Wednesday, January 30, 2013

B.C. Liberals narrow the gap slightly

With a new poll from the B.C.-based Mustel Group hitting the wire yesterday, the projection now shows a slightly closer race between the B.C. Liberals and B.C. New Democrats heading towards the May 14 election. Nevertheless, the probability that the NDP will win the popular vote on that day is still 93.4%, down from 94.6% as of Jan. 18.

The Mustel poll was in the field between Jan. 11 and 21, providing only a little bit of newer information than the Angus-Reid poll that was in the field Jan. 17 and 18. The margin in the poll (10 points) is the narrowest in any survey since March 2012, though that poll was also by Mustel. The last time that the gap between the two parties was 10 points or less on a regular basis was in November 2011. So this poll does show some change.

The projection with all available polling data as of Jan. 21 shows the New Democrats with 46.5% of the vote, down 1.3 points from Jan. 18. The Liberals have 32.9%, up 0.7 points, while the B.C. Conservatives trail with 10.7% (+0.3) and the B.C. Greens with 8.2% (+0.7). The projection range has dropped down for the NDP while the Liberals' range has increased, though the two parties still do not overlap (projected low of 43.7% for the NDP and projected high of 35.6% for the Liberals).

In terms of seats, the NDP has slipped one to 56, with the Liberals increasing to 28. One independent is projected to be (re-)elected as well. Nevertheless, despite the decrease in the NDP's seat projection, the party would still have a 95.2% chance of winning an election held today. The projected seat margin is simply too wide for the Liberals to be able to overcome it, even assuming the polls and the model are faulty (these are taken into account to calculate the probabilities). The projected low for the NDP is still in majority territory, with the party at between 44 and 69 seats. The Liberals could win between 13 and 41 seats, while between zero and four independents could be elected.

Looking forward to May 14, the forecast still considers a Liberal win to be a possibility. Their probability of winning the popular vote has increased slightly from 5.4% to 6.6%, but those are still very long odds. The forecast high for the Liberals is only 41.9%, while the forecasted low for the NDP is 40.5% (note that the projected vote has yet to fall outside the previous update's forecast). But if the numbers did hit those extremes and the regional distribution was especially beneficial to the Liberals, they could win as many as 59 seats. That is an absolutely best case scenario, however. The NDP could also win a landslide of 82 seats. With so much time still remaining between now and the election, almost anything is possible.

Regionally, the New Democrats slipped in metro Vancouver to 47.3% from 48%, and were down two seats due to the Liberals increasing by 1.4 points to 32.4%. The Conservatives took the biggest hit, dropping 1.7 points to 11.6%.

The NDP also dropped on Vancouver Island, decreasing 2.6 points to 50.2%. The Liberals fell, by 1.5 points, while the Conservatives were up 2.9 points to 8.4% and the Greens were up a point to 14%.

In the B.C. interior and north, the New Democrats had a tiny uptick of 0.2 points to hit 41.8%, while the Liberals dropped 1.6 points to 37.2%. The Conservatives were up 2.2 points to 11.7%.
The Mustel poll was relatively small, with a sample of 509 British Columbians. That is standard for the firm, however. They were last in the field Sept. 4-19, and since then the NDP dropped two points to 43% and the Liberals gained one to reach 33%.

Though that is a narrowing of the gap by three points, the +/- 4.3% margin of error needs to be emphasized. Neither of those shifts were statistically significant, and rather than consider that the gap could be even narrower we should, based on other data, lean in the opposite direction.

On the other hand, the swapping of seven points from the Conservatives to the Greens, putting them in a tie at 11%, was outside of the margin of error. It is difficult to imagine why the Conservatives would be shedding voters to the Greens. It could be the 'protest vote', or it could be something else entirely, i.e. Conservatives going to the Liberals, Liberals going to the NDP, and New Democrats going to the Greens, or just a product of the small sample size.

Mustel passed along some interesting regional variations, but the extremely small samples make them almost meaningless. But they do make some intuitive sense: the New Democrats doing very well in the city of Vancouver but in a neck-and-neck fight with the Liberals in the suburbs, and better numbers for the NDP in the north than in the southern interior.

But the numbers on Vancouver Island are the most fascinating. They put the Greens and Liberals in almost a tie for second place, while in and around Victoria itself the Greens are running a very strong second. However, the small sample sizes (with a crippling MOE of +/- 17 points) could make the numbers nothing more than a mirage. Nevertheless, they do match some of the scuttlebutt for the region.

The Mustel poll continues to show a gender problem for Christy Clark, as her party is running only four points behind the NDP among men but 17 points among women. Her approval rating stands at 38% in this poll, unchanged from Mustel's last survey, while her disapproval is 45%. Adrian Dix's approval rating increased to 45%, while his disapproval rating dropped six points to only 24%. But 32% still expressed no opinion.

Having a survey from Mustel is great for the aggregation, as they use live callers. Angus-Reid and Ipsos-Reid (at least in B.C.) use online panels while Forum uses IVR. It is good to have a mix of methodologies. But because of those different methodologies, and the somewhat different results they produce, it is difficult to assess what is going on in British Columbia. Is the gap narrowing? Have the New Democrats moved away from 50% support? We need more data.

Tuesday, January 29, 2013

Pre-Wynne, NDP led in Ontario

Just before the first ballot results of the Ontario Liberal leadership convention were to be announced, the Toronto Star released a Forum Research poll on the voting intentions of Ontarians. The poll showed that the  New Democrats narrowly led over the Progressive Conservatives, and that the leadership of Kathleen Wynne or Sandra Pupatello wouldn't change a thing. Will that hold?
Forum was last in the field on Dec. 17, and since then the NDP increased their support by four points to 35%. That gave them the advantage over the PCs, who slipped one point to 32%. The Liberals were unchanged at 27%, while the Greens were down three points to 5%.

The margin between the NDP and Tories is not statistically significant, but the changes in support of the New Democrats and Greens are (just).

The New Democrats were ahead in the 905 area code with 36%, trailed by the Tories at 30% and the Liberals at 28%. They were also ahead in the wider GTA with 34% to 32% for the Liberals and 28% for the PCs.

In southwestern Ontario, the NDP picked up 12 points to lead with 39%, putting them ahead of the Progressive Conservatives, who were down to 35%. The Liberals dropped eight points to only 18% in the region.

The NDP also had the advantage in northern Ontario with 42% to 30% for the PCs and 22% for the Liberals.

The Progressive Conservatives led only in eastern Ontario, with 42% to 28% for the Liberals and 24% for the New Democrats. The Liberals, meanwhile, were ahead only in the 416 area code, with 37% to 33% for the NDP and 25% for the PCs.

Most of the changes in support were within the margin of error, so we could be looking at a lot of statistical wobbling. On the face of it, though, it does appear that the Liberals experienced an uptick in Toronto while the NDP made gains in rural (eastern, southwestern, and northern) Ontario. Interestingly, the only major change in support occurred in southwestern Ontario, the only region of the province in which Sandra Pupatello's numbers were better than the generic ballot (23%). With her defeat, the Liberals might not expect to recover in southwestern Ontario anytime soon.

But the election of Wynne or Pupatello would not have changed much. Forum found that, under Wynne, the Liberals would have 26% support to 34% for the NDP and 32% for the PCs. For completely logical reasons, a Wynne leadership pushes one point from the Liberals (and the NDP as well, of course) to the Greens. In other words, her leadership of the party should not change anything in the short term - at least according to these numbers. Pupatello had an identical result, except the PCs dropped a point too. Gerard Kennedy would have done better, with 30% to the NDP's 32% and the Tories' 31%, but that is academic at this point.
With Ontarians' current voting intentions province wide, the Progressive Conservatives could eke out a tiny minority with 40 seats, to 37 for the New Democrats and 30 for the Liberals. The margin is close enough that if this were my final projection in an election campaign, I'd only give the Progressive Conservatives a 45% chance of winning the most seats (33% for the NDP and 22% for the Liberals). Put simply, a toss-up forecast.

The regional distribution is still quite stark, with the Tories winning 31 of their 40 seats in eastern, central, and southwestern Ontario, while the Liberals win 26 of their 30 in and around Toronto.

Note that this projection is based on the province-wide numbers. If it were regionally based (as the model will be when the next election rolls around), the NDP could do even better, winning more seats in the GTA and likely enough to put them narrowly ahead of the Tories.

The ace up their sleeve has to be Andrea Horwath, as she has an approval rating that is head-and-shoulders above her rivals. More than half, or 51%, approve of her, compared to 28% who disapprove. Still, one-in-five don't know what they think of her. New Democrats are almost unanimous, though: they give her an 80% approval rating.

Tim Hudak does worse among Tories, with 59% approval. Among Ontarians as a whole, he gets 27% approval to 53% disapproval. This is still a major problem for the PCs, just as Dalton McGuinty was a problem for the Liberals. His approval rating was 21% to 71% disapproval on the eve of his departure, with only 48% of Liberal supporters approving of him.

