Polling ahead of the US election pegged Hilary Clinton as the likely winner but as you know, Trump is now the president-elect.
David Valentin with Mainstreet Research says pre-election polls in the US are faulty compared to Canada’s.
“In Alberta for the last Alberta election, we were the first polling firm to predict the NDP majority and that poll had a sample of 3,000 Albertans,” explains Valentin. “Most national polls for the whole United States have a sample size of 1,000 respondents, that’s one-third of what we would do for Alberta.”
Valentin says there are also laws that make polling more difficult.
“Part of that is because it is so expensive to do polling in the United States,” explains Valentin. “They have regulations down there that don’t allow you to use any automated technology to call cell phones and that means they literally have to have agents who are dialing in cell phone numbers by hand.”
He says another major problem with US polls is that pollsters follow each other.
“A lot of pollsters may have had Trump winning a state or winning the popular vote but may have said – well, everyone has Hilary up three or four, we’ll just change our formula a little so we can say that too,” explains Valentin. “Or they may have said – we’re just not going to publish.”
Valentin says there were four major polling errors in the last 14 months, in Greece and the UK for Brexit, Columbia and now the US election.
He adds Canadian polling agencies embrace technology, which makes polls more accurate, and the US does not. (twd)