B.C. ELECTION: Can We Trust Polls?
Featured / May 5, 2017
The NDP were projected to win by a landslide in 2013, so what are we to think?
Tristan Johnston, Coordinating Editor
When I was speaking to Linda Reid, a B.C. Liberal running in Richmond South-Centre and current speaker of the House, she said that if the Liberals took the polls as gospel in 2013, they wouldn’t have bothered door-knocking.
In 2013, the polls suggested that the NDP would crush the Liberals in the B.C. election, but just the opposite happened. How could this be? Can polling still be trusted?
A big reason polling numbers can be shaky is that they’re gathered on the phone. If you’re having a busy day, or even a relaxing day, the last thing most people want to do is talk to a pollster, or worse, a robot. Everyone these days has caller ID, and some will simply hang up the minute they know it’s just a survey.
After the 2013 election, pollsters considered low voter turnout to be a factor in the inaccuracies of the polls, as well as youth turnout being unreliable.
When it came to the 2016 presidential election in the United States, many pollsters believed that Clinton would win both the popular vote and the electoral college, but she only got the popular vote. The bizarre American electoral system aside, pollsters might have been using models and weights which assumed that certain groups of voters would continue to not show up. For instance, if you had been a strongly right-wing voter, you might have felt that the last few elections in the United States didn’t involve parties or politicians who represented you.
Similarly, pollsters might be cautious when calculating their predictions for some Richmond ridings. In 2013, Richmond Centre had a low turnout in the high 40’s. Seat calculations will also account for the fact that environmental issues might be a lower priority for northern and interior communities in B.C. that need projects in mining and pipelines for employment.
There are other calculations that could be missed. In the last election, one firm calculated their projection based on a turnout of 57 per cent, but the actual turnout was 52 per cent, which was one per cent higher than the year before. Another factor was that both leaders at the time were unpopular with voters.
When looking at polling data for the current election, one will notice Interactive Voice Response as a method, which is an automated way of finding out what voting intentions are like. Pollsters often consider quality and quantity of data, and generally speaking, the more the merrier. For instance, Mainstreet Research polled over 5,000 people with IVR on April 10. The problem this method is that many respondents are likely to consider it spam, resulting in a lot of quick hangups.
There’s also a chance that polling can create false confidence. As of April 25, Grenier’s B.C. poll tracker puts the NDP at 42.5 per cent and the Liberals at 34.7 per cent. In Canada and the United States, conservative voters, typically older voters, are seen as extremely reliable for turning out, while younger, more liberal voters are known for being fickle.
Of course, youth voter turnout increased during the federal election. The models seen currently are probably fairly accurate, and in a sense they were technically accurate during the 2016 presidential election in the United States. Many had predicted Clinton would win the popular vote with a 3 per cent gap, but instead, it was closer to 1 per cent.
Polls were also accurate during our federal election in 2015. EKOS, Mainstreet Research, Forum Research, and Nanos Research were all extremely close to the final result on data released the day before voting day. Nanos had almost perfectly estimated the NDP and Liberal results, and EKOS was precise in their prediction of the Conservative result. Mainstreet, who’s also conducting research in B.C., used a sample of 5,000 with IVR and was within 0.5 per cent of the final results.
Regardless, pollsters have been losing trust. Brexit, the 2016 presidential election, and B.C.’s 2013 polling were all seen as failures by research companies. As time continues, polls will make adjustments to sampling techniques and how they weigh results. But no matter what, don’t trust other voters to do your part. The only way to ensure your riding will go the way you want it to this year is to do it yourself. Go and vote.