Despite a spell of new polls heading in different directions injecting a good deal of uncertainty into what to expect from tomorrow's election in British Columbia, the B.C. New Democrats remain the favourite to win and form the province's next government.
The latest forecast for ThreeHundredEight.com ( full methodology ) and The Globe and Mail gives Adrian Dix's NDP 46 per cent of the vote, compared to 38 per cent for the incumbent B.C. Liberals, 9 per cent for the B.C. Greens, and 5 per cent for the B.C. Conservatives. Another 3 per cent of British Columbians are expected to vote for independent candidates and other parties.
Since the post-debate projection that incorporated polling data up to May 2, this represents a gain of two points apiece for the NDP and Liberals, a drop of one point for the Greens, and a drop of two for the Conservatives. In other words, the margin between the two main parties remains unchanged at eight points.
Based on current polling levels, Mr. Dix appears safely in majority territory with a projected 45 to 56 seats, above the 43 seats needed to form a majority government. Christy Clark's Liberals are projected to win between 26 and 40 seats, while up to four independents should be elected. These numbers are virtually unchanged from the last forecast, while the precise projection of 48 seats for the NDP and 36 for the Liberals (along with one independent) is the same.
However, the polls have become more volatile and are showing less consistency than they did in other stages of the campaign. This makes it difficult to estimate how voting intentions may move between May 10 (the last day that polls were in the field) and May 14. Some final polls are expected to be released before midnight in British Columbia, when the province's election law prohibits the publishing of new opinion surveys. These polls may provide more of a clue of where the trends are heading in the final hours – if voting intentions are still in flux at all.
The polls strongly suggest that Adrian Dix will win tomorrow night. The odds that a gap of eight points can be overcome in the final four days of a campaign are about 3 per cent, based on recent history. The example of last year's provincial election in Alberta comes to mind, but the expectation that such an exceptional event could occur in two of the last three provincial campaigns should not be very high. Nevertheless, a surprise cannot be ruled out. The projection model gives the Liberals a roughly 21 per cent chance of winning the most seats tomorrow night, taking into account the possibility of the polls not proving reflective of the final result and the Liberals' vote being distributed in a disproportionately advantageous way.
This uncertainty would not have been expected when the campaign began and the New Democrats were polling 18 points ahead of the Liberals. But the last series of polls released on Friday were somewhat contradictory, at least in terms of the trends they were showing. Those by Forum and Ipsos-Reid recorded a narrowing gap while the survey by Angus-Reid suggested a widening margin. But in all three of these polls, the change in support was relatively insignificant. As the polls could not agree on a trend, that may instead suggest that the race has stabilized and the shifts in support were due to normal statistical variation.
The strength of the B.C. Liberal vote seems to be the murkiest. The polls have been relatively consistent for the New Democrats, putting them between 41 and 45 per cent support. The Liberals, however, have ranged between 31 and 41 per cent. That is a yawning gap, and is the difference between a very long election night and a wipe-out.
Though Ms. Clark has certainly made the race more competitive, there are some suggestions that she may not be able to pull off a full comeback by tomorrow night. For one, her momentum might be stalling. The debate appears to have boosted her party's support by three to four points, but since then the Liberals have only picked up two more points. Ms. Clark may be reaching the limit of Conservative votes that can be pulled to her side (not to mention those of uncommitted NDP and Green supporters).
Her personal numbers also remain problematic. While she has narrowed the gap or even surpassed Mr. Dix on the question of who would make the best premier (Ipsos-Reid gave it to Ms. Clark by three points, Angus-Reid put Mr. Dix up by five), her approval ratings are dismal. They may no longer be in the mid-20s, as they were in March, but she has yet to do better than 35 per cent approval and her disapproval rating has settled in somewhere between 57 and 61 per cent. By contrast, Mr. Dix has managed about 41 to 42 per cent approval in recent surveys, with a disapproval rating just under 50 per cent. Generally speaking, it is very difficult for a premier to be re-elected with less than 40 per cent approval.
The regional numbers also look good for the New Democrats. Their support in metropolitan Vancouver has hardly budged since the campaign began, and the party there remains comfortably ahead of the Liberals. On Vancouver Island, Green support has slipped a little, reducing the threat to the NDP's dominance on the island. Though a local movement to elect a Green MLA may be un-recordable in province-level polling (and could still occur), the numbers lean towards the Greens remaining shutout.
The one part of the province where the election could be decided is in the Interior and north. Of the five most recent polls, the New Democrats have led in the region in three of them, were tied with the Liberals in another, and trailed by a wide margin in the fifth. If Liberal incumbents in the area prove more resilient than expected, they could reduce the size of Mr. Dix's majority or even put it into question.
But ruling out a late swing in voting intentions – or a colossal error by the polls – the B.C. New Democrats should win a majority government tomorrow night and end the 12-year reign of the B.C. Liberals. However, there is enough doubt about the election's outcome to give British Columbians even more of an incentive to get out and vote, and for observers both inside and outside the province to be prepared for a long night.
All publicly available polls are weighted by sample size, date, and the polling firm's accuracy record. The seat projection model makes individual projections for all 85 ridings in the province, based on the provincial shifts in support since the 2009 election and subsequent by-elections. The probability forecast is based on historical changes in support between polls and election results. The projection is subject to the margins of error of the polls included as well as the unpredictable nature of politics at the riding level.
Éric Grenier writes about politics and polls at ThreeHundredEight.com .