US polling majorly flawed: Mainstreet Research
Posted November 9, 2016 12:45 am.
This article is more than 5 years old.
VANCOUVER (NEWS 1130) – This is a major US polling failure.
That reaction to the election results from David Valentin of Canada’s Mainstreet Research.
Mainstreet declared a liberal majority in Canada for last year’s federal election, which turned out to be accurate. Valentin says every state poll in the US has a high degree of error in the polling.
He says there are a lot of factors that set polling in Canada apart from that in the states.
“One of them is sample size,” explains Valentin. “Here in Canada, we have fairly large polls. Not every firm, but my firm in particular, believes in large sample sizes. A typical national poll for us would be 5,000 respondents.”
Valentin says a typical national poll is about 1,000 respondents. One of the reasons the number of people surveyed is so much smaller in a more populated country is because they can’t use auto-dial technology. In essence, you have to punch in a cell phone number by hand.
“And that really raises the cost of doing business in the US when it comes to doing market research. And what that means is you can’t afford to have these large sample sizes that we have in Canada.”
And of course, there’s alway the chance that people don’t always tell the truth.
“I think that was part of the problem. I think another big part of the problem was pollsters weren’t able to contact enough folks to really build a correct likely-voter screen. And when you see the polls that are being reported in the media, all these polls, they take these people and score them and rank them, and they try to figure out which parts of the population are the most likely to vote. And some pollsters will combine that with early voting data and other factors,” explains Valentin.
“And again they’re not talking to a big enough group of people. They are only talking to, you know, 20 Latinos who are 18-35 without college degrees. They’re only talking to 15 women between the ages of 34 and 49 who work in this industry or that sector. You’re talking about a very small slice of people and you’re hoping that they’re representative of the population at large.”