Will Ontarians warm up to Kathleen Wynne? Her numbers were improving as the convention approached. Only 8% of Ontarians thought she was the best choice to be OLP leader in November, improving to 23% in December and 25% just before the weekend. That still put her behind Pupatello (26%) and Kennedy (33%), but not by the same margins as in previous polls. Among Liberals, she was the choice of 27%, within the margin of error of Pupatello (28%) and Kennedy (33%). By that score, the Ontario Liberals did not make a bad choice.

Undoubtedly, a rash of polls will break out in the coming weeks as we try to gauge where the party stands now that Wynne is the premier and the province could potentially be heading to an election soon. If her numbers are good, we might see the NDP being a bit more amenable to compromise and the Liberal minority could survive. If her numbers are not, both Horwath and Hudak might turn the screws.

Monday, January 28, 2013

How the OLP leadership vote went down

This weekend's Ontario Liberal leadership convention did not have the kind of surprise some had thought possible - even likely - in a delegated convention. But Kathleen Wynne did emerge victorious after trailing on the first and second ballots behind Sandra Pupatello. How did the vote unfold throughout the day?

The initial results were probably the biggest surprise of the day, as the delegate count between Pupatello and Wynne was separated by only two votes, rather than the 41 that separated the two candidates among elected delegates. Many, including myself, expected Pupatello to be more of a favourite among the ex-officio delegates, but instead it appears that Wynne performed best among the party establishment.
It is difficult to assess exactly how the ex-officios voted, as we do not know the number of elected delegates who were present at the convention. It was not 100% turnout, as the total valid votes on the first ballot numbered 2,084. The number of elected delegates should have numbered 1,857, but instead was closer to 1,765, as the OLP has told me that some 320 ex-officios registered at the convention.

So, about 92 elected delegates were not able to make it to the convention. We don't know how these were distributed between the candidates. Could Pupatello's delegates, many of whom came from the northern, southwestern, and eastern corners of the province, have been no-shows in higher proportions?

It is clear that some of Harinder Takhar's delegates did not show-up, as his first round result was 235 votes - nine fewer than his number of elected delegates. But we might be able to use that as a base for determining how many elected delegates did not show up: if we proportion those 92 no-shows by each candidate's share of the elected delegates, we get 12 no-shows for Takhar. Assuming he got three ex-officios is hardly implausible, so using this method to estimate the number of elected delegates seems safe enough.

If we apply this to all of the candidates, we get the number of elected delegates present at the first ballot as 484 for Pupatello, 445 for Wynne, 247 for Gerard Kennedy, 232 for Takhar, 194 for Charles Sousa, 100 for Eric Hoskins, and 64 independents.

That means that of the 384 independent and ex-officio delegates, Wynne got the support of 152, or 40% of them. Pupatello placed second with 115, or 30%, while Hoskins took 13%, Kennedy took 9%, Sousa 7%, and Takhar 1%. Even if Wynne got the support of every independent delegate, she still would have gotten the nod from 28% of the ex-officios, enough to keep her close to Pupatello's 36%. Of course, it is highly unlikely that Wynne got the support of every independent delegate. If we give Wynne some 70% of those independents (as many of them had been Glen Murray's), we end up with a near-tie in the ex-officios between Wynne and Pupatello.

It was Wynne's support among the ex-officios, plus her likely advantage among independents, that gave her the surprising first ballot results and dulled any sort of momentum Pupatello could have built with a strong first ballot performance.

Overall, Pupatello had 28.7% on the first ballot to Wynne's 28.6%, Kennedy's 13.5%, Takhar's 11.3%, Sousa's 10.7%, and Hoskins's 7.2%.

At that point, it was looking far more difficult for Pupatello. But she was still the favourite at that stage, as it was assumed she would pick up the endorsements of Takhar and Sousa, enough to give her the win. But Hoskins's strong support among the ex-officios gave him a bit more weight in the convention, and his endorsement of Wynne helped her more than his numbers suggested. And whereas people were not sure where Hoskins would go before the convention, Takhar's swing to Pupatello was expected. The momentum was Wynne's.
Though Takhar did not officially drop off the ballot, his announcement was as good as an official withdrawal as only 18 of his 235 supporters still cast their ballot in his favour (in all likelihood, they had voted before his announcement was made).

Pupatello made the biggest jump on the second ballot, gaining almost 57% of the new ballots on the table. Wynne took 40%, while 3% did not vote and another 1% went to Kennedy.

Pupatello increased her haul by 218 ballots to 817, or 39.4%. She picked up almost the entirety of Takhar's delegates. Wynne, however, gained 153 votes - more than the 150 that Hoskins had made available to her. She was up to 750, or 36.2%, having apparently taken some votes from Sousa, who dropped 19 to 203, or 9.8%. This should have been Kennedy's chance to make a move, but instead he gained only four votes, giving him 285 or 13.7% of the total. His path to victory was now definitively blocked.

(Note: More than just Sousa's delegates would have supported other candidates on the second and third ballots than the one they were initially elected to support. Estimating that kind of cross-pollination is near-impossible, so for simplicity I assume that delegates stuck with their candidate throughout the convention. If they didn't, the ones swapping from one candidate to the other probably cancelled each other out.)

If Sousa would have gone to Pupatello after the second ballot, she likely could have won it. Kennedy was always expected to go with Wynne, and the sum of their endorsements would have made it incredibly close (1,038 votes to 1,035). How the ex-officios would have swung in such a scenario, in order to give someone a respectable margin, is impossible to guess.

Instead, Sousa and Kennedy went over to Wynne and sealed her victory. She did not get all of their supporters, however, as 11% of the newly available delegates did not vote and 10% went over to Pupatello. She did get 79%, however, more than the 59% she needed in order to move ahead of Pupatello on the last ballot. But the results were definitive: 1,150 or 57% for Kathleen Wynne against 866, or 43%, for Sandra Pupatello.

This sort of delegated convention is exciting, as the turning points are obvious and, with the candidates moving around the floor with their sign-waving supporters, you have the benefit of actually seeing support swing from one side to the other. There were a few moments that pointed towards Wynne's victory: her stronger than expected first ballot results, the endorsement of Hoskins (who himself did unexpectedly well), and the double endorsement of Sousa and Kennedy. It is possible that each of these endorsements could not have occurred without the turning points that preceded them. However, it might be Pupatello's weaker-than-expected first ballot result that sealed her fate. The ex-officios hadn't sided with her as much as everyone thought they would, and that made all the difference.

How will the Ontario Liberals do now that Kathleen Wynne is their leader? We'll find out over the next few weeks, but it may take a lot longer for opinions to firm up. Conceivably, Wynne as Premier makes a snap election less likely, so she may be able to give herself the time to get ready for the next vote. It should be an interesting few months either way.

Thursday, January 24, 2013

With Takhar, does Kennedy stand a chance?

A report in the Toronto Star this morning says that Gerard Kennedy, who finished third in the delegate elections for the Ontario Liberal leadership race, is pushing hard to court Harinder Takhar, who finished fourth. Takhar's support would vault Kennedy into second place among elected delegates, making a path for the leadership much easier to map out. But just how easy is it?

After what happened in the last two delegated conventions for the federal and provincial Liberals, many suspect that a third-place finisher like Kennedy could work his way to the top through the balloting process. But Rob Silver, writing for Maclean's, does an excellent job of showing that Dalton McGuinty and Stéphane Dion's recent victories were exceptions rather than the rule. In addition, the mountain Kennedy has to climb is steeper than either McGuinty or Dion managed in their leadership wins.

But what if Kennedy was able to get Takhar over to his side? Takhar is probably the biggest fish to land in the OLP leadership race, as he is likely to have much more sway over his delegates than those of the other candidates. It is not hard to imagine that Takhar could justify moving over to Kennedy instead of Sandra Pupatello, who many have seen as the candidate he is most likely to support. Takhar would have much more to gain from being the kingmaker for Kennedy than he would being just one piece of Pupatello's victory. With her lead among the elected and ex-officio delegates, Pupatello could probably win without an explicit endorsement from Takhar. Kennedy would have a much bigger favour to return.

When I last mapped out the OLP convention, I did not consider the possibility of Takhar moving over to Kennedy (except in the scenario where I had Charles Sousa and Eric Hoskins endorsing him as well). So let's run the exercise again, with a few different assumptions.

First, we have to re-allocate the ex-officio delegates based on the endorsements that have been made since my original post. Instead of just doing a simple count, I will use the endorsement points system that I am currently employing for the federal Liberal leadership race. Note that the system is designed for federal races, but that it was also calibrated in part with provincial races.

