Thursday, August 1, 2013

Ontario Liberals under pressure in all five by-elections

By-elections are being held in five ridings across Ontario today, from Lake St. Clair in the southwest to the Ottawa River in the east. All five ridings belong to the Ontario Liberals, and all five are at risk of being won by another party by the time the votes are counted tonight.

The By-Election Barometer considers one of the ridings, Scarborough-Guildwood, most likely to remain a Liberal hold, while Windsor-Tecumseh is almost certain to go with the New Democrats. The ridings of Etobicoke-Lakeshore and Ottawa South could go to either the Liberals or the Progressive Conservatives, while London West is considered a three-way toss-up.

But the By-Election Barometer was designed to give people an idea of what to expect in by-elections where little specific information is available. We rarely see by-election polls, so there is never much to go on. But in these five by-elections, we've been treated to a wealth of polling data (25 polls in all). Considering that these polls represent better information than the Barometer itself is supposed to represent, we must consider them much more convincing evidence.

That being the case, let's look at each of the five ridings at stake tonight and what the polls (and the Barometer, which does take them into account) are saying.

(Campaign Research's final polls can be seen here. The final polls from Forum are currently on the first page of their news archive - if you are looking at this post after the by-elections have taken place they will have moved down the list at Forum's site. These polls were all released and made public yesterday, before the new-poll blackout imposed by Elections Ontario.)

We'll start with the riding most likely to remain in incumbent hands: Scarborough-Guildwood. The By-Election Barometer considers it a Likely Liberal win, estimating that Mitzie Hunter has a 75% chance of taking it, against 25% for Ken Kirupa of the Progressive Conservatives.

The polls appear to agree with this assessment. In the five polls done in the riding since the beginning of July, Hunter has managed between 37% and 40% support - a very tight grouping. Her weighted average (which employs the same methodology this site uses during elections) gives her 37.9% of the vote, compared to 31.4% for Kirupa. His support has also been tightly grouped, at between 30% and 32% in the last four polls.

Adam Giambrone of the New Democrats is averaging 25% support, but the latest Forum poll suggests he could have some momentum. The Greens have averaged 4.4%, while support for other parties and candidates has averaged 1.3%.

Applying ThreeHundredEight's turnout model to the Forum poll, however, paints a very different picture. Support for the PCs, Greens, and others is hardly changed, but Hunter drops to 32% and Giambrone increases to 32%, putting the three parties in a three-way tie. Applying the turnout model has improved all of Forum's election-eve polls, so if that trend continues we should expect Giambrone to perform better than the 27% he was awarded in Forum's last poll. (Note: the turnout model cannot be applied to Campaign Research's polls, as they do not release support breakdowns by age.)

The riding in which the Liberals have the next best chance of retaining is Etobicoke-Lakeshore, which the By-Election Barometer considers an OLP/PC Toss-up. The Liberals are given a 56% chance of holding on to it, against 44% for the Progressive Conservatives. But here again, we must consider the riding polls as more revealing information.

They reveal that the race has tightened in the final days of the campaign. Peter Milczyn of the Liberals had fallen behind in mid-July, with Campaign Research putting the deficit at five points and Forum putting it at seven points. Both of their most recent polls show that Doug Holyday's support has remained unchanged, while Milczyn's has increased by a few points, narrowing the gap to four points in Forum's estimation and giving Milczyn the edge by one according to Campaign. Holyday is still ahead in the weighted average, however, with 45.7% to Milczyn's 42.2% and P.C. Choo's 8% for the NDP. The Greens are averaging 3.6% while support for other parties has averaged 0.5%.

This is one riding where the calls made by Forum and Campaign differ. I will assess how the polls (and this site) did in the by-elections tomorrow, and this is one of the ridings that will put the two firms to the test.

The turnout model suggests Holyday has it, improving his vote haul to 49%. Milczyn's also improves to 44%, while the NDP and Greens drop to 4% and 2%, respectively.

The Ontario Liberals' hope of winning a third riding starts to decrease significantly as we turn to Ottawa South, formerly the riding of Dalton McGuinty. The By-Election Barometer does give the Liberals a 55% chance to the Tories' 45%, but the riding-level polling suggests that this is far too kind to Liberal candidate John Fraser.

