile for josh     Send New Private Message      Edit/Delete Post
quote:
Originally posted by Stockholm:
and how many of those 278 correct calls were safe seats for one party or another than any idiot could have predicted accurately?

Yeah, but they predicted better than all the other idiots.


From: the twilight zone between the U.S. and Canada | Registered: Aug 2002  |  IP: Logged
election model 2004
rabble-rouser
Babbler # 6305

posted 03 February 2006 04:14 PM      Profile for election model 2004   Author's Homepage     Send New Private Message      Edit/Delete Post
quote:
Originally posted by adma:
Oh, and speaking of electionprediction, 278 out of 308, 90.3% accuracy...

yes, milton did well also.
DemocraticSPACE = 283 out of 308 = 92%
Election Prediction Porject = 278 out of 308 = 90%


From: CA and ON | Registered: Jun 2004  |  IP: Logged
election model 2004
rabble-rouser
Babbler # 6305

posted 03 February 2006 04:24 PM      Profile for election model 2004   Author's Homepage     Send New Private Message      Edit/Delete Post
quote:
Originally posted by peterjcassidy:
Much appreciated. Can you give us some sense of the minor shifts that made a big difference?

For example, from some limited evidence, I believe there was a slight shift from the NDP to the Libs in Ontario,particularly in the Hamilton region, maybe a percentage point or two, in the last few days of the election. If that shift had not occured the NDP maybe would have picked up 2 or 3 more seats in Ontario from the Libs, giving us the balance of power. (This is also one reason HESC was closer than it should have been, 500 or so of our supporters probably went to the Libs to "stop Stephen Harper".)

Then what about Quebec? Would a percentage or two staying with the Bloc or Liberals in the Quebec city region have made a big difference for the Cons?

Was the variance in particular ridings where the atual result differed from the predicted generally within the margin or error- e.g. the predicion was party X wins by less than 5%, they lose by less than 5%?


And what your model provides is not just the provincal breakdown, but the regional. So if the model was out one or two seats in a region, does that mean that the region went though a shift in the last few days or that the winning party was stronger locally than projected? Any thoughts would be appreciated.

Thanks again. And what is your next project?



there weren't that many surprises, actually. certainly starting about 4-5 days prior to the election, we started to see a shift back to the liberals in ontario - but this was as much at the expense of the tories as the NDP. the NDP did just about as well as could be expected in ontario. the biggest shift away from the NDP came in northern ontario, where they didn't substantively improve on their 2004 results, but it was not surprising given what i saw in the polls (i forecast them to pick up 1, but they didn't pick up any). i was surprised the tories didn't do better in mississauga/brampton - i had a few seats as neck-and-neck, and expected the tories to pick up a couple seats here, but they were shut out completely (and convincingly). i missed 4 seats in mississauga, but all 4 were in the too close to call category for me (they turned out not to be close at all).

the shift from bloc to tory was a little more pronounced in the final days in and around quebec city. i had levis-bellechasse and beauport-limoilou as neck-and-neck, so no surprise that they went tory. i was surprised at roger clavet (louis-hebert) and richard marceau (charlesbourg-haunte-saint-charles) losing their seats, since they were high profile people for the bloc, particularly marceau. so those 2 seats were surprises. i missed 4 seats in quebec city, 2 were surprises, 2 were not.

those 2 cities account for 1/3 of the incorrect calls. the others were ones that were neck and neck and could have gone either way.

my next project will be to catch up on my real work! i'll track the 2007 ontario election, for sure.


From: CA and ON | Registered: Jun 2004  |  IP: Logged
Stockholm
rabble-rouser
Babbler # 3138

posted 03 February 2006 05:44 PM      Profile for Stockholm     Send New Private Message      Edit/Delete Post
quote:
I believe there was a slight shift from the NDP to the Libs in Ontario,particularly in the Hamilton region, maybe a percentage point or two, in the last few days of the election. If that shift had not occured the NDP maybe would have picked up 2 or 3 more seats in Ontario from the Libs,

Excpet that there were no real "near misses' for the NDP anywhere in Southern Ontario. It would have taken more than an extra percent or two to have taken Welland or Oshawa or even beaches-East York. Northern Ontario was a bit of a heartbreak, but the NDP did gainm ground in that region, but none of those seats were super close in 2004, so it was a big hill to climb. The North tends not to swing as much as the south and there is more loyalty to incumbents.