ENDORSEMENT RANKINGS

Sandra Pupatello - 212 points - 48.8%
Kathleen Wynne - 131 points - 30.1%
Gerard Kennedy - 37.5 points - 8.6%
Eric Hoskins - 29.5 points - 6.8%
Charles Sousa - 17 points - 3.9%
Harinder Takhar - 7.5 points - 1.7%

This is one case where I suspect the endorsement rankings will lean too heavily towards the frontrunner, and will almost certainly reverse the order of the bottom three. It is extremely unlikely that Pupatello will come that close to winning, or even approaching 40%, on the first ballot. Instead, we will use this to distribute the 419 ex-officio delegates, with Pupatello taking 205, Wynne 126, Kennedy 36, Hoskins 29, Sousa 16, and Takhar seven.

We also have to distribute the 67 independent delegates that were elected. Instead of giving Wynne 70% of them, as I did last time due to Glen Murray's endorsement, I will give her 50% and distribute the rest proportionate to their elected and estimated ex-officio delegate support. That gives us the following first ballot estimate:

FIRST BALLOT ESTIMATE

Sandra Pupatello - 729 delegates - 32.0%
Kathleen Wynne - 628 delegates - 27.6%
Gerard Kennedy - 302 delegates - 13.3%
Harinder Takhar - 256 delegates - 11.2%
Charles Sousa - 224 delegates - 9.8%
Eric Hoskins - 137 delegates - 6.0%

Compared to the first ballot estimate I made in my previous post, Pupatello and Sousa picked up an extra point while Kennedy dropped one.

Now that we have made a plausible estimate of first ballot support, let's see how Takhar's support could change the race for Kennedy. We will assume that Hoskins and Sousa drop out and decline to make an endorsement, releasing their delegates. We will distribute their delegates proportionately to the top three candidates only. That gives us the following result:

Sandra Pupatello - 887 delegates - 39.0%
Kathleen Wynne - 765 delegates - 33.6%
Gerard Kennedy - 368 delegates - 16.2%
Harinder Takhar - 256 delegates - 11.2%

And now we see the problem. If Takhar publicly endorsed Kennedy and asked his delegates to support him, even with all 256 of them voting for Kennedy he would end up with only 624 delegates and 27.4% support. That still leaves him behind Wynne and forced to drop-off the ballot. Without higher ex-officio delegate support, he would need Takhar's endorsement to influence a large number of delegates that have already supported Pupatello and Wynne to go his way.

Could he have picked them up after the first and second ballots from Sousa and Hoskins? Instead of getting 18.3% of their delegates, as I awarded him earlier, he would need to get some 38%, and all of them coming from Wynne, dropping her to 18% of the Sousa and Hoskins delegates. That does not seem like a plausible scenario - if Kennedy took a larger share of the Sousa and Hoskins delegates, some of them would undoubtedly come from Pupatello instead of Wynne, increasing the number he would need.

Kennedy would stand a chance if he managed to capture some 50% of the delegates that initially supported Sousa and Hoskins, or if he was able to attract those delegates committed to Pupatello and Wynne on the first ballot. But if he was able to take 50% of the delegates released by Sousa and Hoskins, with the rest going to Pupatello and Wynne proportionately, we would get this result:

Sandra Pupatello - 826 delegates - 36.3%
Kathleen Wynne - 711 delegates - 31.2%
Gerard Kennedy - 483 delegates - 21.2%
Harinder Takhar - 256 delegates - 11.2%

Mathematically, it is now possible for Takhar's support to vault him into second place, pushing Wynne off the ballot. But it is not exactly an easy task: he needs 229 of the 256 delegates alloted to Takhar in this scenario, or 89.5%, with all of the remaining delegates going to Pupatello. If even 10 of those delegates went to Wynne instead, Kennedy would need over 93% of Takhar's delegates. In addition to Kennedy taking half of the delegates who supported Sousa and Hoskins, could Takhar really deliver almost the totality of his elected delegates? It starts to stretch the imagination.

Kennedy needs the convention to go incredibly well for him. Unless he somehow becomes an establishment favourite and picks up a swathe of ex-officios (which, considering his history, seems improbable), he probably needs Takhar to endorse him immediately after the first ballot if he can't somehow get Hoskins or Sousa to do so. With Kennedy managing such a coup after the first ballot, he might be able to show himself to be the candidate with the momentum, giving him the necessary support from Hoskins and Sousa. Otherwise, it is difficult to see why those delegates committed to Sousa or Hoskins would vote for Kennedy instead of one of the frontrunners, or why the delegates committed to Pupatello or Wynne would suddenly jump ship. The numbers are there for Kennedy, but the odds are very slim.

Tuesday, January 22, 2013

Conservatives lead in two polls

It is not a common occurrence for two polls to be released on the same day with both being in the field at the exact same time. But that happened on Friday, as Forum Research and Angus-Reid both published their federal polls conducted on Jan. 16-17, surveying some 3,600 people. And, most shockingly, the polls were not in disagreement about much.
We'll start with the Forum poll. The firm was last in the field Dec. 18, and since then the Conservatives increased their support by five points to 36%, a statistically significant lead. The NDP was unchanged at 28%, while the Liberals were down two points to 25%.

The Bloc Québécois was up one point to 7%, while the Greens were down two to 4%.

The shifts in support for the Conservatives and Greens were outside their respective margins of error.

The Conservatives led in Ontario, where they were up six points to 40%. The NDP and Liberals were tied at 27% in the province. The Tories also led in Alberta with 63%, while the New Democrats jumped nine points to 20% and the Liberals placed third with 11%.

In British Columbia, the Conservatives were narrowly ahead with 40% to 36% for the NDP. The Liberals fell eight points to 15%. The Tories also had the advantage in the Prairies, with 49% (+13) to 33% for the NDP and 11% for the Liberals.

The Liberals led in Atlantic Canada with 46%, putting them well ahead of the Conservatives at 28% and the NDP at 23% (a drop of 13 points). The Liberals were also in front in Quebec, with 29% to 26% for the NDP, 25% for the Bloc, and 16% for the Conservatives.
Angus-Reid has been busy this month, as they were last in the field only a few weeks ago on Jan. 2-3. The Conservatives have not budged since then, holding at 35% support. The NDP was down four points to 29%, while the Liberals were up three points to 22%.

The Greens and Bloc Québécois were both unchanged at 6% apiece.

My complaints with both of these firms are the same: Forum should be releasing unweighted data, while Angus-Reid should at the very least be releasing sample sizes for their detailed breakdowns, in addition to their unweighted samples.

The Conservatives also led in this poll in Ontario with 38%. The NDP dropped six points to 29% while the Liberals increased to 26% in the province. In Alberta, the Conservatives led with 62% to 18% for the NDP and 11% for the Liberals, while in the Prairies the Tories led with 53% to 25% for the NDP and 15% for the Liberals.

The Conservatives and NDP were tied in British Columbia with 35% apiece, while the Liberals made an eight-point gain to hit 19% in the province.

The New Democrats led in Quebec with 32%, though that represented an eight-point drop since earlier this month. The Bloc trailed with 25%, while the Liberals were at 21% and the Conservatives at 17%.

The Liberals were in front in Atlantic Canada with 35% to 31% for the Conservatives and 29% for the New Democrats.

These are two very similar polls, with a variance of no more than one point for the Conservatives, NDP and Bloc at the national level, two points for the Greens, and three for the Liberals (here's probably why). We see the same gender split as well (39% or 41% for the Tories among men, 25% or 28% for the NDP, and 31% or 30% for the Conservatives among women, 30% for the NDP). The Conservatives held statistically significant leads in Ontario and Alberta and were ahead in the Prairies, while the race in British Columbia is neck-and-neck. The Liberals were ahead in both polls in Atlantic Canada, and Quebec is a jumble of a three-way race between the NDP, Liberals, and Bloc (pegged at 25% in both polls).

There are some things to note, however. The results in Quebec are particularly interesting. Forum has the Liberals at 29% in the province, a very high score. In fact, in the six polls that have put the Liberals at 29% or more in Quebec since September, five of them have been from Forum. We could be looking at a methodological bias, or that everyone else is under-estimating Liberal support in the province.

Atlantic Canada is a region to keep an eye on, as it appears to be in flux. According to the monthly averages, the New Democrats were leading in the region between April and September, before moving into a tie with the Liberals in October and November and falling to second in December. Both of these polls put the New Democrats in third in Atlantic Canada. The NDP has been in second or third in 12 of the last 15 polls now, and were third in almost half of those.
The seat outcomes from these two polls are mildly different, and only because of the results in Quebec. With Angus-Reid's numbers, the Conservatives would win 162 seats on the 338-seat map, falling just short of a majority. The NDP wins 96 seats, the Liberals 65, the Bloc 14, and the Greens one. Awkwardly for the opposition, the NDP, Liberals, and Greens can combine for only 162 seats - meaning the Bloc Québécois would hold the balance of power.
With poor results in Quebec, the NDP falls to third party status with Forum's numbers, as the Liberals edge them out with 75 seats to 70. The Conservatives win 163 seats, again just short of a majority, while the Bloc returns with 29 seats and the Greens hold their one.

