BC Liberal Leader Christy Clark waves to the crowd following the BC Liberal election in Vancouver, B.C., on May 10, 2017.JONATHAN HAYWARD/The Canadian Press
Paul Fairie is a political scientist in Calgary.
Unless absentee ballots or recounts grant either the Liberals or NDP a majority government, neither party will be fully satisfied with the result of the provincial election in British Columbia. However, one group with an interest in the election result will surely be happy with how things turned out: pollsters.
Four polling firms released results in the final days of the campaign, and all four seemed to agree on the final result: It was a close race. The polls measured the Liberals at between 39 and 41 per cent of the vote, the NDP at between 40 and 41 and the Greens at somewhere in the 17 to 20 range.
Though a significant number of ballots remain to be counted and a seat or two might still change hands, the popular-vote shares for each of the parties should not change significantly from where they sit now: The Liberals won 41 per cent of the vote; the New Democrats won just shy of 40 per cent; while the Greens won their highest share ever, 17 per cent.
This represents a strong success for those in the polling industry, a sharp contrast to their less than successful ventures estimating voter intention during the 2013 provincial election. In the final days last time, the NDP led every publicly released poll throughout the campaign with polled leads of between 6 and 9 percentage points in the final days. Yet, after all of the ballots were counted, the Liberals ended up winning the popular vote by more than 4 percentage points, and received a healthy majority of the seats in the legislature.
Some might still criticize pollsters because three of the four final polls showed small NDP leads. However, these polls stopped surveying British Columbians between one and three days before election day, and small numbers of voters changing their minds in the last few days is very common.
The leader in the poll when the margins are so small is less important than the fact that the polls showed a tight race. Often, calling the winner correctly might appear to be more important than estimating voter intention, such as in the recent second round of the French presidential election. There, polls showed eventual winner Emmanuel Macron ahead of Marine Le Pen by about 25 percentage points, while the results revealed the ultimate gap to be 32.
While few seemed disturbed by this miss because the polls identified the winner correctly, the difference between polls and result was larger than in last year's Brexit referendum where pollsters were roundly criticized. For B.C.'s results to match the polls so closely makes it a good day for pollsters involved in the race.
Ultimately, it is always important to remember that polls, even those taken in the final days of a campaign, are subject to many sources of inaccuracy. For example, the normal and inescapable sampling error that arises when randomly sampling a population can be measured with a margin of error. While these details are often found at the ends of stories covering opinion surveys, they are still just as important to read.
Others are more difficult to quantify. Sometimes certain voters are more reserved about revealing their own opinions. In the 1992 United Kingdom election, "Shy Tory" syndrome led some Conservative Party supporters to conceal their vote intentions because they felt it was an unpopular preference to hold. Other times, voters can simply change their minds, even in quite significant numbers, in the final days of a campaign after polling has stopped.
This is not to say that polls cannot be useful. Indeed, pollsters should be very pleased with their efforts in British Columbia over the last few weeks. However, despite this good result for pollsters, it remains important to always cast a critical eye on the results of polls so that neither their successes nor failures lead to too much confidence or excessive cynicism.