With six months to go before the next provincial election in Ontario, two narratives are forming in the polls. In the first, the Ontario Liberals are trailing the Progressive Conservatives by 10 points or less and will struggle to hold onto power. In the second, Liberal support has sunk so low that the party is not only on track for a massive defeat, but could place third behind the New Democrats.
With the exception of Patrick Brown's PCs being favoured to win the next election, these two narratives are difficult to reconcile.
The latest survey from Forum Research, conducted Nov. 29-30 and published by the Toronto Sun on Monday, put the PCs at 40 per cent, followed by the NDP at 26 per cent and the Liberals at 24 per cent.
Those results are typical for Forum, which has put Kathleen Wynne's Liberals behind the NDP in six of eight polls published so far this year.
But they are odds with the numbers that have been published by the two other polling firms consistently active on the provincial scene in Ontario.
A poll by Campaign Research conducted in early November put the PCs at 35 per cent to 32 per cent for the Liberals and 23 per cent for the NDP. The Innovative Research Group, polling in mid-November, put the Tories ahead with 41 per cent to 31 per cent for the Liberals and 19 per cent for the New Democrats.
Again, these were broadly typical for 2017.
Since the beginning of the year, the PCs have averaged a 6.2-point lead over the Liberals in Campaign's seven polls. In five polls released publicly by Innovative, that Tory lead has averaged 6.7 points. In none of these surveys have the Liberals been in third place.
In Forum's eight polls, however, the Tories have averaged a lead of 19 points over the Liberals, with Andrea Horwath's New Democrats edging out the Liberals by 1.7 points on average.
Method behind the madness
The reason for this discrepancy is unclear. But the methodology each pollster employs might be behind it.
Forum uses interactive voice response (IVR), in which people punch in their answers on their keypad in response to recorded questions (For Progressive Conservatives, press 1. For Liberals, press 2, etc.). For its voting intentions surveys, Innovative uses live callers to conduct its interviews, while Campaign employs an Internet panel, getting respondents to take their surveys online.
Why, however, these methodologies are producing these different results is not obvious. In its final poll of the 2014 election, Forum over-estimated PC support by four points but also over-estimated Liberal support by two points.
Four online pollsters over-estimated the PCs by an average of two points and under-estimated the Liberals by an average of four points. Only one telephone pollster was in the field in the last days of the campaign and it also under-estimated the Liberals.
Neither Innovative nor Campaign publicly released polling results prior to the 2014 election results.
But this is contradictory to the current situation as it would suggest the Liberals' stronger results in the online and telephone polls should be even better, while the poor Liberal results in Forum's IVR poll should be even worse.
Methodology or volatility?
Richard Ciano, principal at Campaign Research, says that his company moved away from IVR because of its concerns over difficulties in reaching younger respondents, difficulties he says have increased in recent years.
Greg Lyle, president of Innovative Research Group, points to some of the volatility in the Liberal and NDP vote within his own polling.
"There is more movement among centre-left voters than on the centre-right," according to Lyle, "which means different polls could get different reads on the relative strength of the NDP and Liberals, depending both on their methodology and on the media dynamic during the field window."
Forum Research President Lorne Bozinoff would not comment on the results of other polling firms.
A blue majority or a minority of some shade
Either narrative could be correct. In 2014, Forum's numbers pointed towards a Liberal majority — the actual outcome — while online pollsters were showing a much closer race. But online pollsters have also had success in other recent elections, as have pollsters using live callers.
The implications of the two narratives, however, are significantly different. A projection model using the same methodology as the CBC's federal Poll Tracker shows why.
Using only Forum's numbers, the PCs would have a 93 per cent chance of forming a majority government and a 98 per cent chance of winning the most seats. The Liberals would have less than a two per cent chance of winning, and would be projected to win fewer seats than the NDP.
With Innovative's polling, however, the odds of a PC majority government, though still high, fall to 82 per cent. Liberal odds would increase to only six per cent, but they would be heavily favoured to form the official opposition instead of the NDP.
And with Campaign's results, the PCs would stand only a 15 per cent chance of forming a majority government. A minority government headed by either the PCs or the Liberals would be the most likely outcome, with the Liberals' chances increasing to 25 per cent.
Six months out from an election campaign, these competing narratives might not have important consequences. Voters aren't paying as much attention to Ontario politics as they will next spring. Once the campaign begins, however, these polls would paint very different portraits of the race — with potentially serious implications on media coverage and voters' decisions.
Hopefully things will clear up before then. The polls do show that Wynne's Liberals are in trouble. But beyond that, there is no consensus on exactly what the political landscape in Ontario looks like.