What is costing the Conservatives their majority in these two scenarios? British Columbia and the Prairies, primarily - but also Ontario, where the Tories are not winning the new seats added to the province.

Liberal leadership race

The two polls also quizzed Canadians on the on-going Liberal leadership race, giving us plenty of data to work with. The most interesting to me was the breakdown by Angus-Reid of just how much Canadians know about the candidates:
Aside from Justin Trudeau and, to a lesser extent, Marc Garneau, the answer is not that much.

Trudeau's numbers are by far the best, as 22% claim they "know his background and the ideas he believes in very well". That is a question that is very similar to what CROP recently asked about the Quebec Liberal leadership race, oddly enough.

Garneau was second at 9% on this classification, while everyone else was at 2% or less - political junkies only.

When you add the people who know their background and ideas a little, which is about what you should expect from Canadians this far from an election, Trudeau's advantage is amplified. Fully 48% have a decent idea of him, 70% if you include those who know him very well. That is huge, and Garneau hardly competes with a combined 42% awareness. Martin Cauchon places third with 15% (13% know him a little) while Martha Hall Findlay ranks fourth at 13% (11% know her a little). No one cracks 10% after that.

One in five Canadians recognize the name only of Trudeau, Cauchon, Hall Findlay, and Deborah Coyne (perhaps confused by her columnist cousin), while one in four recognize Garneau's name (likely due to his astronaut past). But over 62% of Canadians have never heard of the candidates outside of Trudeau and Garneau, and that increases to over 70% for the bottom five and 80% for the bottom three.

Forum finds that 34% of Canadians think Trudeau is the best option, though that is a drop of five points since their December poll. He is followed at length by Garneau (10%), while no one else does better than 3% (Hall Findlay and Cauchon managed that). Among Liberal supporters, Trudeau's lead increases to 63% to 6% for Garneau. Hall Findlay and Joyce Murray managed 3%.

Accordingly, 55% of Canadians think Trudeau will win (according to Angus-Reid). Another 7% think Garneau will win, while no more than 1% give any of the others a chance. Trudeau and Garneau are the most appealing to voters, as 40% told Angus-Reid a Trudeau win would make them more likely to vote Liberal. 23% said the same for Garneau, while no more than 7% said another candidate would make them more likely to vote Liberal.

More specifically, in a head-to-head match-up with the other parties Forum gives a Trudeau-led party the lead with 35% to 33% for the Conservatives and 21% for the NDP. Fair warning, however: that is a drop of four points for the Trudeau Liberals since December. Angus-Reid finds something similar, giving the Trudeau Liberals 34% of the vote to 33% for the Tories and 22% for the NDP - a drop of eight points since earlier this month.

Forum didn't ask about the other candidates, but Angus-Reid did: Garneau improves Liberal fortunes slightly to 25%, while Hall Findlay drags them down to 19% and Cauchon to 17%.

The votes are there for the taking, though. Angus-Reid founds that 38% say they are very or moderately likely to vote Liberal in the next election. To be precise, 14% said they were very likely to vote Liberal (the base) while 24% said they were moderately likely (the swing voters). These polls suggest that only Trudeau and Garneau are likely to attract many of those swing voters to the Liberal fold.

Monday, January 21, 2013

Liberals gain in Interior but B.C. NDP still heavily favoured

The probability of the B.C. New Democrats winning the popular vote and, quite likely, the election on May 14 has increased from 91.7% to 94.6%, despite the B.C. Liberals making important gains in the B.C. Interior and North.

With the release of a new poll by Angus-Reid, the vote and seat forecasts for the upcoming British Columbia election have been updated. As of Jan. 18, the last day that Angus-Reid was in the field, the NDP is projected to win between 45 and 74 seats, with the most likely outcome being 57 seats. The B.C. Liberals are projected to win between nine and 40 seats, with 27 seats being the most likely result. If an election were held today, the New Democrats would have a 96.4% chance of winning it.

Since the last projection, which was based on polling data as of Nov. 30, the New Democrats have dropped 0.7 points to 47.8% support, while the Liberals have dropped 0.9 points to 32.2%. The B.C. Conservatives have hardly budged, down 0.1 point to 10.4%, while the B.C. Greens are up 1.7 points to 7.8%. (It is worth noting that all of these results are within the range forecasted for Nov. 30's data). The projected vote ranges for the Conservatives and Greens do overlap (at 8.4% to 12.4% and 6% to 9.6%, respectively), while those of the NDP and Liberals do not. The Liberals are projected to take no more than 35.3% of the vote, compared to a low of 44.5% for the New Democrats.

The likely forecast also does not foresee a scenario in which the Liberals move ahead of the New Democrats in the popular vote, with a forecast high of 41.2% for the Liberals on May 14. Nevertheless, there is still an outside chance that the Liberals could win as many as 57 seats on election night, with the NDP winning as few as 25. That is, of course, a very unlikely outcome. In fact, the forecast high for the Liberals has dropped by six seats, while the forecast low for the NDP has increased by three. The scenarios in which the Liberals come out victorious are dwindling.

Indeed, the likelihood of the Liberals finishing first in the popular vote are extremely low: 5.4%, down from 8.3% on Nov. 30. In other words, in only one out of 20 cases would the Liberals be expected to overcome a 15.6-point deficit in fewer than 120 days before an election. The amount of runway remaining for Christy Clark is running out.

There is some good news for the Liberals, however. Their projected support in the Interior and North has increased 5.2 points to 38.8%, putting them less than three points behind the NDP, who are down 4.3 points in the region. As a result, the Liberals are now projected to win 16 seats, an increase of seven. The NDP dropped seven seats to 15, meaning the Liberals are in a better position to win more seats than the NDP in the region as a whole. However, the forecasted high and low results for the NDP are more favourable.

The NDP's position has also improved in Vancouver, where the party is up one point to 48% support. The Liberals dropped 4.2 points to 30.7%, and in the process dropped three seats to the New Democrats. The NDP is forecast to win no fewer than 17 seats, and as many as 38, in the metropolitan region on May 14.

The Greens made the biggest gain of any party in any region on Vancouver Island, up 6.1 points to 13%. They have displaced the Conservatives in third place, while the NDP and Liberals were both down more than a point (the NDP picked up a seat). The Greens are still not projected to be in range of any seats on Vancouver Island based on current levels of support, but the forecast does put two seats in play for the party. The most likely scenario, though, is an NDP sweep.
The poll that prompted the projection update did not have too many changes from Angus-Reid's last survey from November. Since then, the New Democrats dropped one point and decreased to 46% support, while the Liberals were up two points to 31%. The Conservatives were down two points to 10%, while the Greens were up one to 10%. Support for other parties and independents increased one point to 3%.

All of these shifts are within the margin of error (assuming random samples, of course, which an online poll does not have). Angus-Reid also still does not put sample sizes in their regional breakdowns, something which absolutely should be done.

The Liberals are certainly trending upwards in Angus-Reid's polling: 22% in July, 25% in September, 26% in October, 29% in November, and now 31% in January. But the party has been gaining one to three points each month, and only four months remain between now and the election. That means if the trend continues, the party could pick up between four and 12 points - putting them between 35% and 43%. Over that time, the NDP has scored between 46% and 49% in every poll, without any discernible trend. If the Liberals do get to over 40% on election night, much of it will likely have come from the Conservatives. It won't be enough unless they start seriously eating into NDP support.

Regionally, the New Democrats led in Vancouver with 46%, and were trailed by the Liberals at 29% and the Conservatives at 13%. The NDP was also in the lead on Vancouver Island with 51% (down 11 points). The Liberals were second with 27% and the Greens had 17%. The NDP was ahead in the North with 45% to 32% for the Liberals and 13% for the Greens.

In the Interior, however, the Liberals moved into a tie with the NDP at 39%. The Greens placed third with 11%, while the Conservatives dropped seven points to 9%. That drop in Conservative support appears to be real, but the gains of the Liberals and the small loss incurred by the NDP could be statistical wobbling. It does make intuitive sense, though, to assume that the gain made by the Liberals is not an anomaly and that Conservative support is indeed drifting disproportionately to the Liberals in the Interior. That is not necessarily trouble for the NDP as they still hold wide leads in Vancouver and on the Island, but it is the first step towards a Liberal recovery, if one is to occur.

The poll showed no significant changes in personal ratings for the leaders, with Adrian Dix leading on the Best Premier question with 29% to 19% for Clark and 5% for Cummins. Dix still has the best approval rating at 46% to 34% disapproval. Clark scored 31% and 56% on approval and disapproval, while John Cummins's approval rating is a woeful 13%. His disapproval rating is almost as high as Clark's, at 51%, and this despite 36% not sure of their opinion of him (compared to 13% not-sures for Clark).

