In a close election like this one, the polls can cause considerable emotional angst, especially in the week before Election Day. Here’s your guide to avoid freaking out, apart from shutting all of this out completely and avoiding the polls altogether.
Let's start with a summary as of 6 p.m., Tuesday, May 29, 2018. It’s a virtual tie between the NDP and the PCs with the Liberals still in distant third place and continuing to decline in voter support. Because of how the vote is distributed in the 124 ridings, the PCs have an edge in the hunt for the crucial number of seats needed for a majority -- 63.
To determine how things will work out in the final eight days, the poll zealots are looking for evidence as to the impact of the leaders debate on Sunday. The first returns came in this afternoon from Innovative Research. It showed that since they reported their numbers last week the NDP is up by five percentage points to 36%. The Liberals are down four points to 22% and the PCs are down two points to 34%. How much of this is due to the debate is unclear though much of the polling was done after the debate. Mainstreet Research’s Daily Tracking Poll suggests the NDP has gained most in the post-debate period with the PCs losing most. In any event Rule No. 1 applies here: don’t rely on a single poll to draw any conclusions. And who knows, the NDP might have gained more if the debate hadn’t happened.
In the larger picture, seven of the last 10 polls, mostly done before the debate, have the NDP ahead generally by a point or two. One is an exact tie and two have the PCs up slightly. There’s a couple of other straws in the wind so tentatively it’s looking reasonably good for the NDP with a week to go. Not so good for the Liberals – terrible for them in fact. And it’s a time of high anxiety for the PCs whose lead has melted away and is continuing to head in a southerly direction with all hands bailing as hard as they can.
The NDP total vote averages around 37%. It has to get over 40% and build a lead of four or five percentage points over the PCs to be sure of stopping Ford and of forming a Horwath government.
So, I'd like to offer some tips for surviving this final week.
1. Again, do not believe one stand-alone poll. A poll that suddenly proclaims a big lead for one party or another may be completely wrong or be a harbinger of polls to come. Wait and see what the other polls report.
2. Refuse to put any stock in slight variations especially one or two or even three percent swings that change the lead from one party to another. Small swings are likely to be polling noise, the natural imprecision of polls. Actual election results are by definition accurate. Polls are not. Polls published showing results to a decimal point are just proclaiming a precision that can’t exist. When you find a poll with decimal points, round up or down depending on whether it’s your party or theirs. You will feel better. Just don’t bet on decimal points.
3. Two main problems lead to the fuzziness of polls. Pollsters have to guess who will vote. Demographic groups differ sharply. Pollsters place their respondents into groups. They then try to create a duplicate world of who will vote, the same percentage of women, young people, minorities, etc. as will be in the final voting population. Voters cross them up by turning out in unpredictable numbers. Secondly people often guess when they are asked how they vote. Engage a voter for more than 30 seconds and many will contradict themselves. It’s because they are often cross-pressured and unsure. That’s one reason the same company can poll on the same night and get different results. They could even poll the same people and get different results. We’re dealing with human beings here, many of whom have lost their traditional moorings when it comes to voting. That’s why the NDP can come out of the blue and challenge for government.
4. In a close election the “margin of error” or the fundamental imprecision of polling can make all the difference on the big question, who forms government and will it have a majority.
5. Some experts say there is a “shy voter” effect which causes some of those polled to hide their true choice because they are embarrassed. Or perhaps voters will have an underlying caution that in the end on the way to vote will lead the voter to reject the riskier choice. Usually these attitudes aren’t picked up in polling and underestimate the votes going to conservative parties.
The PCs have stumbled badly during the campaign. Doug Ford’s party was in the mid-40% range in March and early April, immediately after the PC leadership election. That would have given Ford a solid majority. Now they are in the mid thirties. That might still give them a majority but it is a very near thing at best. As I have noted in earlier posts, with anything less than a majority for the PCs the politics of the situation will almost certainly leave them in the opposition. The remnants of the Liberals and the one possible Green MPP would not likely support a Ford government. They will either work out a formal agreement with the NDP allowing Andrea Horwath to form a minority government or enter a two or even three party coalition.
