Push polling in Canada too

Allow me to preface this with an explainer.  

A lot of people think of push polling as a poll with a result that is designed to influence public opinion, which is technically not correct but still substantially correct. In reality, a push poll is an ostensible opinion poll in which the true objective is to sway voters using loaded or manipulative questions. In other words, it uses misleading questions to sway voter opinions. The misleading part is they skew the poll results by framing the question in a way that manipulates those polled. Those being polled may themselves change opinion based on the framing of the question.  Inevitably the poll result is skewed and publishing it further skews public opinion (or attempts to do so).

All that said, ultimately the public sees polls that could be very different from what the reality of public opinion is. People may question their own opinions as a result. It's a way to manipulate voters, and/or public opinion, plain and simple. But it's not real polling, it's an agenda driven exercise in propaganda.

Okay, on to Canada.

Pierre Poilievre, Canada's common sense conservative candidate has been decimating the Liberal party in polls. This of course is in large part due to Justin Trudeau's stench of putrid leadership and governance.  That's not hyperbole by the way. But it is also not entirely on Trudeau; his ineptitude, narcissism and tyrannical approach to governance merely opened the door. Trudeau, despite that, has defeated other conservative leaders prior to Pierre Poilievre.  Pierre Poilievre is different. He's changed. The Liberals have not. 

Justin Trudeau stepping aside as leader and using parliamentary tricks to stall the forcing of an election was supposed to help the Liberal party recover from its long, unimproved disastrous polling. It appears it did and Canadians are dumb enough to vote Liberal party again. But... push polling:

It turns out, the poll showing a Liberal recovery minus Trudeau, was a push poll.

My take on the state of the race

A lot of polls being included in the RCP polling average for the swing states, are of little value.  For example the NYT poll does not show a Margin of Error.  When polls exclude details like that, or survey size, for example, they are either not confident in their results or are hiding something. I exclude those polls based on threshold criteria, along with polls that are too small, or include perhaps all adults instead of Registered or Likely Voters, or ones that may oversample Democrats, which often happens.  Those types of polls are not exactly push polls for the respondents but are a form of push polls for the voting public.  For example polls trying to generate astroturfed momentum for Kamala Harris.

I have not in this exercise even looked at the crosstabs and these polls may indeed still be overemphasizing the positive look for Harris. All that to justify what I have included in my polling calculations.  What I have done is aggregated the polls that are worth including (minus the bad crosstabs filter) into a vote total for each state to see what the good polling is actually showing across multiple pollsters with similar, acceptable methodology/disclosure polls.

Here's my criteria (these are relatively loose criteria, but if they were any tighter, there would not yet be enough to share any insights):

 

Here's what I see.

Based on the above, here's what I would project, a Trump win.

Keep in mind there are a number of states that do not have enough polling to report.  Georgia is one such example but others I have defaulted to most likely based on previous cycles.  These include New Hampshire, Virginia, Minnesota and Maine.  These were leaning Trump when he was facing Let's Go Brandon, but have since not shown any polling results.

My honest opinion is that this will break more for Trump by November but that remains to be seen.

Push Polling the polls the day before?

I haven't talked about persuasion polls much (if at all).  Unlike typical push polls, a lot of political polls are not intended to sway those being polled but rather those viewing the results of the polls. That could be what's happening right now.  The RealClearPolitics average of generic congressional ballots has seen the Republican advantage slip back to +2.5% since pesking well above that mark last week.  Why?

It is coming from the Economist/YouGov poll (which has historically had a Democrat tilt in it's results) showing only R+1%, the outlier NBC poll showing D+1% and the ABC News/WaPo poll showing R+2%.  These all have dragged down the Republican advantage and all come out in the final days before the midterm election.  All of these outfits are pro-Democrat.

In my filter above I've left them in and am still seeing R+3.2%.  What if we take them out as dubious inclusions?

Even leaving the ABC/WaPo poll in, it still rises to R+4.2%.  But then what about Rasmussen which has a slight Republican tilt? If we remove that, Republicans still lead by 3.5%:

There's no bad way to slice this but it makes predictions difficult.  Here's my take on final numbers:

Senate: Republicans 53 or 54.

House: Republicans gain 28-48

Governors: Republicans 30-32

It's optimistic compared to some, but the polls, even the likes of YouGov are still shifting towards Republicans, and hopefully, that's not done yet.