Dix still holds a major advantage over Clark, as he is seen as the best leader on the issues of health care, crime, the economy, education, and federal-provincial relations. He only trails Jane Sterk on the environment, as one might expect. The top issue is still the economy for British Columbians (increasing to 28%), and Dix still has the edge over Clark on a question that is supposed to be an NDP leader's Achilles' heel. It is hard to see a route for a comeback for the Liberals.

Friday, January 18, 2013

What fate awaits the OLP under Wynne or Pupatello?

Barring a big surprise next weekend, either Kathleen Wynne or Sandra Pupatello will be the new leader of the Ontario Liberal Party and premier of the province. How might they do in the next election?

I wrote about this for The Huffington Post Canada today, and I invite you to take a look. In the article, I spell out how I think it could play out. Here, let's take a look at what the numbers show - or at least what can be done with them.

The most recent data about who Ontarians like for the leader comes from a December 17 poll from Forum Research. It asked who Ontarians thought would make the best leader of the Ontario Liberal Party. Gerard Kennedy placed first with 36%, followed by Wynne at 23% and Pupatello at 20%. No other candidate hit double-digits. The poll also broke the numbers down by voting intentions, giving us an idea of what supporters of other parties think of the race.

Let's run a hypothetical exercise using those numbers to determine the potential support Wynne and Pupatello could draw from other parties. Forum did not disclose how many respondents said "I don't know" or "None of the above", but from the sample sizes we can estimate that number.

Kathleen Wynne was the favourite choice of 10% of NDP voters (20% of NDP voters who expressed an opinion, but 10% of the entire pool of NDP supporters) and 6% of Tories. She was also the favourite of 16% of Green voters and 10% of voters who said they intended to support an other party. In a best case scenario where Wynne is able to draw all of those voters over to the Liberals, and using the current Ontario poll averages as a base, we get the following outcome:

WYNNE VICTORY SCENARIO

Liberals - 33.5%, 46 seats
Progressive Conservatives - 32.1%, 35 seats
New Democrats - 27.4%, 26 seats
Greens - 6.2%, 0 seats
Others - 0.8%, 0 seats

Wynne draws away a good deal of support from the New Democrats - and even a few Tories. It keeps the Liberals in power, but makes reliance on another party for survival all the more important. Taking votes from the Greens also helps the Wynne Liberals quite a bit.

Pupatello is not so fortunate. Her numbers among PC voters are the same - she takes 6% of their supporters - but she only attracts 8% of New Democrats and 3% of Greens. That makes it a bit more difficult for her to put the Liberals over the top, which she does by a hair:

PUPATELLO VICTORY SCENARIO

Liberals - 32.2%, 44 seats
Progressive Conservatives - 32.1%, 36 seats
New Democrats - 27.7%, 27 seats
Greens - 7.2%, 0 seats
Others - 0.8%, 0 seats

One of the problems here is that the PC vote is pretty much locked in - only 28% said they thought one of the Liberal candidates would be a good option to lead that party, compared to 49% of New Democrats and 60% of Liberals. It was almost as if the vast majority of Tories would not even consider the question. That bars the door to Pupatello somewhat, as her appeal among New Democrats is more limited.

Nevertheless, in both of these scenarios the Liberals survive. That is a far cry from where they are in the polls right now. But what if things take a turn for the worst? Who could salvage things best?

This is a bit trickier to determine, and requires an even bigger assumption to be made. For the sake of the exercise, let us assume that if Wynne wins, the OLP supporters who said they preferred Pupatello decide to vote for another party, and if Pupatello wins, the Wynne OLP voters jump ship. These can be portioned out according to how PC, NDP, and Green voters considered the race, the assumption being that OLP voters who like a certain candidate probably agree more with the supporters of other parties who also like that candidate. It is a crude way to go about it, but I think it is the best we can do under the circumstances.

Contrary to province-wide opinion, Pupatello was the second choice among Liberal voters behind Kennedy, rather than Wynne. That means that a Wynne victory pushes 16% of OLP supporters to other parties. 52% of them go to the New Democrats, 39% of them to the Tories, and 4% apiece to the Greens and other parties. That results in:

WYNNE COLLAPSE SCENARIO

Progressive Conservatives - 35.8%, 57 seats
New Democrats - 32.7%, 37 seats
Liberals - 22.8%, 13 seats
Greens - 7.6%, 0 seats
Others - 1.1%, 0 seats

The Tories win a majority government while the NDP is vaulted to Official Opposition status. The Liberals are reduced to a rump. This is not an unthinkable scenario, as if Wynne comes out of the gate and trips those left-of-centre OLP voters could easily move over to the NDP. This happened federally as Liberal supporters saw that the NDP became the more viable anti-Conservative option. And coming from the centre-left of the party, Wynne pushes more voters over to Tim Hudak as well.

PUPATELLO COLLAPSE SCENARIO

Progressive Conservatives - 35.2%, 55 seats
New Democrats - 32.1%, 36 seats
Liberals - 23.7%, 16 seats
Greens - 8.0%, 0 seats
Others - 1.1%, 0 seats

If Pupatello wins and she falls flat on her face, the 13% of OLP voters who consider themselves Wynne supporters abandon the party. 48% go to the NDP, 31% go to the PCs, 16% go to the Greens, and 5% to the other parties. That still delivers the PCs a majority, but only just, and the rump the Liberals are reduced to is somewhat larger. This is also a plausible scenario, as Pupatello is probably less likely to lose centrist and centre-right OLP supporters than Wynne would, which might make her seem like a more viable anti-Hudak option than a Wynne-led party that is collapsing.

This exercise shows the risks and rewards that come with a Wynne or Pupatello leadership. Wynne can probably attract more support from the left, putting them ahead of the Tories, who are relatively solid. She potentially has more upside. But by pushing the party to the left, she might make the choice to jump over to the NDP easier than under Pupatello, meaning a Wynne collapse could be worse for the Liberals than a collapse under Pupatello.

All of this assumes the status quo, however. If Hudak's campaign falls apart, Pupatello might be better placed to scoop up those disillusioned PC voters than Wynne. If Andrea Horwath's campaign collapses, disappointed NDP voters may be more willing to cast their ballot for the Wynne Liberals than they would a Pupatello-led party.

But what about Kennedy? Polls show he is more widely liked than either Pupatello or Wynne, even if the OLP itself is not all that keen on him. If he does pull off an upset next weekend, what could happen then?

KENNEDY VICTORY SCENARIO

Kennedy scored very well among NDP voters in the Forum poll. He pulls 23% of NDP votes to the OLP. He also draws 9% of Tories and 18% of Greens.

Liberals - 38.6%, 60 seats
Progressive Conservatives - 31.0%, 29 seats
New Democrats - 23.4%, 18 seats
Greens - 6.1%, 0 seats
Others - 0.9%, 0 seats

His upside among PC voters is still rather limited, but he completely erases the gains the New Democrats have made since the last election. He gives the Liberals a majority government. These might be fairy-tale numbers due to Kennedy's name recognition alone (Hudak and Horwath are also well-known at this point, dulling his potential upside), but it is hard to argue that Kennedy wouldn't stand the better chance in a snap election than Pupatello or Wynne. If you have to spend the first two weeks of a campaign introducing yourself, you will be two weeks behind the other leaders.

Of course, this exercise is highly hypothetical and makes plenty of assumptions that may or may not be warranted. But the results do make a lot of intuitive sense. Unless Wynne or Pupatello can find some support across the aisle, we will probably find out sooner rather than later how they will do.

Wednesday, January 16, 2013

Couillard favoured in PLQ race

Over the weekend, La Presse reported on a poll done by CROP on the PLQ leadership race in Quebec. The results were not surprising, but did show that Philippe Couillard is the odds-on favourite to win it. It also showed that a Couillard-led Liberal Party could oust the PQ from government.
Since CROP's last poll taken Dec. 3-10, the Parti Québécois and Liberals have traded five points, with the PQ dropping to 31% and the Liberals increasing to 30%. The CAQ was up two points to 27%, while Québec Solidaire was down two points to 8%. Another 4% said they would vote for another party.

These results are virtually identical to the last election. CROP did not release regional data, so a seat projection is not possible. We can safely assume, however, that the seat breakdown would also be virtually identical to the last election on these numbers.

The game changes dramatically with Couillard at the helm of the Liberals, as he boosts the PLQ to 36%, taking three points from the PQ and two from the CAQ (and one from the magic of rounding). Because of their generally lower numbers among francophones, the Liberals can do less with this amount of support than the PQ but this would likely deliver a slim majority to the PLQ.

Raymond Bachand also improves things for the Liberals, picking up one point to increase the party's support to 31%. The CAQ gains two to hit 29% and the PQ falls three points to 28%. That is a bit of an odd result - why would the PQ lose support to the CAQ if Bachand becomes leader of the PLQ? It could just be a statistical anomaly or respondents changing their mind mid-stream for no particular reason, but it could also be possible that Bachand (a former sovereigntist) pulls a few voters from the Parti Québécois and pushes a few Liberals to the CAQ. That might be reading a bit too much into the numbers, though.