Assuming the Liberals, the Greens and the others don’t move much or slide further, the action in the remaining eight days is between the NDP and the PCs. The final numbers coming out of the ballot box the evening of June 7 will be determined by these six factors:
1. Momentum. The big Mo has been with the NDP. Does it continue, slow down, stop or reverse itself? History is no guide. All four possibilities have occurred in the last week of election campaigns that I have followed.
2. The underlying poll numbers offer hope for both sides. The NDP has substantial second place votes particularly among Liberals, who told Mainstreet Research recently that a third of them were either likely or somewhat likely to change their vote by Election Day. Most of those would change to the NDP. The NDP still gets high marks on a range of factors that may be vote determinative. Ford has a high level of solid support, though still some room to slip even more. He has a much more efficient vote i.e. his total provincial vote is better distributed to win seats. As noted he could tie the NDP in votes, or even lose by a few percentage points, and still win a majority. Shades of Trump and the U.S. Electoral College.
3. Turnout. Conventional wisdom, backed by past experience, has it that older voters who in this election favour Ford, are more likely to vote than younger voters who strongly favour the NDP. In a couple of recent elections younger voters have flocked to the polls. In part that may have been a function of sophisticated turn-out-the-vote campaigns. There’s not much evidence of intense GOTV campaigns in this Ontario election.
4. The likelihood of a Liberal incumbency effect. One theory is that in seats held by the Liberals and where the incumbent is seeking reelection they will be able to pitch to anti-Ford voters to vote strategically for them no matter how badly the Liberals are polling across the province. That will enable them to outperform the general voting numbers unearthed by the polls. If it does, will that depress PC or NDP projected seat totals?
5. The instinct of voters to return to the comfort of their traditional electoral habitat.
6. If some voters flirt with change during a campaign and then opt for a more cautious vote in the end, which party is the more cautious vote? Counter to conventional wisdom it may not be the Conservatives, led by Doug Ford. Kathleen Wynne or Andrea Horwath may have a better shot at that type of voter, to the extent they exist in this “change” election.
For my part, I am interested in hearing what Polly has to tell us in this final week. Polly lives in the artificial intelligence world programmed by Advanced Symbolics, an Ottawa AI firm. The firm’s CEO Erin Kelly has been on Steve Paikin’s TVO show telling viewers about Polly’s conclusions to date. Polly apparently analyses social media and has a great track record in calling elections including the last federal election in Canada, Brexit and the Trump election, none of which were well-called by the traditional pollsters.
Polly says that this election is too close to call at this point. She reinforces the polling conclusion that the Liberals are in deep trouble. Her handlers promise that she will tell us before election day who will emerge as the next Premier of Ontario. She adds that Doug Ford lost potentially decisive ground in the final debate especially with his failure to offer a costed program.
It's too soon to start cheering, but all the signs suggest we should keep calm and carry on to June 7th.
Postscript: When I asked John Graham, my son-in-law of 35 years, what he wanted to talk about during what turned out to be our last conversation, he looked at me intently from his bed at Princess Margaret Hospital and replied without hesitation: "The polls." He died after a shockingly short struggle with pancreatic cancer on Friday, May 25th. He was a remarkable guy, really exceptional. John and my daughter Gill lived and worked in Namibia and Ethiopia for over 25 years where they raised my grandchildren, Danielle and Iain. He was the Country Director for Save the Children International in Ethiopia until last June when they embarked on a new stage in their long careers in international development and humanitarian assistance. But that was before he was diagnosed with cancer. You can read all about him (and consider making a donation to the causes he served over his lifetime) here. I loved him very much and miss him already. I will really feel his absence when the returns start coming in on Election Night. But John was an expert in resilience and I owe it to him and his memory to "carry on". This post is dedicated to him.
Originally written on John Brewin On Politics