Fraser has not led in a poll since the beginning of July, when he was ahead by four points. In every poll since the deficit has increased. Campaign put it at one point in mid-July, but now reckons it to be seven points. Forum put it at 14 points on July 24, but now sees it as a gap of 16 points in favour of Matt Young.

Young is averaging 48.7% in the polls, compared to 35.4% for Fraser, 10.6% for Bronwyn Funiciello, 4.1% for the Greens, and 1.3% for the other candidates. If Fraser manages to win the riding - and it is still generally considered a toss-up, no matter what the polls say - he will have proven the polls wrong. Either Campaign or Forum will miss the call to some extent, however, because the gap between their numbers is so large. This will be the riding that gives either one of these firms the 'win'.

The turnout model suggests Young's advantage is under-stated in the polls, as his support is increased to 55% against 34% for Fraser. In other words, the turnout model believes that if Forum is going to miss the call, it will be because they didn't give Young enough of a lead. That seems hard to swallow.

From here, we move to ridings where the Liberals are not realistically in the running to hold on to their seats. The By-Election Barometer still considers the OLP a factor in London West, however, due to the party's history in the riding and their decent polling numbers in the southwestern part of the province. It gives them a 39% chance of winning, compared to 35% for the Tories and 27% for the NDP, making it an OLP/PC/NDP Toss-up. That seems far too optimistic for the Liberals, who have been tanking in the polls.

Liberal candidate Ken Coran hasn't been a factor in the polls since he took up the Liberal banner. The poll from February giving the OLP 30% was before his candidacy was known. He has been dropping during the race, and has averaged 16.5%.

The real contest is between PC candidate Ali Chahbar and NDP candidate Peggy Sattler. Chahbar's numbers have been pretty steady in the mid-30s, while Sattler's were consistently around 30%. But the last Campaign poll put the gap at three points, and Forum put it at only two points. The weighed average gives Chahbar 36.6% to 33.5% for Sattler. That makes it the closest race of the night. Support for other parties and independents (Al Gretzky, primarily) has averaged 7.1%, while the Greens have been at 6.4%.

The turnout model gives Chahbar a slightly wider lead than the Forum poll suggests, with his support boosted to 40% against 37% for Sattler. But that is still very close. It seems London West will be the race to watch.

The race that will not be the one to watch is Windsor-Tecumseh, which everyone concedes will be won handily by the New Democrats. The By-Election Barometer considers the riding the only Strong one on offer tonight, giving the NDP a 95% chance of winning (the OLP gets the remaining 5%, based on their history and nothing else).

Percy Hatfield of the NDP has had majority support in every poll since his candidacy was made official, and he's averaging 52.7% going into tonight's vote. The PCs have placed second in the last three polls, and look set to take 24.5% of the vote. The Liberals are well behind, averaging 13.9%. The Greens and others are averaging 5.8% and 3.2%, respectively.

The turnout model suggests both Hatfield and Robert de Verteuil could take a higher proportion of the vote, boosting Hatfield to 54% and de Verteuil to 30%. But it seems that the NDP should have no trouble taking the riding.

So the Liberals will be under a lot of strain tonight. They have a good shot (but by no means a certain one) of retaining one of the ridings, and are in the running in at least one more, and possibly two more. It means they are expected to win one riding, but could win three or none. The Progressive Conservatives are set to be tonight's big winners as they are well positioned to win as many as three ridings, but none of these are slam dunks. They could potentially be shutout if the Liberals have a good night. The New Democrats will win Windsor-Tecumseh and put up good numbers in London West and Scarborough-Guildwood, so it seems they will be under the least strain as they watch the votes come in, since they can safely assume they will have a decent night.

But will it be a good night for Forum Research and Campaign Research? They have put their necks out on this one, with polls in the final days of the campaign. They should be able to call the race correctly, or at least ballpark it considering the unpredictable turnout levels in a mid-summer campaign. Tomorrow, I'll take a good look at how they performed.