Pierre Moreau is a drag on Liberal fortunes, decreasing their support to 29% (a point made up by Québec Solidaire, of course). Broadly speaking, the Moreau and Bachand scenarios are similar enough to current voting intentions to conclude that neither of them does much to change the political landscape in the province, unlike Couillard.

But is Philippe Couillard just the latest shiny new leader that Quebecers like but will eventually reject? François Legault was supposed to be Premier by now instead of leading the third party in the National Assembly. His novelty quickly wore off once he became a politician again - a situation that could repeat itself if Couillard takes over the party. It would be less likely to occur under Bachand, who is a sitting MNA and, as former finance minister, already well known as a politician.

That is not to say that Couillard is an unknown - he isn't. Out of the three candidates, he had the lowest number of people saying they did not know him at 8%. Bachand was an unknown to 13% while Pierre Moreau was unknown to 47%. One of these things is not like the other.
The numbers are little different among PLQ supporters: 8% don't know Couillard, 11% don't know Bachand, and 44% don't know Moreau.

Worse for Moreau, however, is that those who know him are mostly just familiar with his name. Fully 54% of PLQ supporters said they knew Couillard's history and the ideas he defends, while 43% said the same of Bachand. Only 9% said they knew that much about Moreau, compared to 47% of PLQ supporters who said they know just his name. That number was also high for Couillard (39%) and Bachand (46%), but they are more intimately known by a much larger proportion of the population.

This speaks to how much of a long-shot Moreau is in this race - it really is between Couillard and Bachand. And both appear to be good options for the party, as 65% of all Quebecers have a good opinion of Couillard while 52% have a good opinion of Bachand. Only 36% have a good opinion of Moreau, mostly because 47% don't know who he is. Bachand is a bit more of a divisive candidate, undoubtedly due to his recent political history, as 35% have a bad opinion of him, compared to 27% for Couillard. Among Liberal supporters, Couillard splits 84% good to 8% bad opinion, not unlike Bachand's 77% to 12% ratio.

Couillard is generally seen as the person who embodies change, honesty, and action, while Bachand is the good economic administrator (according to another set of questions). Both perceptions are good for winning elections, but Couillard's is probably what can win him a leadership race.

And that is what people expect to happen. 48% of Quebecers think Couillard will win, compared to only 18% for Bachand and 5% for Moreau. That changes little among Liberals, as 46% think Couillard will win while 23% give Bachand the better odds.

Either way, they don't want an election. Only one-in-three Liberal voters would want the next leader to force the downfall of the Marois government, the majority preferring that the Charbonneau Commission finish its work. That puts us in 2014 for the next election, which seems to be the most likely outcome at this stage.

By then, Couillard or Bachand will have had a year to get comfortable in the job and for Quebecers to get a good idea of what kind of Premier they might be. They generally like what they see right now, but there's no telling what they will think in a year's time.

Tuesday, January 15, 2013

Why the ballot box is a viable route for change for First Nations

With protests, hunger strikes, and the obstruction of transportation links, Canada’s first nations have forced themselves to the top of the political agenda for 2013. While their underlying concerns have deeper roots, the recent target of their ire has been the Conservative government of Stephen Harper. But Canada’s aboriginal peoples have an easier path to force change: voting.

You can read the rest of the article at The Globe and Mail website.

The article spells out why the ballot box is a viable means for Canada's aboriginal peoples to effect change in this country. They do have the size to change outcomes, and if the Idle No More movement turned its attention to electoral politics it could have a huge influence.

If the movement had been active in May 2011 and targeted its efforts to get the vote out, they could have defeated nine Conservative MPs by convincing less than one-in-four aboriginal non-voters to cast their ballot for an NDP or Liberal candidate. More broadly, as mentioned in the article, targeted efforts could have reduced the Conservatives to a minority. More hypothetically, the aboriginal population is large enough to potentially sustain an aboriginal party that could hold the balance of power in a minority parliament. By looking at the results of the last election, it is possible to see what could be possible for the movement going forward.

This was an interesting article to research. Thanks to Elections Canada, who provided me with a list of polling divisions partly or completely located on a First Nations reserve, it was possible to make an estimate of how on-reserve aboriginals voted in the last federal election. The results were not exactly surprising, but the extent to which the New Democrats dominated between Saskatchewan and Quebec was remarkable.

In fact, the national totals mentioned in this article (43% for the NDP, 37% for the Conservatives) actually disguise the size of the NDP lead in many native communities. The Conservatives' on-reserve vote was located primarily in British Columbia, where the on-reserve (or at least those who voted) population is quite high. B.C. voters represented almost half of the entire pool.

Going through the returns, I noticed that the Conservatives were competitive on or near many reserves, but when a reserve voted for the NDP they voted en masse. It was not unusually to see the New Democrats taking 80% or more of the vote in a reserve polling division. As I mention in the article, the NDP took over 90% of the vote in Attawapiskat. That pattern was repeated in many native communities, whereas it was very rare to see this level of support for the Tories in any polling division.

But the analysis I made here is somewhat hamstrung. Elections Canada included any polling division that partly contained a First Nations reserve, making no distinction between polling divisions in which the on-reserve population represented 95% of voters and those in which only a tiny portion of voters were on a reserve. They did distinguish those polling divisions located entirely on reserves, however. But without a very detailed study of each polling division (and there are 1,300+ reserves), it is impossible to know how much the sample was skewed due to it including non-First Nations voters. It was likely not skewed to a huge extent, as many reserves are located some distance from non-aboriginal populations, but I suspect that a not insignificant portion of the Conservative vote tally actually came from non-aboriginals.

The results in Quebec particularly stood out, as the Bloc Québécois managed 19% support. This is a case where much of that Bloc support probably existed in polling divisions that were partly on a reserve and partly not on a reserve, as in some locations the Bloc's vote was virtually non-existant. But discounting the Bloc's vote on reserves in Quebec entirely would be jumping to a false conclusion: there were actually a few polling divisions that were located entirely on reserves in which the Bloc took a normal chunk of the vote. This makes it difficult to assess how much of that 19% is real First Nations support and how much of it was drawn from non-aboriginal Quebecers who merely live near a reserve. Here again, the Bloc's support is likely (but not completely) inflated, pushing the NDP's proportion up even more.

Nevertheless, this analysis captures virtually every First Nations voter who lives on a reserve (but as Elections Canada points out, it is possible that some on-reserve natives were directed to polling booths off of their reserve). Might we expect aboriginals who do not live on a reserve to vote differently? That is something we do not know, but we have some indication that they may vote similarly. As mentioned in the article, an Elections Canada study found similar turnout levels between on- and off-reserve aboriginals.

There are also eight ridings with large aboriginal populations (10,000+) but few reserves: Winnipeg North, Winnipeg Centre, Abitibi-Baie-James-Nunavik-Eeyou, Western Arctic, Vancouver Island North, Prince Albert, Saskatoon-Rosetown-Biggar, and Nunavut. On average, the NDP took 41% of the vote in these eight ridings, with the Tories taking just under 40%. Not dissimilar from the national on-reserve totals. And the NDP managed 43% or more in five of the eight ridings (and averaged 44% if we exclude Nunavut). So it stands to reason that the on-reserve estimate is probably not unrepresentative of off-reserve voting habits as well.

It will be interesting to see what political consequences the recent protests will have. Will aboriginal leaders consider the ballot box as a means for change? Will the federal parties start prioritizing First Nations issues in order to get their support? I suspect that we will be discussing these issues for quite some time to come.

Monday, January 14, 2013

Who has the easier path to OLP victory?

The voting for the Ontario Liberal leadership took place over the weekend, with Sandra Pupatello and Kathleen Wynne emerging as the two frontrunners. Though the counts differ depending on the source, it appears that Sandra Pupatello has emerged with 27% of elected delegates and Kathleen Wynne with 25%. That puts either of them in the best position to win the race to be premier on January 26.

Gerard Kennedy placed third with 14% of the delegates, followed by Harinder Takhar at 13%, Charles Sousa at 11%, and Eric Hoskins with 6%. Another 4% were elected as independents, many of them supporters of Glen Murray.

If the last Ontario Liberal leadership convention is any guide, this could potentially still be anyone's game. More realistically, however, the race is between Pupatello and Wynne. In the 1996 OLP leadership race, Dalton McGuinty did win after placing fourth on the first ballot and Stéphane Dion won the 2006 federal Liberal race after placing third out of the gate. But in both cases, they started out with 18% support and were 11 or 12 points behind the leader.

Kennedy, Takhar, and Sousa are all well below that 18% and Kennedy trails Pupatello by 14 points among the elected delegates. It is a bit of a stretch to imagine a scenario where anyone but Wynne or Pupatello wins, though it is not impossible, as we will see.

