The CPC Leadership Race Is A Four-Way Tossup, Despite What Inaccurate Polls Say

A new Mainstreet Research phone poll was released late last night and it supposedly shows a close race for the Conservative Party leadership between Peter MacKay and Erin O’Toole with Derek Sloan and Leslyn Lewis only in a position to decide which one of the two wins. 

On the first ballot Mainstreet had MacKay at 40.5%, O’Toole at 32.1%, Lewis at 16.2%, and Sloan at 11.2%.

In a simulated ranked ballot scenario Mainstreet determined that MacKay would edge out O’Toole on a third ballot to win the leadership race, but when one puts the poll under any sort of scrutiny the numbers tell a very different story.

First off it should be a red flag when people self identified as “Harper Conservatives” are supporting MacKay, and self identified “Red Tories” are supporting O’Toole. It seems as if people who are not engaged in the race are just remembering MacKay from the days of Prime Minister Harper, and remembering O’Toole from the 2017 race when he actively marketed himself as a red tory.

Also those self identifying as “Anti Candidate” voters are saying they are voting for one candidate to stop another, a sort of less or two evils type of voting strategy. It seems odd then that those voters would favour MacKay when he has been seen by many members as the main candidate people call “Liberal lite” and have spawned hashtags against, like #NeverMacKay.

One thing that severely limits the accuracy of the poll was a lack of balance between members in different ridings, meaning those who were polled are likely disproportionately voters in urban areas where memberships are more common. 

The website iPolitics stated:

Mainstreet asked Conservatives who their first choice, second choice and third choice is for leader of the party, offering each candidate as an answer. The poll was not directly representative of the makeup of Conservative leadership voters per riding and excluded the ridings in each of the territories.

Even if the poll was equally balanced among all the different ridings, excluding the territories, then with 5,267 members polled, only 15.7 people would have been polled in each riding. 

When polling 335 ridings, 5,267 respondents is not actually that high, especially considering cities like Toronto will be polled above the 15.7 average due to higher membership numbers. Most rural riding will have had less than 15.7 people deciding where their riding’s 100 points go, which doesn’t exactly seem like the makings of an accurate poll.

Despite the previously mentioned flaws in the poll the biggest issue with this phone poll is just the simple oversight that the majority of CPC members, at the time of this poll, had not voted in the race.

Around 129,000 CPC members have voted in the leadership race at the time this poll was conducted, but that’s out of a total of 269,469 members, which is only 47.8% of the membership.

The Mainstreet telephone poll never indicated whether or not those they polled had voted or not, so we can only assume that around 47.8% of those polled actually sent in a ballot. Added onto that we can assume that the people who did not vote, likely having not been very engaged in the race are not too familiar with Lewis, and extremely unlikely to know much about Sloan. 

Out of the 52.2% of respondents that likely had not voted it is very likely the vast majority voted either, MacKay, or O’Toole, with maybe a small number voting for Lewis due to increasing positive mainstream media coverage. This is all to say that Sloan’s 11.2% of the vote on the first ballot is likely concentrated in those who actually did send in ballots.

It would be strange to have people not motivated enough to research the candidates and send in a ballot to choose Sloan, someone they likely have never heard of before. This is to say if the poll was reduced to only the 47.8% of people who likely voted then Sloan would probably still have somewhere around 11.2% which would be 23.4% of the vote on the first ballot (11.2/47.8). 

Lewis’ numbers are probably watered down a bit with people who did not vote, but she too would likely be scoring in the low to mid 20’s in a poll adjusted for actual motivated voters. 

This would mean that all four candidates are likely neck and neck, which validates why we are seeing every campaign pushing as hard as possible for every single last vote before the cut off on August 21 at 5pm. 

This lines up with how Mainstreet Research during the 2017 election under-predicted candidates like Brad Trost polling him only at 1-5% (with anything around 5% being the rarity) and then on election day Trost came away with 8.35% on the first ballot.

For those in Western Canada, who could also forget about Mainstreet predicting a landslide victory for Bill Smith over Naheed Nenshi in Calgary’s mayoral race in 2017, that actually turned out to be the complete reverse on election day.

As a general rule with polls, especially for leadership races, is that it is incredibly difficult to poll and predict those who will actually show up and vote. Older voters and newer Canadians are notoriously hard to get a read on, and often those who do answer polls are not actually keeping up with the race they are being polled on.

In something like the CPC leadership race turnout is the key to winning and soft polls like the one from Mainstreet are unlikely to be accurate due to several factors that it is unable to take into account when trying to get a read on motivated voters’ intentions. 

Could the Mainstreet poll still be accurate? Maybe, but highly unlikely.