Yesterday's Angus Reid poll suggests the New Democrats are neck-and-neck with the B.C. Liberals in this provincial election. But the seat projection model prepared by SFU political scientist Kennedy Stewart – which is based in part on an average of five recent polls – continues to predict that Gordon Campbell's Liberals will win another majority.
Stewart now predicts the Liberals will win 52 seats, with the NDP taking the remaining 33 ridings.
"Using my model, it looks like the Liberals will retain a majority – perhaps even gaining a slightly higher proportion of legislative seats," he said.
"My confidence in this projection is boosted by the fact that the Liberal majority projections hold when the last five polls are averaged, and when projections are based on Angus Reid's most recent poll which has the Liberals at a lower support level than either Environics or Mustel."
Stewart’s regression model works by plotting a historical correlation between votes and seats for the B.C. Liberals, then applying that formula to an average prediction of the popular vote drawn from five recent polls. The model assumes that the NDP will claim all seats not won by the Liberals.
"As predicted, the polls have more or less converged," Stewart said. "Mustel's Liberal estimate has dropped from 52 per cent to 47 per cent, and Angus Reid's Liberal estimate has climbed from 42 per cent to 44 per cent. Environics also helps confirm this Liberal support, matching Mustel's 47 per cent estimate."
Nonetheless, the battle for bragging rights continues between Angus Reid and Mustel. Angus Reid is pioneering an online sampling technique, while Mustel is using traditional telephone polling.
"Mustel's reputation is most at risk as the company has consistently had the Liberals at a much higher levels of support," Stewart said. "The perception will be that they have been quite far off and consistently inflating the Liberals' support among voters, although it would be a mistake to make too much of this."
Monte Paulsen reports for The Tyee.
11
Login or register to post comments
G West
3 years ago
Hmmm?
And I see these guys have moved to 37 (LIB); 28 (NDP) and 20 (too close to call)...this is getting very interesting....
http://www.electionprediction.org/2009_bc/index.php
WHAT
3 years ago
With all due respect Mr. Stewart
How does your poll include the other influences of the day?....corruption, will to change...I see your results are based on the past, what about the present day political scene around the world?...the change in the US political arena. What about the fact that in 2001 people wanted change (or thought they needed it)and a landslide happened, did your methodology reflect that....now it seems that corruption as it did in the US turned the tide...
You are right the swing voters may be the deciding votes, but it could also be a landslide again, "the will to change seems to be in the air again". Now a model is a model! Why are you the only one who is jumping on this new methodology. The best way to find out is to ask people,,that's how you assemble the truth. Theory is great until you actually try to apply it. Human nature is hard to capture, so nobody takes it into account when devising a model. That's why I take all the factors/influences into account when I manage a project or define business requirements....Theory is perfect on paper, but not so in the implementation phase..I suggest our academics search for the truth not fame....
This is not about any party, this is about the FACTS, honestly taking the bias out of any methodology, science makes an attempt to do this at all costs, why not political science. I have a background in the physical sciences, political science, and computer science and am dedicated to the facts...You might fail me if i handed in a paper with unsubstantiated facts right! Facts count more, presentation is just decoration.
Pls I am not jumping down your throat, I am asking for answers, facts. It's ok to state a model, but you must always list the assumptions/strengths and weaknesses.
Van Isle
3 years ago
As I wrote earlier in the
As I wrote earlier in the week, it seems that these polls are more about trying to influence people on how to vote. Wonder how many more poll results are being released between now and election day?
secret cove
3 years ago
@Kennedy Stewart......
What a clown,what spin,what bogus methadology........
Your statement Kennedy ...." As predicted the polls more or less have converged"
What does that mean,more or less?
You updated your prediction yesterday to 51 Liberal seats to 34 NDP seats.....
And later in the day the Angus reid poll comes out calling it a virtual dead heat and your prediction didn`t change?
Explain that Kennedy?........
And...Mr What is absolutely correct,your model is based on the past,on voting patterns......You haven`t considered the most important thing...ISSUES,like Corruption,Run of River.......Let me put this to you Kennedy Stewart........
