Federal election 2015: Can polls 6 months out tell us who will win?
Polls do have some predictive power, even this far out from the vote
Politicians are fond of saying the only poll that matters is on election day. But polls taken six months before that day can tell us quite a bit, too.
If we look at polling done in pre-election periods over the last few decades, some patterns emerge. The first is that they actually do a very good job of suggesting who will win the vote six months later. The second is that, despite this predictive capability, campaigns matter. Voting intentions can shift dramatically.
And there's a third pattern, which does not bode well for Liberal Leader Justin Trudeau.
Back to the first pattern. Over the last six federal election campaigns where public polling records are relatively robust, only once did the party leading in polls conducted five to seven months before the election fail to secure victory. If we stretch that to the last 11 elections using historical polling data from Environics, the trend continues. In that span beginning in 1979, only twice — in 1988 and 2006 — has the leading party failed to win the most votes six months later.
Those two elections saw big changes during the campaigns. Going into the 1988 free-trade election, the contest was still a relatively close three-way race. It was an entirely different affair in 2006, the only other recent instance of a party losing the lead it held six months before the vote.
Paul Martin's Liberals enjoyed a comfortable nine-point advantage in mid-2005. Though an average of 36 per cent support put a majority government in doubt, the polling margin over the Conservatives was large enough to make it possible. However, the polls shifted in the midst of the 2005-2006 campaign, and the Conservatives won their first victory. Perhaps it was not a coincidence that their last win had come in 1988, when the party had also trailed six months before the vote.
Liberals see frequent drops
The failure of the Liberals to secure re-election in 2006 was just another case of the party faltering in the run-up to an election. In each of the last six elections, the party has dropped significantly from its pre-election standing. Some of the decreases were relatively modest: five points in 2000 and 2008. Others were significant, with the party dropping 10 points or more in the run-up to the elections in 1997, 2004, and 2011.
With the Liberals and Conservatives currently neck-and-neck in voting intentions, the tendency for polls to accurately predict the winner six months out is nullified. But the trend of the Liberals under-performing their pre-election polling at the ballot box tilts the odds in the Conservatives' favour.
The Liberals have averaged a drop of about six points per election since 1979, if we include the more limited polling prior to 1997. Since 1997, that drop has averaged eight points. That is virtually identical to the gains made by the Conservatives and their predecessor parties over that period of time.
In 2011, it was in fact the New Democrats who gained the most at the Liberals' expense. That was the exception. Between 1997 and 2008, the pre-election polls always proved to be predictive of the NDP's eventual vote tally to within two percentage points.
History, then, is not always fated to be repeated. The New Democrats broke the mould in 2011 in more ways than one, and it has been almost 30 years since three parties held a legitimate shot at forming government this far out from the vote. But if the Liberals are to win the federal election six months from now, they will have to break some old trends.
This article reviews trends in national public opinion surveys. Methodology, sample size and margin of error if one can be stated vary from survey to survey and have not been individually verified by the CBC.
ABOUT THE AUTHOR