One of the factors complicating the calculations are the ex-officio delegates. These are current and former MPPs, party brass, current Ontario MPs, and others. They are not tied to any candidate on the first ballot, like the elected delegates are.

In order to make some estimates on how the voting might go at the convention, it is necessary to distribute the ex-officio delegates which, again depending on the source in this wonderfully opaque process, number between 419 and "about 600". From what I have seen, however, 419 is probably closer to the real number.

Since these delegates are from the party establishment, it makes sense to distribute these delegates according to the endorsements each of the candidates have piled up. If we look only at current MPs and MPPs and former MPPs, we get the following distribution of ex-officio delegates:

Sandra Pupatello - 47%, or about 200
Kathleen Wynne - 30%, or about 125
Gerard Kennedy - 12%, or about 50
Eric Hoskins - 9%, or about 35
Charles Sousa - 3%, or about 10
Harinder Takhar - 0%

But before estimating the first ballot, the independent delegates need to be portioned out. A look at the 1996 OLP and 2006 LPC races suggests that, when an endorsement is made, about 70% of those delegates will go where they are directed. The remaining 30% generally drift to the other candidates evenly. That means that Kathleen Wynne picks up 47 of the independent elected delegates, with the others getting four apiece.

Now that we have established the number of ex-officio and independent delegates each candidate might get, we can make an estimate of what the first ballot could look like:

FIRST BALLOT ESTIMATE

Sandra Pupatello - 708 votes, 31%
Kathleen Wynne - 635 votes, 28%
Gerard Kennedy - 311 votes, 14%
Harinder Takhar - 248 votes, 11%
Charles Sousa - 212 votes, 9%
Eric Hoskins - 143 votes, 6%

Where do we go from here? Let's assume that both Hoskins and Sousa drop out of the race at this point, and that any endorsements will result in 70% of votes going to the endorsed candidate and the rest being distributed evenly. Let's also assume that the vote of the 4th place candidate, in this case Takhar, drops on the second ballot. In the 1996 and 2006 races, support for candidates outside of the top tier dropped on the second ballot, undoubtedly due to their prospects for winning looking so slim.

Sandra Pupatello has the best position, as if she maintains her lead throughout the balloting (like Thomas Mulcair did in the 2012 NDP leadership race), she will win. In other words, if no candidate makes an endorsement, Pupatello wins in this exercise. Momentum could also work in her favour, as the longer she stays in front on the balloting the more likely she is to get more support from ex-officio delegates. But Wynne has some easy paths to victory as well. Let's look at the scenarios.

WYNNE VICTORY #1 - Hoskins, Sousa, Kennedy to Wynne, Takhar to Pupatello

In this first scenario, Torontonians Hoskins and Sousa decide to lend their support to Wynne, resulting in this second ballot (in this and other scenarios, the total number of votes may not remain constant, as is usually the case when trying to corral 2,000+ to vote):

Kathleen Wynne - 884 votes, 40%
Sandra Pupatello - 760 votes, 34%
Gerard Kennedy - 363 votes, 16%
Harinder Takhar - 213 votes, 10%

Takhar goes to Pupatello but she does not take the lead on the third ballot, when Kennedy drops off and lends his support to Wynne. That results in the following final ballot:

Kathleen Wynne - 1,122 votes, 52.9%
Sandra Pupatello - 998 votes, 47.1%

In this case, Wynne's victory is at its largest due to the momentum swinging to her on the second ballot. If something like this occurs, she will likely win.

WYNNE VICTORY #2 - Hoskins and Sousa neutral, Takhar to Pupatello, Kennedy to Wynne

The second scenario sees neither Sousa nor Hoskins lending their support to any of the candidates. In that case, the second ballot looks like this:

Sandra Pupatello - 826 votes, 38%
Kathleen Wynne - 753 votes, 35%
Gerard Kennedy - 429 votes, 20%
Harinder Takhar - 159 votes, 7%

Takhar drops off the ballot and lends his support to Pupatello. If that occurs, the third ballot looks like this:

Sandra Pupatello - 937 votes, 43%
Kathleen Wynne - 777 votes, 36%
Gerard Kennedy - 453 votes, 21%

Now Kennedy drops off the ballot, and decides the race by endorsing Wynne:

Kathleen Wynne - 1,094 votes, 50.5%
Sandra Pupatello - 1,073 votes, 49.5%

Theoretically, Pupatello could win if Kennedy does not endorse anyone but one assumes that a good portion of Kennedy's support would go to Wynne rather than Pupatello. The result, then, is that Kathleen Wynne wins.

WYNNE/PUPATELLO TIE - Hoskins, Takhar to Pupatello, Sousa, Kennedy to Wynne

Another interesting scenario would result in a virtual tie, meaning either candidate would have a good shot at winning. This assumes that, after the first ballot, Hoskins lends his support to Pupatello and Sousa gives his to Wynne.

Sandra Pupatello - 839 votes, 38%
Kathleen Wynne - 803 votes, 36%
Gerard Kennedy - 352 votes, 16%
Harinder Takhar - 217 votes, 10%

At this point, Takhar goes to Pupatello and Kennedy goes to Wynne, resulting in this final ballot:

Sandra Pupatello - 1,106 votes, 50.1%
Kathleen Wynne - 1,104 votes, 49.9%

There are more than enough assumptions made in this analysis for this scenario to be way too close to call.

PUPATELLO VICTORY #1 - Sousa, Takhar to Pupatello, Hoskins, Kennedy to Wynne

If Pupatello gets Sousa and Takhar to her tent, she has a much better chance of winning. Here is how the second ballot plays out if Sousa goes to Pupatello and Hoskins to Wynne:

Sandra Pupatello - 876 votes, 40%
Kathleen Wynne - 766 votes, 35%
Gerard Kennedy - 352 votes, 16%
Harinder Takhar - 217 votes, 10%

With her larger lead, the support that Takhar gives her makes the difference. Kennedy does not have enough support to give Wynne the victory:

Sandra Pupatello - 1,143 votes, 51.7%
Kathleen Wynne - 1,067 votes, 48.3%

It is still rather close, and again the number of assumptions being made gives this entire exercise a rather large margin of error.

PUPATELLO VICTORY #2 - Sousa neutral, Hoskins, Takhar to Pupatello, Kennedy to Wynne

If Sousa stays neutral after the first ballot but Hoskins goes over to Pupatello, then the second ballot looks like:

Sandra Pupatello - 882 votes, 40%
Kathleen Wynne - 719 votes, 33%
Gerard Kennedy - 395 votes, 18%
Harinder Takhar - 185 votes, 8%

Here again, the lead that Pupatello holds makes Takhar's support the deciding factor.

Sandra Pupatello - 1,138 votes, 52.2%
Kathleen Wynne - 1,043 votes, 47.8%

PUPATELLO VICTORY #3 - Hoskins neutral, Sousa, Takhar to Pupatello, Kennedy to Wynne

Pupatello's victory gets larger when Hoskins stays neutral and Sousa goes to her camp. In that case, the final ballot would be:

Sandra Pupatello - 1,167 votes, 53.2%
Kathleen Wynne - 1,028 votes, 46.8%

And, of course, her victory gets even larger if she gets Hoskins, Sousa, and Takhar on her side throughout the balloting. At any point, Pupatello's chances of victory get slimmer if Takhar stays neutral but Kennedy sides with Wynne. If Kennedy stays neutral, then Wynne is in a difficult position. It makes horse-trading absolutely necessary.

But, within the bounds of this theoretical exercise, other candidates could win as well.

SOUSA VICTORY - Everyone to Sousa

Charles Sousa does have a path to victory, but it is not an easy one. He would need every dropped-off candidate to endorse him, and deliver their delegates, at each ballot. Assuming Takhar also picks up support on the second ballot, and with Hoskins's support going to Sousa, that second ballot would be:

Sandra Pupatello - 718 votes, 32%
Kathleen Wynne - 645 votes, 29%
Gerard Kennedy - 321 votes, 14%
Charles Sousa - 312 votes, 14%
Harinder Takhar - 258 votes, 11%

For the third ballot, Takhar would need to send his support Sousa's way in order for him to overtake Kennedy:

Sandra Pupatello - 744 votes, 33%
Kathleen Wynne - 671 votes, 30%
Charles Sousa - 493 votes, 22%
Gerard Kennedy - 347 votes, 15%

At this point, Kennedy would also have to endorse Sousa:

Sandra Pupatello - 796 votes, 35%
Charles Sousa - 735 votes, 33%
Kathleen Wynne - 723 votes, 32%

But you can see how fraught this path to victory is for Sousa. He ends up ahead of Wynne by 12 votes, a margin that is very small considering the assumptions being made. At this point, though, he could easily win with Wynne's support:

Charles Sousa - 1,241 votes, 54.8%
Sandra Pupatello - 1,025 votes, 45.2%

What are the odds, though, that the candidate one up from the bottom would receive an endorsement at each step of the way? And would the ex-officio delegates perhaps not move strongly way from Sousa  - too strongly for Sousa to over-take Wynne on the third ballot?

