Dr. Stewart Prest
Lecturer, UBC Political Science Department
It’s a real challenge deciding whether to vote strategically, but a lot of it comes down to two things: the existence of one candidate that you really don’t like with a good chance to win, the presence of a tolerable alternative candidate you’re confident has the best chance to hold them off.
For voters paying attention it’s straightforward enough to figure out whether there’s a candidate you hate. Party track records and platforms, statements by the leader, and the candidate’s own record and behaviour can all provide indicators as to when there is a candidate or party we really don’t care for in the race. The true hang-up for the would-be strategic voter is figuring out who among your tolerable alternatives has the best chance to win. Who will your fellow voters line up behind?
What can be done about this? The frustrating but honest answer is that there’s no foolproof way around the problem, short of having other candidates drop out of the race. To be sure, there are sources of information that we can look to: polling, both national and regional, and polling aggregators like 338Canada.com. We’ll discuss each of these alternatives.
A poll or survey, simply put, is an attempt to gauge the public’s view on a given issue.It’s not easy to get a representative sample for many reasons—everything from who is likely to answer the phone, to the phrasing of the question, to the array of options given in response can affect the outcome. Good polling firms have strong track records of managing these considerations well, and producing accurate responses.
A poll can be national in scope, or provincial, or even down to the level of riding—the latter are rare, however, because each poll is costly, and it’s that much harder to get a representative sample among such a small population.
That said, local level polling can be quite accurate when done well, though with two caveats. First, even an accurate poll likely has a small sample, leading to less precise estimates. Second, even an accurate poll reflects intentions *at the time of polling* and people might later change their minds.
A couple of riding level polls by Researchco during last month’s elections illustrate both points. In two island ridings, North Island–Powell River and Cowichan–Malahat–Langford, Researchco asked 400 voters, weighted by age and gender (to ensure representativeness), who they would support if the election were held that day.
Both polls were largely accurate within the margin of error to election day results. They were both good polls! More importantly for strategic voters, both polls nailed the relative order, correctly identifying the NDP candidate as the second most popular option in both ridings.
However, after the poll, many people changed their minds—in particular, “undecided voters” made up their minds, and many of them followed the national trend and voted Liberal. In both ridings, the Liberal candidate did better than the poll suggested they would once the undecideds decided. In both ridings, the actual Liberal result was significantly higher than at the time of polling—though in both cases, the Conservative candidate ended up winning.
Simply put, if voters had been interested in strategic voting, local riding polling would have helped identify the second-best option. However, even when that polling was available it seems that undecided voters were looking elsewhere for guidance—we can’t be sure where, but a good guess would be the state of the race nationally, given that so many joined the Liberal bandwagon.
What about a polling aggregator like 338Canada? These sites take individual surveys by different pollsters, and create an average among them—a kind of “poll of polls”. Many go a step further as well, using that average to build a model predicting the vote in each riding in the country. The models are quite sophisticated, attempting to control mathematically for a variety of factors including riding history, demographics, regional variations, and even the presence of star candidates.
Accordingly, at the level of the country as a whole, such exercises are accurate, though even there a small national shift can make a big difference. At the level of individual riding things are less predictable, especially in any riding with an unusual local factor like, say, a prominent Green candidate.
A local candidate—especially one belonging to a small national party like the Greens—will likely be affected by all kinds of idiosyncratic factors that the model can’t capture. Everything from the effectiveness of the local ground campaign, to lingering frustrations over decisions of that local politician, to the vagaries of the weather might tip the scales dramatically one way or another.
And, looking at the actual performance in last month’s election, we can see that while 338 was quite right about the likely winner in Nanaimo–Ladysmith, it was off the mark in predicting who the “best bet” to beat that winner was. In the event, 338 called the Conservative win, but consistently suggested Paul Manly of the Greens would be the second most popular candidate, on track to get about 25% of the vote, based on previous Green strength in the riding.
In reality, he finished a disappointing fourth, behind both the NDP and the Liberal candidates, with about 18% of the vote. To the surprise of many (including the polling aggregator!) the Liberal candidate finished a strong second, despite the party’s relative lack of success in Nanaimo–Ladysmith and indeed across the island.
The strong performance of the Liberal candidate again suggests voters were more focused on the national race than local considerations in casting their ballot in this high-stakes election.
Of course, Liberal domination on the centre-left wasn’t absolute. The NDP held seats in their Vancouver East fortress, for instance, and Elizabeth May once again beat the odds to win in Saanich–Gulf Islands.
So what does it all mean? First, it reminds us that voters vote for many different reasons. Not everyone is a well-informed strategic voter. Second, if you really want to figure out who to vote for strategically, you may want to look at the national trends for guidance, since your local undecided neighbour is probably doing the same.
Ultimately, as in any election, your best bet may be to figure out who you think is the best candidate, and do everything you can to convince your friends, family, and neighbours to vote for them.
Dr. Prest teaches political science as a lecturer with the Department of Political Science.
Made possible by: Covering Canada: Election 2025 Funding
Recent Comments