My prediction is 48 NDP seats 1 independent 36 Liberal seats......
If your wrong Kennedy Stewart.......
You will come to the Tyee and state to all of the readers that " I Kennedy Stewart am not smarter than a Canadian 5th grader!"
And after you make that statement you can go CRAWL back under your Rock!
frenchy mcswede
3 years ago
I really can't help but
wonder why mr stewart seems so pathetically eager for the ndp to lose, unless it's a combination of ego about not being wrong, combined with sour grapes about having done such a lackluster job when he ran for the ndp federally in vancouver-burrard.
And the angus reid poll should not be put on an equal par with the others, but given far more weight instead, being both more recent and much more scientific. It was more scientific, because it used online polling in every riding rather than outdated, older demographic favoring, telephone landline polling, and because it was more specific in that it asked: "Given the candidates in your riding who would you to be more likely to vote for?" This poll also showed james rapidly picking up support in the personal respect category, while the premier's numbers in this area, were static,or declining. It therefore showed POSITIVE MOMENTUM for james and NEGATIVE MOMENTUM for campbell ...anyone who can think knows how biased the recent mustel poll is, which canwest seems determined to use, (as of the late news last night) while ignoring the far better angus reid poll -another sign of desperation from bctv, who also ludicrously tried to claim campbell won the tv debate, "because he scored better on the economy..." I did NOT see any such scoring myself...
frenchy mcswede
3 years ago
Could someone who knows how post
links to the angus reid poll? there were other positives for james as well...I have only a vcr recording of ctv's report...which reported that people are beginning to warm to james, (now that the media can't just run edited 5 second clips of her, or tweaked, uncomplimentary photos...)
freebear
3 years ago
Polls Peddled to swing peoples votes to the (predicted) 'Winners
Its about political manipulation and attempting to influence/coerce voters before they actually vote.
Many people want to vote for a winner so they let the polls decide.
Whose poll you may ask?
Ahh there lies (pun iontended) the rub!
If no one spoke of there private vote preference, the polls would all be gone like ghosts!
Why do climate change models get challenged by neocons but not polling models?
We need a poll to answer that question!!!!
Wilfred Laurier
3 years ago
Accurate
I believe this seat projection will prove accurate.
Freddy Hutter
3 years ago
Swingometers are extremely accurate
The swingometer methodology being used by Kennedy is extremely accurate in jurisdictions across the world that are predisposed to two party dynamics. I have little doubt the 2009 results will follow the illustrated pattern.
Again, please note that the model tracks RESULTS ... not pre-election polls. It is not a predictive model.
The challenge for Kennedy is to determine which poll or combinaton of polls to input that will most accurately reflect the Election Day results.
Seat projections are an art form as much as a science. For them to have predictive value, one must be savvy in employing the correct input. Should he use the 2005 BC winner (Mustel) or the 2008 Federal winner (Anugus Reid)? Or an average?
On Wednesday nite, each modeler will reflect on shoulda/coulda/woulda.
At TrendLines, we have our favourite models, but each has let us down drastically at some time. The Ontario Election was a case in point. Chan's ElectionPrediction & Morrow's DemocraticSpace were disasters.
Even my own model efforts have sucked at times. This is why we highlight the AVERAGE (of models) on our website. Over time, this strategy produced the most consistently accurate results...
www.trendlines.ca/electbc.htm
G West
3 years ago
My pleasure frenchy
here's the whole Magilla:
http://www.theglobeandmail.com/v5/content/pdf/bcelectionpol050809l.pdf
And here's another, just fresh today, from Glen Robbins:
http://www.robbinssceresearch.com/polls/poll_590.html
Looks an awful lot like a dead heat to me - the party who gets the last woman to the polls is likely to win....
Freddy Hutter
3 years ago
6-model AVG seat projection
With three new polls today, the 6-model avg drops the Liberal tally to 53 as of today.
Projection for Election Day: 59 Liberals & 26 NDP
www.trendlines.ca/electbc.htm