From: Toronto | Registered: Sep 2002  |  IP: Logged
Gir Draxon
leftist-rightie and rightist-leftie
Babbler # 3804

posted 04 February 2006 05:34 AM      Profile for Gir Draxon     Send New Private Message      Edit/Delete Post
I definately thought Allan Cutler was going to win. That was disappointing. I would have loved to see him as the Minister of Public Works.
From: Arkham Asylum | Registered: Feb 2003  |  IP: Logged
Thrasymachus
rabble-rouser
Babbler # 5747

posted 04 February 2006 05:54 AM      Profile for Thrasymachus     Send New Private Message      Edit/Delete Post
quote:
I definately thought Allan Cutler was going to win. That was disappointing. I would have loved to see him as the Minister of Public Works.
Apparently the guy is an unbelievable jerk in person. His negative charisma actually caused him to end up with results that were similar to the 2004 Ottawa South candidate who was caught driving to an all candidates meeting on a DUI related suspended license.

From: South of Hull | Registered: May 2004  |  IP: Logged
the bard
rabble-rouser
Babbler # 8375

posted 04 February 2006 03:39 PM      Profile for the bard     Send New Private Message      Edit/Delete Post
David McGuinty may have lost a bit of his "natural support" at the time of the extreme unpopularity of his brother's government.

Not to mention Ottawa South has a lot of civil servants who by and large don't like the idea of a Harper government. I under-estimated the number of Liberal seats by 10, but I was sure they'd win Ottawa South.


From: Toronto | Registered: Mar 2005  |  IP: Logged
nicky
rabble-rouser
Babbler # 10066

posted 05 February 2006 01:24 AM      Profile for nicky     Send New Private Message      Edit/Delete Post
There is another explanation for the "unexpected" Liberal strength in Mississauga / Peel - long term demographics. The authors of The Emerging Democratic Majority argue that older suburbs have been tending Democratic for various demographic reasons. This seems to be reflected in the inner ring of the GTA. The Con success in winning seats was limited to the very fringe of the GTA. The inner ring stood resolutelyb against them.

I suspect if someone crunched the numbers for Peel, Vaughn, Richmond Hill, Markham, Pickering, Ajax,( as opposed to Burlington, Milton, Uxbridge, Clarington,) over the last three elections the relative swings would be much more favourable to the Liberals than the province as a whole.I suspect this would also apply to Richmond, Burnaby and Surrey.

Peel was very marginal territory for years but that seems to be ending. It has become a Liberal stronghold with margins comparable to many Toronto ridings. McGuinty also piled up enormous margins there.

The trend is explained by people and jobs moving from the inner city to the immediate suburbs as well as by high immigration and immigrants becoming voters. The suburbs are becoming more like the core in their demographics and their voting patterns as well.


From: toronto | Registered: Aug 2005  |  IP: Logged
inkameep
rabble-rouser
Babbler # 3357

posted 05 February 2006 02:35 AM      Profile for inkameep     Send New Private Message      Edit/Delete Post
quote:
Originally posted by nicky:
I suspect if someone crunched the numbers for Peel, Vaughn, Richmond Hill, Markham, Pickering, Ajax,( as opposed to Burlington, Milton, Uxbridge, Clarington,) over the last three elections the relative swings would be much more favourable to the Liberals than the province as a whole.I suspect this would also apply to Richmond, Burnaby and Surrey.
In Richmond, Burnaby and Surrey the Liberal share of the vote fell by 3% between the 2004 and 2006 elections, closely reflecting the Liberal decline in the province as a whole.

Richmond, Burnaby, Surrey (7 ridings)
Percentage of valid votes by party

2004 2006 change

Conservative 33.2% 36.0% +9%
Liberal 33.0% 32.0% -3%
NDP 23.3% 27.3% +17%
Ind (Cadman) 5.0% nil n/a


From: Vancouver | Registered: Nov 2002  |  IP: Logged
Stockholm
rabble-rouser
Babbler # 3138

posted 05 February 2006 01:18 PM      Profile for Stockholm     Send New Private Message      Edit/Delete Post
I think that the strength of ther NDP in BC masks the pattern a bit. But it is worth noting that in 2000, the Tories won every single seat in the GVRD apart from those in the City of Vancouver and Svend Robinson in Burnaby-Douglas. Then in 2004, the lost North Vancouver and Richmond to the Liberals (actually its a bit complicated in Richmond because of party switching) and the NDP took the new seat of Burnaby-New Westminster. Then in 2006, the NDP also picked up New Westminster Coquitlam and Surrey North and the Liberals added on West Van and Newton-North Delta. In two elections the Tories have gone from having lamost total control of the Vancouver suburban seats to being reduced to just a handful of ex-urban seats.
From: Toronto | Registered: Sep 2002  |  IP: Logged
inkameep
rabble-rouser
Babbler # 3357

