The timing of Justin Trudeau’s decision to call an election certainly had all the trappings of shrewd political calculus, but given the Liberal party’s then-lead in the polls it was a bet that seemed likely to pay off. Pundits and pollsters alike suspected an easy win for the incumbent Trudeau Liberals and a probable return to majority status.
As we approach the halfway point in the campaign the ground appears to have shifted dramatically with the Liberals seeing a precipitous decline in vote share and the Conservatives gaining significant ground — if the polls are to be believed, that is.
Polling in Canada has had a notoriously spotty record when it comes to predicting election outcomes. One need only consider the Nova Scotia provincial election weeks ago where the Progressive Conservatives defied expectations and unseated a Liberal government that was unanimously favoured to win in the polls.
In fairness to the craft and those who practice it, perennial problems of probability sampling and modelling voter turnout are increasingly complicated in the digital era. Moreover, pollsters have been harshly condemned for vote share predictions that are at times off by only a percentage point or two from actual election results (results which themselves often fall within the margin of error). But in an electoral system like Canada’s those percentage points can make or break a party’s chances to form government.
In an attempt to produce more reliable election forecasts poll aggregators have become something of a cottage industry in Canada. These “polls of polls” work by modelling the systematic bias of each polling firm using its historical projections. As a crude example, if a given polling firm systematically underpredicts the Conservative vote share in previous elections a poll aggregation model would correct for this bias and adjust forward-looking projections from said firm accordingly. Adjusted projections from each firm are then aggregated in order to generate an election forecast.
During the 2015 Canadian federal election, Vox Pop Labs partnered with the Toronto Star to launch its own poll aggregator called The Signal. Our model differed from most others in Canada, taking inspiration instead from the approach of people like Nate Silver and Drew Lizner in the United States. Our 2015 forecast was ahead of the pack, but a single success does not merit bragging rights. So, after a hiatus in 2019 we have relaunched The Signal in an effort to demonstrate that its initial success is reproducible. Already we have seen it deftly track shifts in the election campaign thus far, several of which warrant close attention as they may ultimately define the outcome of this race.
The consensus among firms that have fielded polls in the last week is that the Conservatives have pulled ahead of the Liberals in terms of vote share, but this does not necessarily translate into enough seats for the Conservatives to form government. Recall that in 2019 the Conservatives ended up more than a percentage point ahead of the Liberals when the votes were ultimately tallied but wound up with 36 fewer seats in the House of Commons. Even the pollsters most bullish on the prospects of a Conservative victory presently see the party only having enough seats to secure a minority government. As of earlier this week, however, The Signal has the Conservatives ahead of the Liberals in both vote and seat share albeit still out of reach of a majority government.
Conservative gains have come mainly at the expense of the Liberals in British Columbia and Ontario. Erin O’Toole’s significant overtures to Quebecers have seen the Conservatives make subtle inroads in the province, but not yet the watershed that they are clearly seeking.
The Liberal vote in B.C. appears to remain in a free fall with the party plummeting from first to third place in the provincial polls — behind the NDP — since the writs were issued. In Ontario there are early signs of stabilization in support for the Liberals after the Conservatives overtook them early in the second week of the campaign.
The close race in Ontario is one of several factors that suggest that the outcome of this election is far from a foregone conclusion; another is the ostensible rise of the People’s Party of Canada. Various polling firms have recently pegged the PPC at record high levels for the upstart party, although polling aggregators — including our own — struggle to validate these estimates for largely technical reasons. Inconsistencies in the way polling firms have included the PPC in their reporting complicates efforts to accurately model their support. These challenges are exacerbated by a lack of observations of past PPC performance in federal elections. If the PPC vote is indeed as large as some pollsters are suggesting, it may serve to materially split the small-c conservative vote. At present, however, our model indicates that PPC support is far from the levels needed to win seats in the House of Commons.
The next week will be critical in terms of determining whether the Conservatives can sustain their momentum and if the Liberals can regain ground by appealing to anything-but-Conservative voters. Justin Trudeau will no doubt hope to repeat his performance of 2015 when the Liberals ascended from third place in the polls to a majority status in the final stretch of the campaign. Of course, circumstances are different now: Trudeau is the incumbent instead of an emergent challenger. Moreover, Jagmeet Singh is no Tom Mulcair and will not be readily outflanked on progressive policy positions.
JOIN THE CONVERSATION
Friends, this isn’t the time to be complacent. If you are ready to fight for the soul of this nation, you can start by donating to elect Joe Biden and Kamala Harris by clicking the button below.
Thank you so much for supporting Joe Biden’s Presidential campaign.