Skip to main content

NDP Leader Tom Mulcair poses for a photo with street hockey players after addressing the media at a morning announcement during a federal election campaign stop in Kamloops, B.C. Wednesday, Sept, 2 2015.JONATHAN HAYWARD/The Canadian Press

UPDATE Oct. 18: The final day of the campaign sees the Liberals with the largest chance of victory, given the state of the polls and the distribution of votes around the country, though the Conservatives still have a one-in-five chance of ending up with the most seats on election night.

The NDP also appear to have little chance  winning the most seats. Over the past few days, the Forecast has given the party less than a 1 per cent chance of doing so. Today, they led in zero of the 1,000 simulations. This doesn't mean that there is literally no chance of them winning the seat count, but that their chances are very low indeed. This is based on their poor polling performance in recent days. Since October 10th, no poll has had the NDP closer than 10 percentage points from the lead. If they do end up with the most seats, it would require either an unusual and optimal distribution of votes around the country, or a polling failure larger than has ever been seen.

Oct. 14: The Globe Forecast's projection of an increased chance of the Liberals winning the most seats is due to some strong polling numbers posted by the party over the last few days. Three different pollsters have recently released polls showing the Liberals with a lead of between 4.5 and 6.9 per cent over the Conservatives. The Liberals have also consistently polled very strongly in Ontario, where a number of polling firms are now showing them having a more than 10 percentage point lead.

Oct. 12: The final week of the campaign sees a significant tightening of the chances by the Liberals and Conservatives of winning the most seats. This is primarily the result of the consistently strong polling numbers posted by the Liberals in surveys with end dates of October 6th, or later. It is their strongest polling period of the campaign.

Methodological note: This last seven days also represents the last change to the model's parameters, now only including a weighted average of the last week of polling, instead of the last three weeks. (This was planned from the beginning, when the model first included six weeks of numbers.) In either the one or three week averages, the chances of a Liberal or Conservative victory are essentially the same.

Oct. 5: Polls released in the last week seem to be in general agreement on the slide by the NDP, driven largely by a decline in their vote share in Quebec, which is what the new Globe Election Forecast reflects. However, the polls have disagreed about what else is happening in the national race.

Polls by both Forum and Angus Reid showed the Conservatives leading by a reasonably healthy six-to-seven-percentage-point margin, with the NDP and Liberals tied for second. If these polls ended up being the final result, the Conservatives would likely win a strong minority government, with the two remaining parties battling for Official Opposition status.

Just as this narrative was emerging, three pollsters have, in the last few days, put the Liberals in first by a small margin. Léger and Innovative Research had the Liberals ahead by two percentage points, while the latest Nanos three-day rolling poll puts the Liberals ahead of the Conservatives by five, and a full 13 points ahead of the NDP.

These last few polls might be picking up the beginnings of a Liberal surge, but in the context of the last week of polls, it's still not quite clear enough. If there is a real increase in the Liberal vote, they will surely be able to repeat their performance over the next few days of polling. If this is the case, the forecast will accordingly become more favourable to them. If, instead, the mixed messages of the last week are repeated, the Forecast should remain steady.

Sept. 25: The Conservatives have re-taken the lead in the Globe Election Forecast for two reasons. First, a strong performance in an EKOS poll gave the party 35 per cent of the national vote, compared to just 26 per cent for the Liberals and 25 for the NDP. If these results are repeated by other firms in the next few days, this is very good news for the government; if the poll is an outlier, this will become clear in short order, and its effect in the Forecast will wash out. Second, the NDP have polled somewhat weaker than usual in Quebec, most notably in the most recent Léger poll. While still in first, and down compared to earlier polls in the range of eight percentage points, losing grip on even 10 seats in a three-way race reduces any party's chances of winning the most seats.

Sept. 22: Public opinion data has been streaming in since the federal leaders' debate last Thursday, and all evidence suggests that voters remain as divided as ever.

Winning a debate isn't the same thing as winning an election. A better measure of who won can be seen by looking at who moved the most votes. Here, too, signals are mixed. While the Nanos 3-day tracking poll showed its usual three-way race, Ipsos had the Liberals taking a small lead. The last time they had the Liberals in first was back in late May when they were tied at 31 per cent with the Conservatives. Similarly, the Liberals continue their gradual improvement in the Globe Election Forecast.

Sept. 14: As the polls draw even to a three-way split in the popular vote, so do the odds of each party winning the most seats. While the NDP and Conservatives remain ahead, the Liberals continue to improve their chances of winning the largest parliamentary caucus primarily as a result of their recent strong polling performances in Ontario.

Sept. 8: The close three-way race in the federal election has become even tighter in the last week. A diminished Conservative vote coupled with growing Liberal support now gives all three parties with a realistic shot of winning the most seats in October. A consequence of this three-way race is seen in the Election Forecast's estimate of the likelihood of a majority government: just 2.2 per cent.

Sept. 2: The Globe's forecast now predicts that the NDP are the most likely party to win the largest number of seats, with the party leading in 53 per cent of the simulations. This follows a string of seven consecutive national polls each showing a lead of between 1 and 10 percentage points for the New Democrats.

The seven poll lead was reported by seven different pollsters, using three different methods: traditional telephone, interactive voice response (IVR) and online surveys. The New Democrats have only had such a string of good polling on two separate occasions during this parliament: earlier this year in June, and in the May-June period of 2012.

In good news for the Liberals, three recent polls, by Nanos, Ipsos Reid and Forum, have showed the party in second place, ahead of the Conservatives. Furthermore, polls consistently suggest the gap between first and third place is under 5 percentage points.

This all reinforces how unusual this election is: the best a third-place party has ever done in terms of vote share was in 1988, when the Ed Broadbent-led NDP won 20.4 per cent of the vote. Currently, we're in a situation where whatever party is polling in third is earning 25 per cent popular support.

Paul Fairie is a University of Calgary political scientist who studies voter behaviour, who designed The Globe's Election Forecast.

Interact with The Globe