Quote:
Originally Posted by Iowa_Flames_Fan
Based on topline margin Mainstreet’s last tracking poll (UCP +4) is actually closer to the last Leger numbers (NDP +2) than it is to Janet Brown. I don’t see them as “pretty much in line.”
To me the two that stand out are Abacus and Janet Brown. Everyone else is kind of within a standard deviation of one another.
One weird thing about the Janet Brown poll is it’s not even “pretty much in line” with Janet Brown. Her last poll showed the NDP ahead in Calgary by 5 points, in early April. She’s effectively saying that over the last 6 weeks, significant numbers of Calgarians watched Danielle Smith and thought “yes, I’d like more of that please.”
To me, that seems way less plausible than just saying “this is an outlier”, which is just a thing that happens sometimes and in no way a criticism of her poll or her methodology.
|
I wouldn't worry about a single poll. No one - even a good pollster is going to nail everyone of them. It gets tougher now because hardly anyone answers calls from a number/name they don't recognize. Even if you have set it up so it comes up as 'Polling company' on the display name - you are biasing your sample because only people who want to do a poll are going to answer.
Its also more difficult when voting trends shift - if you look in the US - a state like Florida that has shifted red in recent elections as closer than actual because you are doing assumptions that x person based on this demographic is going to vote this way based on past trends and your sample of 3 people who actually answered the phone in that demographic.
She could be right, she could be wrong - but in general - typically the polling average is decent. That 2008 example given in the article - if you look at the wikipedia article and the polls showed - they seem to be pretty close to end result -
https://en.wikipedia.org/wiki/2008_A...neral_election