Election Writ 4/7: When polls disagree
Is it a Liberal landslide, a modest Liberal lead or a neck-and-neck race?
When the polls start to diverge in an election, the campaign becomes a bit of a Choose Your Own Adventure.
On the one hand, the race might be tight — really tight. Or it could be a Liberal stomping. Or it could be somewhere in between.
Let’s try to make some sense of it.
The Numbers: Week 2 goes to the Liberals, too - On Saturday, Philippe and I recorded our weekly bonus episode of The Numbers that is reserved for members of our Patreon. You can join and listen to the episode here.
Also in today’s newsletter: A deep-dive into some B.C. regional polling numbers, how Albertans view separation and The Writ’s Campaign Bus stops in Victoria.
Since Saturday’s update, six polling firms have put out new numbers. They can be grouped into three categories:
Liberal landslide: Surveys by Ipsos (for Global News) and MQO Research (for Global Public Affairs) give the Liberals a double-digit lead over the Conservatives. Ipsos put the gap at 12 points (46% to 34%) and MQO put it at 10 points (44% to 34%). With numbers like that, the Liberals would be virtually guaranteed to easily clear the bar of 200 seats.
Neck-and-neck: A new poll by Abacus Data put the Liberals and Conservatives tied at 39% apiece. This wasn’t a tightening of the race as Abacus had shown a tie in its previous survey. Despite the national tie, the Liberals’ regional distribution of support would still make them the heavy favourites to win the most seats, though not necessarily a majority government.
Modest Liberal lead: The daily trackers are somewhere in between, with Liaison Strategies, Nanos Research and Mainstreet Research putting the Liberal lead at between four and six points this morning. These three polls, despite using wildly different methodologies, are all producing broadly similar results. The Liberals would win a majority government with these numbers.
It should be noted that these polls weren’t all done at the exact same time. When Ipsos was in the field between April 1 and 3, both Liaison and Nanos were showing a wider Liberal lead than they do now — seemingly corroborating Ipsos’s finding that those were good polling days for the Liberals.
But Abacus was also in the field on the same dates as Ipsos. And the new MQO poll was done on the same dates as the latest trackers. Field dates don’t seem to be the reason for the discrepancies.
So, we’ve definitely got some disagreements here. The net effect in the Poll Tracker, though, is that the trendline is steady — Ipsos and MQO balance out Abacus, while the trackers are broadly in line with where the projections had already been.
But let’s try to figure out whether these differences between the polls really matter — and what could be behind them.