Polling in 🇨🇦 is wild. They literally tell you nothing but sets the narrative for discussion.
Just 1k people interviewed, 67% aged 50 or older.
No worries, the small sample of young adults will be projected, creating a 50% margin of error—like flipping a coin
OTOH … that poll appears to track with the historic results of EVERY Canada election since that Harper guy … aka Mitt Romney of the Great White North. Your country is a solid supermajority of leftist numbskulls … just like here in the decaying State of CA.
Finally,,, a poll breaking out Saskatchewan from Manitoba.
Two completely different groups of voters.
Saskatchewan, now showing a higher Conservative percentage than Alberta. Not surprised by the numbers,,, considering, with who now lives in Calgary and Edmonton.
Or Regina? Or Saskatoon?
When you look at pollsters you should never look at individual results but at the trend. On the same day or couple of days apart one pollster will differ from another because they over sample or under sample some demographics. Or even when their sampling design is the same they will just have a different draw as a result of a small sample. no big deal.
Sampling design depends on pollsters perception on who is more likely to vote not directly on proportions in the population. Ceteris Paribus, if you have 50% of green people and 50% of Blue people in the population but you believe that blue people are three times as likely to vote you will sample 750 blue and 250 green. Or you will sample them equally but give higher weight to blue. Same thing. It is always a bit of a guessing game at the end.
So when you look at results what should concern you is if the pollsters show similar trends. Graphically you want to see if the data cloud moves in the same direction over time and if the size of data cloud remains more or less constant (meaning are pollsters diverging or converging to same estimates).
To better see the picture I suggest plotting results over time one straight (not curved) line per pollster omitting pollsters like Ekos or Liaison Strategies entirely. That is the best way to see what is happening.
With that in mind, trends do not look promising at this moment for Conservatives. This may change and it has changed already more than once. The long run trend shows a drop for Conservatives. The single most important factor behind the drop is reactions of Canadian electorate, in particular of boomers to Trump.
Conservatives need to get young people to vote and need to throw as many bones in the direction of boomers as possible. Do they have enough time to turn things around? Yes, I believe so. Will Carnage’s support drop. The more he opens his snout the more he will lose, that is why he called the election as soon as he could when the initial impression created by MSM is largely positive. But this cannot last too long an he does not have the hair to hide behind. https://x.com/MarkCarney_Ego/status/1905090495090479471
But it all can go to shit if Trump continues inserting himself into Canadian politics.
A 1,000 sample set is ample enough for an accurate survey, provided there aren’t extraneous factors.
However, a national poll of a thousand leaves many questions.
Most of the polls are just too small. They need to be approx 1,000 per province to get a fair assessment. Most polls running right now are void of being completely random.
That said, PP keeps disappointing, thinking being anti-Trump is a winning hand. He needs to be more neutral.
Whatever, read what I posted with understanding or not.
One more thing, and that is anecdotal but perhaps illustrative. A pollster (well known) called me few days ago. On a landline! Why I have a landline is unimportant. Most people these days don’t have landlines and those that do tend to be elderly. Amused I contemplate answering a few questions and ask how long will it take. “about 25 minutes” they tell me.
Now ask yourself what demographic has time for a 25 minute interview on a landline? Sample response bias anyone?