posted 05 February 2006 04:43 PM      Profile for inkameep     Send New Private Message      Edit/Delete Post
Yes, this time around the Conservatives lost 3 of the 7 seats they held in the inner suburbs of Vancouver. But these losses appear to have been caused by small shifts in close three-way races, not by a general erosion of Conservative support. In fact, overall the Con share of the vote in Vanc’s inner suburbs increased from 34.5% in 2004 to 36.0% in 2006.

The other thing worth noting is that although the City of Vancouver did not elect a single Conservative MP, the Con share of the vote in the city went up by more than 2 percentage points (to 22.1%) while the NDP share fell by 2 percentage points to (30.0%).


From: Vancouver | Registered: Nov 2002  |  IP: Logged
Stockholm
rabble-rouser
Babbler # 3138

posted 06 February 2006 12:02 AM      Profile for Stockholm     Send New Private Message      Edit/Delete Post
But keep in mind that the Conservatives managed to lose 3 (4 if you count Cadman's seat) in the Vancouver area in an election where in every part of the country the Conservatives were making major gains. Imagine, what would have happened to the Tories in BC if the Liberal national campaign hadn't fallen apart and if the election had been more like 2004? The Tories could easily have done far worse and the fact that BC was the only place in Canada were they have lost ground in each of the last two elections, while increasing everywhere else shows that they are on a longterm slide.

Next election, when BC residents turn with a vengeance on the Tories for being the incumbents, watch them drop another half dozen seats there.


From: Toronto | Registered: Sep 2002  |  IP: Logged
Pogo
rabble-rouser
Babbler # 2999

posted 06 February 2006 02:20 AM      Profile for Pogo   Author's Homepage     Send New Private Message      Edit/Delete Post
A bit of thread drift, but I don't think it deserves its own thread. Talking with people I find that a great number of people vote quite erratically and even more consider it. Is there any research of how consistent individuals are in their voting pattern. For example when the NDP gets 20% in a riding two elections in a row, how many of those votes were the same person voting NDP in both elections?

[ 06 February 2006: Message edited by: Pogo ]


From: Richmond BC | Registered: Aug 2002  |  IP: Logged
abnormal
rabble-rouser
Babbler # 1245

posted 06 February 2006 07:34 AM      Profile for abnormal   Author's Homepage     Send New Private Message      Edit/Delete Post
Pogo,

That's a good point. I suspect that there are more individuals that vote for different parties in different elections than one might think. For those who are not tied to a particular party because of that parties ideology the decision of who to vote for depends on a lot of things. Party platforms are only one of them. Things like "Who is the local candidate (as an indifidual)?", "Which individual will best represent my riding and me personally?", "What's the current hot button? (adscam, etc.)", "Do I like the party leader?", "Does the individual I'd like to elect stand a chance or should I vote for my second choice?", etc.

I'm sure this is not an insignificant percentage of the voting public.


From: far, far away | Registered: Aug 2001  |  IP: Logged
Reality. Bites.
rabble-rouser
Babbler # 6718

posted 06 February 2006 08:52 AM      Profile for Reality. Bites.        Edit/Delete Post
quote:
Originally posted by Pogo:
Is there any research of how consistent individuals are in their voting pattern.

Problem is such a study would be based on self-reporting. I suspect if you took a poll of Americans, for example, you'll find that Bush only got about 40% of the vote and there was 90% turnout.


From: Gone for good | Registered: Aug 2004  |  IP: Logged
Pogo
rabble-rouser
Babbler # 2999

posted 06 February 2006 11:56 AM      Profile for Pogo   Author's Homepage     Send New Private Message      Edit/Delete Post
I just find the talk about core support such as 'the NDP has regained its core support' very problematic. I know a lot of people who vote in circles, giving each party a try and then giving up on them when they are not the perfection they demand.
From: Richmond BC | Registered: Aug 2002  |  IP: Logged

All times are Pacific Time  

   Close Topic    Move Topic    Delete Topic next oldest topic   next newest topic
Hop To:

Contact Us | rabble.ca | Policy Statement

Copyright 2001-2008 rabble.ca