A Liberal win was the expectation. A win of this magnitude, maybe not. 

The surprise in last night's results came in the breadth of the Liberal victory, with unexpected gains in British Columbia, Quebec, and a sweep of Atlantic Canada. With the Conservatives performing only slightly below expectations in the seat count, these surprise wins came largely at the expense of the NDP in some of their most secure ridings, as the Liberals' momentum swept up strategic voters to carry them over the majority threshold. 

The polls did not miss the call, but many of them missed the last bit of gains the Liberals made in the final days — the gains that made the difference. In fact, polls conducted only on Sunday (those by Nanos Research and Forum Research) nailed the call almost exactly. Polls conducted earlier in the week or including data from earlier in the weekend had the Liberals too low to make a majority government a likely possibility.

Trudeau crowd

Liberal Leader and prime minister-designate Justin Trudeau shakes hands as he leaves following his victory speech in Montreal on Oct. 19, 2015. (Justin Tang/Canadian Press)

The CBC Poll Tracker projected the Liberals at 124 to 161 seats, but the maximum projection put the Liberals at 185 seats, recognizing the potential for this sort of outcome (the Liberals won 184) but not considering it a high probability result. At 99 seats, the Conservatives fell just outside their likely low range (which bottomed out at 100 seats), but well inside their minimum range. The NDP's 44 seats put them below their low range (which was 51 seats), but also within their minimum projected range.

The most significant source of this discrepancy came from the poor performance of the New Democrats. Their disappointing showing in Quebec helped deliver 40 seats to the Liberals (10 to 19 more than expected). The polls were pointing to a gap between the Liberals and NDP in that province closer to three percentage points, rather than the 10 points it actually turned out to be. With the NDP's support distributed across the province, rather than concentrated in a few regions, the party took just 16 seats.

The Liberals also had unexpected vote efficiency. They had the momentum, but the party did not seem to have the electoral map to give them 170 seats. In 2011, the party earned 82,000 votes for each of their 34 MPs, whereas the Conservatives needed about 35,000 votes per seat. In 2015, the Liberals were the vote efficiency victors, needing under 38,000 votes per seat. The Conservative requirement ballooned to 57,000 votes per seat, far higher than it has been in any election since before 2004. The Liberal vote got out where it needed to. 

But perhaps more striking was the defeat of a large number of NDP incumbents in ridings that would have been considered safe for them, and where there was no danger of Conservatives being elected.

Ridings like St. John's East (where Jack Harris was the incumbent), Halifax (Megan Leslie), Gatineau (Françoise Boivin), Ottawa Centre (Paul Dewar), Toronto–Danforth (Craig Scott), Winnipeg Centre (Pat Martin), and a number of Northern Ontario ridings, to name but a few. That the New Democrats lost such a large number of votes to the Liberals in these ridings is a strong suggestion that progressive swing voters swung hard for the Liberals. Polls in the last week showed that the NDP's voters were less committed than those of the Liberals, and Justin Trudeau's party was, by far, the one Canadians expected would win on election night.

What of the Conservatives? There was no significant "shy Tory" effect or organizational advantage that can be seen in the numbers. The Tories did outperform their polls a little, particularly in Ontario, but it wasn't enough to make any difference. They were hit hard in British Columbia, taking 30 per cent of the vote (about as expected) but just 10 seats. For the most part, however, the Conservatives lost the seats they were expected to lose. Strategic voting at the national level helped the Liberals against the Conservatives, but at the riding level it was primarily a problem for the NDP.

The performance of the polls

Overall, the performance of the polls was largely positive. The Liberals were assessed to have between 38 and 40 per cent by five of the seven pollsters that were in the field in the last week of the campaign (the Liberals took 39.5 per cent). All pollsters had the Conservatives between 30 and 33 per cent (the party took 31.9 per cent). And all pollsters had the NDP between 20 and 22 per cent (instead of the actual 19.7 per cent). The Poll Tracker pegged the three parties at 37.2, 30.9, and 21.7 per cent, respectively.

So the calls of most pollsters were virtually all within the margins of error of the polls. But these numbers show why a call of a majority government was difficult to make. Two pollsters had the Liberals at 35 or 36 per cent, and three others had them at 38 per cent. That would not have been majority territory for the party. The slight over-estimation of the NDP (from a few tenths of a percentage point to just over two points) appears to have been behind the slight under-estimation of the Liberals. That alone was enough to turn a minority into a majority.

The result highlights the importance of momentum in the late stages of a campaign, and how polling needs to be done up to the last moment to capture the final snapshot of a fluid and shifting race. The Liberals were on track to win this election almost two weeks ago. They might have only won the majority in the last few days.