Wednesday, 17 July 2019

Ipsos/Global News: CPC 37, LPC 31, NDP 18, GPC 7, BQ 5

A new Ipsos/Global News poll has been published this morning, and it shows that voting intentions in Canada have remained stable in the last month - contrary to the general trend detected by other Canadian pollsters in the same time period.

The full tables have not yet been released (I expect they will in the next day or so), but here is the article from Global News. (I must warn you: this article is a somewhat confusing. It goes from party standings to leader's approval numbers back and forth and throws a bunch or regional and demographic numbers in what looks like random order).

Here are the results at the national level:



Every party is stable compared to Ipsos' previous poll back in June. The main difference is that the Greens and the PPC were included this time (which explains the +7 for the Greens).

While it is not surprising to see Ipsos give the Conservatives the lead (CPC has led every Ipsos poll since February, even by a ten-point margin back in March), it is strange that Ipsos has not seen the gap narrowed one bit like other pollsters.

The most header-scratching data however is that fact that Ipsos has the Conservatives leading among women. This is highly inconsistent - even Angus Reid latest poll, which has similar overall numbers to Ipsos - has the Liberals leading among women.

Nevertheless, the regional numbers that were presented in the article are all within current ranges, with the notable exception of Ontario - where Ipsos has the CPC ahead by 6 points.

Still, as I wrote yesterday: "We have to look at the movement of the tide, not just the water level... and not overreact."

We add this one to the pile and carry on.




Philippe J. Fournier is the creator of Qc125 and 338Canada. He teaches physics and astronomy at Cégep de Saint-Laurent in Montreal. For information or media request, please write to info@Qc125.com.


Philippe J. Fournier est le créateur de Qc125 et 338Canada. Il est professeur de physique et d'astronomie au Cégep de Saint-Laurent à Montréal. Pour toute information ou pour une demande d'entrevue médiatique, écrivez à info@Qc125.com.

Tuesday, 16 July 2019

Look at the Tide Movement, Not Just the Water Level

At the end of the latest 338 federal projection published on Sunday, I wrote: "As we enter the middle of summer, Canadian pollsters will probably take a much-needed break before the fall campaign".

This may have been just a little bit of wishful thinking on my part.

Five new federal polls, yes five, were published in the last 24 hours from five different polling firms: Angus Reid, Nanos, Campaign Research, Innovative Research, and EKOS.

The full tables for Innovative and EKOS have not been made public just yet, so we will hold on to these two for now, but let's take a quick look at the other three.


(All federal polls are listed on this page.)


Nanos Research



The Nanos weekly tracker has been updated and it doubles down on the Liberal-comeback narrative. The LPC leads the way with 36% of voting intentions, up six points from the middle of June. The Conservatives, which Nanos measured as high as 36% last May, lose four points in the past four weeks and stand at 30% nationally.



The New Democratic Party climbs again this week according to Nanos and now stands at 19% - which is somewhat higher than the current polling average for Jagmeet Singh's party.



Angus Reid Institute


The Angus Reid Institute had the Conservatives with a 13-point lead in May, and an 11-point lead in June. Its latest poll still has the Conservatives ahead, but the gap has been closing to 8 points. The CPC leads with 38% and the Liberals come second with 30%. (See Angus Reid report here.)





Campaign Research


Finally, Campaign Research has a statistical tie between the Liberals and Conservatives with 33% and 32% of national support, respectively. The most notable movement in this poll compared to Campaign's previous poll in June is the Conservatives shedding three points.




What are we to make of all there numbers? Whereas Campaign Research's numbers align closely with the current 338 weighted average, Nanos' and Angus Reid's numbers diverge significantly.

Taken alone, Nanos' numbers would probably mean a strong minority or even perhaps a majority for the Liberals.

Taken alone, Angus Reid's would assuredly lead to the Conservative winning the most seats, although perhaps not enough for a majority.

However, if we look at the movement detected by these polls and not just the raw numbers, we all have the same tendency: a tightening race.

I published in May a brief meta-analysis of Canadian polls from the past three years, as well as those just before the 2015 federal election (which you can find here). While Nanos tends to very modestly lean liberal compared to the overall polling average, it was near perfect in its final 2015 poll. On the other hand, Angus Reid has had the Liberals much lower than other pollsters on average in the past three years, and had also underestimated the Liberals by almost five points in 2015. I do not, and I repeat, do not dismiss Angus Reid, but every polling firm occasionally displays a "House Effect"*. Angus Reid is no exception.

*Displaying a house effect does not mean the skew is intentional. Whatever the narrative you want to believe: no, it's not a conspiracy.

Again, remember: we have to look at the movement of the tide, not just the water level.

And not give too much weight to a single poll.

And not overreact.

* * *

All these numbers will be added to the model for this Sunday's 338 update. Have a great week, and thank you for supporting 338Canada.



Philippe J. Fournier is the creator of Qc125 and 338Canada. He teaches physics and astronomy at Cégep de Saint-Laurent in Montreal. For information or media request, please write to info@Qc125.com.


Philippe J. Fournier est le créateur de Qc125 et 338Canada. Il est professeur de physique et d'astronomie au Cégep de Saint-Laurent à Montréal. Pour toute information ou pour une demande d'entrevue médiatique, écrivez à info@Qc125.com.