TAKHAR  VICTORY - Everyone to Takhar

Takhar's path is similarly unlikely, especially considering that he has not a single endorsement to his name at this stage of the race. After the first ballot, he could overtake Kennedy with just Hoskins's support, with Sousa then dropping off the second ballot and lending his support to Takhar as well:

Sandra Pupatello - 743 votes, 33%
Kathleen Wynne - 670 votes, 30%
Harinder Takhar - 496 votes, 22%
Gerard Kennedy - 346 votes, 15%

And then Kennedy would need to go Takhar's way, which is far less imaginable than Kennedy choosing Sousa over Wynne:

Sandra Pupatello - 795 votes, 35%
Harinder Takhar - 738 votes, 33%
Kathleen Wynne - 722 votes, 32%

And now we need to assume that the ex-officios do not go en masse to Pupatello to prevent Takhar from winning (recall, she currently leads endorsements) and that Wynne prefers Takhar to Pupatello, in order to get:

Harinder Takhar - 1,243 votes, 55.1%
Sandra Pupatello - 1,012 votes, 44.9%

This is the scenario that is hardest to envision, as Takhar is not seen as anyone's consensus second choice. Sousa could conceivably fill that role, but again we're stretching the imagination.

KENNEDY VICTORY - Everyone to Kennedy

But what about the candidate that Ontarians would choose if they were voting? It is not too much of a stretch to imagine the losing candidates deciding to go with the popular choice, at least in terms of name recognition. After all, the next election might be weeks away - no time for an introduction to Pupatello or Wynne.

Sandra Pupatello - 749 votes, 34%
Kathleen Wynne - 676 votes, 31%
Gerard Kennedy - 569 votes, 26%
Harinder Takhar - 217 votes, 10%

Joining Hoskins and Sousa, Takhar decides to go against the establishment and sends his supporters to Kennedy for the third ballot:

Sandra Pupatello - 781 votes, 35%
Gerard Kennedy - 721 votes, 33%
Kathleen Wynne - 708 votes, 32%

At this point, a path to Kennedy's victory is easy to see. Having edged out Wynne, she could lend her support directly to Kennedy and give him a big victory:

Gerard Kennedy - 1,216 votes, 55%
Sandra Pupatello - 994 votes, 45%

But if Wynne decided to stay neutral, Kennedy would only need to carry a little more than 54% of her supporters in order to beat out Pupatello. That is not difficult to imagine, as Kennedy and Wynne share more of a constituency than Wynne and Pupatello.

This means that the leadership is very much up for grabs, and that the machinations on the convention floor will be hugely important. Kathleen Wynne and Sandra Pupatello have the easiest paths to victory. Gerard Kennedy and Charles Sousa also have a not-implausible way to win it, while Harinder Takhar could mathematically emerge as a victor as well.

Much will depend on where the ex-officio delegates go - if they plump for Pupatello by a larger margin on the first ballot, it will be hers. If they instead go to Wynne, she will be in the much better position. Couple that with the potential for candidates to lose control of their delegates through the balloting process, and the race really is anyone's to win. But with their large lead in elected delegates and advantage among the party establishment, the safest money is on Pupatello or Wynne.

Friday, January 11, 2013

Liberal leadership endorsement rankings

With the deadline for becoming an official candidate about to pass on Sunday, the time has come to launch the endorsement rankings for the 2013 Liberal leadership race. With 87.3% of endorsement points, Justin Trudeau kicks off the race in a dominating position.

The page linked to above (and accessible by clicking on the 2013 LPC leadership graphic in the right-hand column) gives a detailed description of the endorsement rankings system, and a little bit of the rationale behind it. It also lists the endorsements that have been awarded so far.

I think that the endorsement rankings proved their worth in the races for the Bloc Québécois and NDP leadership in 2011 and 2012. The Bloc race was harder to track due to its low profile, but the rankings did pick Daniel Paillé to win it.

The NDP race was far more interesting, with endorsements coming from all levels of the party throughout the country. By the day of the convention, Thomas Mulcair was leading the rankings and was followed closely by Brian Topp. In the end, Mulcair beat Topp on the fourth ballot - an outcome that was projected during the interminable leadership vote. The endorsement rankings did quite well in assessing who the frontrunners were, with the notable exception of Nathan Cullen.

That was an interesting result, however. Of the candidates, Cullen was the most populist and least 'establishment', at least in terms of his proposal to co-operate with the Liberals and Greens. His campaign resonated with the rank-and-file membership, and he vaulted to third place ahead of Peggy Nash and Paul Dewar, who received far more establishment support. Cullen received few endorsements, especially outside of his native British Columbia, and so the rankings could not gauge him correctly.

Or did it? If the endorsement tracker is about gauging establishment support, with the assumption that establishment support will generally align with membership support, perhaps it was on the money. By under-estimating Cullen's first ballot support it gave a stark demonstration of how much of a membership favourite he was, and how much he was going against the grain of the party establishment.

Or perhaps Cullen's campaign was simply an example of one that cannot be tracked by something like the endorsement points system. The endorsement rankings are not a prediction of final outcomes, but rather one way to quantify the race and a method that will usually reflect how the race is going and where the candidates stand. While it missed out on Cullen, for example, it nevertheless pegged Mulcair as the frontrunner, Topp as the runner-up, Nash ahead of Dewar, and Niki Ashton and Martin Singh to be the bottom two. Not a bad result, and during the course of the convention the endorsement rankings only gave Mulcair the win on the fourth ballot.

At this stage of the Liberal race, however, the endorsement rankings suggest that it won't even go to a second ballot. Justin Trudeau has the lion's share (and a big lion it is) of endorsements so far, with the support of one-third of the Liberal caucus.

Trudeau starts the race with a commanding lead of 87.3% of endorsement points, followed at a distance by Marc Garneau at 4.9% and Joyce Murray at 4.4%. Martha Hall Findlay has 1.2% of endorsement points, tied with George Takach. David Bertschi rounds out the top six with 0.9%. Neither Deborah Coyne or Karen McCrimmon have a single endorsement to their name yet.

What to make of this? No one doubts that Trudeau has a wide lead over his rivals, and his impressive fundraising gives his campaign more than its fair share of credibility. I suspect, however, that if the vote were held today Garneau and Hall Findlay would take a much larger share of the vote. Perhaps that will be reflected as the race goes on and the two candidates start to get some endorsements. There's a very good chance that some of these candidates already have a few endorsements in the bag and are waiting for the right time to reveal them (say, after the first debate).

It should also be pointed out that Garneau, Murray, and Hall Findlay get a large portion of their points (and Hall Findlay all of them) from their own "self-endorsement". This is meant as a reflection of the pull they have within their party all on their own. Murray and Garneau are in their second term as MPs, and Murray was a provincial cabinet minister. Hall Findlay is a former MP (and leadership candidate). That is worth something in the endorsement rankings, as well as in the real world.

But the Liberal leadership race is not using a simple one-member-one-vote system. Instead, each riding will be given a value of 100 points and those points will be allocated according to how the members and supporters vote in each riding. So, for example, the vote of 200 members in a downtown Toronto riding will carry the same weight as the eight members in a riding in rural Alberta.

I will be doing the same thing for the endorsement rankings, except I will be breaking down the endorsements by province rather than riding. Here is how that breaks down (readers dating back to the NDP leadership race will recognize the graphics, which are based on CNN's presidential primary graphics. I used them in a post at the time, and liked them too much to not use them again).
Provinces without any endorsers yet (Alberta and Saskatchewan, as well as the territories) are allocated zero points at this stage, so the provincially weighted total is based on 26,300 points, rather than the full 30,800.

Broken down this way, Trudeau still leads by a wide margin. Even without the points from Alberta and Saskatchewan and the territories included, he has more than the 15,400 that will eventually be needed to win. He currently holds 71.5% of the total share of available points, with Murray trailing at 12.6%.

She gets bumped up from third to second due to her (self) support in British Columbia, a province worth 3,600 points. Garneau increases his share to 9.7% due to his (self) support in Quebec. Hall Findlay and Takach follow with 2.2% apiece, while Bertschi has 1.7%.

Hopefully the endorsement race will get a lot closer than this as the campaign unfolds. While I have little doubt that Trudeau would win if the vote was held right now, I am somewhat more doubtful that he would win with more than 70% on the first ballot. I suspect that it might go to a second or even third ballot at this stage. The endorsement tracker, then, is currently reflecting how much of an early juggernaut his campaign really is. The other candidates will need to gain some traction - and fast - in the first few debates if this is to become a real race.

(Note to readers: I based the list of endorsements on the Wikipedia page for the campaign, paying close attention to citations. Now that I have begun the tracker, I will be watching for endorsements myself. If there are any errors in my initial list of endorsements, please let me know.)