‘Not talking to enough people:’ why polls predicted Clinton, not Trump

The polls said Hillary, but the ballots said Donald Trump in an election outcome that stunned many.

Early Wednesday morning, Trump was named president-elect of the United States.

David Valentin, executive vice-president of Mainstreet Research, pointed out several factors that could have skewed polls leading up to election day.

One reason, Valentin said, is the move to online polls. Another reason – smaller sample sizes.

“(They) are really affecting the way that people are able to drill down into the data.”

Valentin said polls in the United States will poll as few as 1,000 people – compared to the 2,500 to 5,000 people typically polled in Canada, a far less populated country.

“We’re talking about a very small sample of people, especially when you compare it to the actual population size in America.”

Valentin said polls in the United States talk to so few people due to the cost of polling over the phone.

“They have these very strange regulations they brought in the 1920s-1930s,” he explained.

“What that means is the way you call a cellphone right now in the States for a poll interview, you actually have to have the operator punch in the number by hand.”

In Canada, there are programs and technologies that dial the numbers, which speeds up the process.

Polls wrong – again 

This isn’t the first time polls for high-profile elections have been wrong.

“This is actually the fourth time the polls have been wrong for a major electoral event in the last 14 months,” Valentin said.

Polls also incorrectly predicted the outcomes of the Greek bailout referendum, the Brexit vote and the Colombia-Revolutionary Armed Forces of Colombia (FARC) peace vote.  

Valentin summed up polling errors: “This all comes down to the factor of not talking to enough people.”