Are the Liberals on the rise?

Week two and we finally, finally have our first true, complete poll. This is from Ipsos and you can find the details at the end of this article. As a reminder, until now, we've had one Forum poll done on a Sunday between 10am and 4pm, another Forum for the GTA-only done on a Friday during the same hours, and one Mainstreet poll for BC only. We also got a Nanos, but they do a weird 4-week average which makes the numbers pretty much useless. So it's nice to finally have a Canada-wide poll, done over multiple days. The fact we have less polls during this campaign than we did before the writs were dropped is puzzling, but let's move on.

The new Ipsos numbers show a close three-way race with a slight edge for NDP, at least in terms of votes. At 33%, the NDP has a 2 points lead over the incumbent Conservatives, and a 5 points gap ahead of the Liberals.While it might not look like it at first sight, Justin Trudeau and his party are actually the ones making gains. Indeed, compared to the most recent Ipsos poll, they are up 3 points while the NDP and Tories are both down. Given the sample sizes, this is actually a significant change.

Moreover, this poll seems to confirm a small trend observed before where the Liberals appear to be slowly rising. It's way too early to tell, but here are the evidence so far. The above mentioned GTA-only poll had the Liberals slightly higher than expected. Sure, this poll had serious limitations, but we can't completely discard it. Using the Mainstreet numbers, we get the Liberals at 27% in BC (among people who watched the debate though), higher than the previous poll average and 8 points higher than in the previous poll from the same firm. On top of this, we now have had three polls with information regarding the Maclean's debate and the performance of each leader. In all three, Justin Trudeau is seen as the winner by most. The table below shows my usual "debate index" where I divide the % of people picking one candidate as the winner by the % of people who want to vote for this party. Therefore, a score above 1 means the leader managed to convince people outside of his party support while a score below 1 indicates the opposite. The idea of the index is to remove the bias caused by the fact that a lot of voters will see "their" leader as having done the best (i.e: trying to remove some subjectivity). It's also there to see which leader might have done the most in order to convince other voters to switch party.

Forum (GTA only)
Mainstreet (BC only)
Ipsos (Canada wide)
Stephen Harper / CPC
Justin Trudeau / LPC
Thomas Mulcair / NDP
Elizabeth May / Green

As you can see, Elizabeth May is clearly the "winner" using this measure. She's by far the leader who managed to reach outside of her party base. Of course, this index is usually quite favorable to small parties as they can't really do worse (plus, let's face it, a lot of people might judge the leader of a small party differently. For instance because she doesn't "have" to look like a Prime Minister). Nevertheless, May can be very satisfied with her performance.

But among the main three parties, Trudeau is the closest to 1 while Mulcair is the furthest in average. We aren't talking about a debate performance with a knockout or anything, but Trudeau seems to have done at least relatively better than his two direct opponents. Even if you don't like my index, Trudeau was chosen as having had the best performance by the most voters in all three polls.

What I'm trying to say here is that so far, there hasn't really been any bad news for the Liberals. While the polls were not Trudeau's friends this summer, the Liberal leader seems to have at least stopped the decline. The only negative was that the LPC was technically down this week in the Nanos poll, but the sample size is so small that it's hard to really take anything from it.

Of course, being (slightly) up is good, but the Liberals were starting third with virtually no chance of winning the most seats. So they had a long way to go. Let's see the projections we would get using the latest Ipsos numbers (given the lack of polls, we might as well use this survey only. No need for a poll tracker when there aren't any poll).

Voting intentions; Seat projections with 95% confidence intervals; Chances to win the most seats.

First thing first, yes, the Conservatives have more chances to win the most seats despite being projected with less wins. This can happen because the distributions of seats are not symmetric. In other words, the average of the projections is not the same as the projections of the average voting intentions. If you prefer, just interpret this as if the polls are slightly wrong (like underestimating the CPC), then the total number of seats of this party could increase quickly. On the other hand, the NDP wouldn't gain as much of an small underestimation. Also, and this is important, in case of a tie between the two parties, the model gives the win to the incumbent Stephen Harper (as is the rule of our system, at least in theory). If we look into the 5000 simulations, we actually find 2662 cases where the Conservatives win the most seats, the exact same number of scenarios where it's the NDP, and 171 cases where it's a tie.
At the end of the day though, don't read too much into this. The situation with this poll is that it's a close race between Tories and New Democrats with the Liberals far behind.

Also, you may have noticed the percentages for the voting intentions are slightly different from the published Ipsos poll. This is because I allocate undecided differently. I give more of them to the CPC because the incumbent is often underestimated (and small parties receive none). It's not a major change but it's enough to make a small difference. If you don't think we should do that and you trust the polls (or you really distrust them), remember you can always use your own numbers in the simulator.

But speaking of the Grits, they are at least back to a situation where it's possible for them to win. That wasn't the case at the beginning of this campaign. It's still a long shot, but it's better than nothing. It does show that Trudeau and his party are having a relatively good start of a campaign. Had it been otherwise (as well as win by Mulcair at the debate), the Liberals were at risk of becoming irrelevant very early on in this election.

In conclusion, nothing major has happened so far in this campaign. But if we had to pick one party/leader with a better start, it seems that would be Justin Trudeau and the Liberals. We are far from a come back or surprise win, but the slow and continuous decline observed in the recent months seems to have been stopped and even inverted. We'll hopefully get more polls soon to see if there is really a trend.

Details about the poll:
Ipsos poll for Global News. Conducted between August 7th and 10th among 2022 respondents from the Ipsos online sample. Margins of error for a probabilistic sample of this size would be 2.03% 19 times out of 20 for the NDP and less for the other parties. The question asked was: "Thinking of how you feel right now, if a federal election were held tomorrow, which of the following parties' candidates would you, yourself, be more likely to support?". More details available here.