By the Numbers: What are the chances B.C.'s election results could change after the final count?

If you don't know me, I was a journalist at The Vancouver Sun for 17 years, including many years as a data journalist. I now teach journalism and data visualization at KPU. I've got a lot of experience number crunching election data. But one thing I don't have any more is an editor. So if you notice a mistake in what I've written here, or something isn't clear, please reach out at cskeltondata@gmail.com

As folks in B.C. woke up on Sunday morning, with all the election-night votes counted, the BC NDP had 46 seats, the Conservatives had 45 and the Greens had 2.

A party needs 47 seats to have a majority and, in practice, 48 since the governing party usually needs to appoint the Speaker. So as the results sit right now, the BC NDP could only govern if they had the support of the Greens. The Greens seem unlikely to support the Conservatives and, even if they did, that would only give those two parties a combined 47 seats — not enough to govern if they also needed to appoint a speaker.

But these are only the preliminary results. The final count doesn't take place until this coming weekend. And as Elections BC notes, there are about 49,000 votes still left to count. That's about 500 votes on average in each riding, which is a lot more than the margin of victory in some of the closest races.

(UPDATE: On Oct. 24, Elections BC provided an updated estimate that 65,000 votes remain to be counted. That works out to more like 700 votes on average in each riding rather than 500. Elections BC said the actual number of uncounted votes in each riding will be released on Friday, Oct. 25.)

In particular, the NDP's margin of victory in Juan de Fuca-Malahat is just 20 votes and in Surrey Centre it's just 95 votes. Similarly, the Conservatives are leading in Surey-Guildford by only 103 votes and in Kelowna Centre by just 149 votes. (Note: Some of these results have shifted a little since election night. The chart below is from CBC News)

(Pro-tip: To make any chart or image in this post larger, just click on it.)

Given how close the results are, even one or two seats changing hands could have a huge impact on who governs B.C. — of if we head into another election.

If the NDP picked up just two of the Conservative's seats, they wouldn't need the support of the Greens to govern. And if the Conservatives picked up two of the NDP's, they'd have a majority (though they may not be able to govern since they'd have to appoint a speaker).

If this all sounds familiar, it's because we were in a very similar situation seven years ago. In 2017, no party had a majority on election night: the BC Liberals had 43 seats, the BC NDP had 41 and the Greens had 3. But the NDP held one of those ridings, Courtenay-Comox, by just 9 votes. If the Liberals were able to flip that seats in the final count, they would have (just barely) held onto their majority.

As it happened, though, when the final count was completed, the NDP's lead actually grew to 189 votes.

And it grew for a simple reason: The NDP did better in the ballots that were counted late than those counted on election night.

(I've uploaded all the charts I created in Tableau here, where you can also download the workbook and play around with the data yourself if you know how to use Tableau Public. All data for this post comes from this great data file of election results going back to 2005.)

The Greens also did better on ballots counted later. The Conservatives, then a pretty marginal force, saw their vote share go down from 7.6% to 6.8%.

As it turns out, the NDP's performance in the final count shouldn't have come as that much of a surprise. Pretty much in every election going back to 2005, the NDP did better in the ballots counted late — absentee ballots and mail-in ballots — than those counted on election night. The Greens have also usually done better, while the BC Liberals, the main right-of-centre party in B.C. until this year, have done worse.

(I'd take the 2020 numbers — where the NDP did a lot better in the final count — with a grain of salt. More than 30% of people voted by mail in 2020 due to the pandemic and how people felt about the risk of voting in person may well have correlated to which party they voted for. More on this later in the post.)

The NDP bump in the final count is also pretty consistent. In 2017, the NDP did better in votes counted late than those on election night count in 73 of 87 ridings — sometimes by just a bit, sometimes by as much as 10 percentage points. In only 14 ridings was their share of the vote lower in the final count than on election night. And things were even worse for the BC Liberals. There were only four ridings — North Coast, Peace River South, Powell River-Sunshine Coast and West Vancouver-Sea To Sky — where the Liberals' vote share was higher in the final count than on election night. And those Liberal gains were marginal. In all other ridings, the BC Liberals did worse in the final count.  (Yes, that means there were some ridings where both the NDP and the Liberals did worse in the final count than on election night. That's usually because some third party, like the Greens, did better.)

Why does the NDP so consistently improve on its vote share in the final count? I've heard a few possible explanations. One is that a fair number of absentee voters — those voting in a district other than their own — are university students. Unionized workers with irregular schedules, like nurses, might also be more likely to vote outside of their district. And those who like to travel could be more likely to ask for a mail-in ballot. It's not hard to imagine how some of these groups might be more left-leaning and so be more likely to vote NDP.

If you're sitting in NDP HQ, you might be looking at these numbers and feel optimistic that, when the final count is in, the NDP could pick off a couple more seats in places like Surrey-Guildford and Kelowna Centre, where the Conservative lead is less than 150 votes. After all, in 2017 the NDP gained 180 extra votes in the final count in Courtenay-Comox. How hard could it be to pull that off again?

Pretty hard, actually.

The main reason for that is there are a lot fewer votes left to count this time than there were in previous elections. Even putting 2020's crazy pandemic numbers aside, in both 2013 and 2017, there were more than 170,000 votes left to be counted during the final count. In contrast, this time there are only about 49,000.

As Elections BC spokesman Andrew Watson has explained, in previous elections, about 10% of all votes weren't counted until the final count. In 2024, it's only about 2%.

This is mainly due to the new technology that Elections BC adopted in this year's election. In previous elections, if someone voted absentee in a district other than their own, their vote couldn't be counted until Elections BC manually checked that the person hadn't voted in their home district. This meant their vote wasn't counted until the final count. But this year, Elections BC was able to check the master list of voters online from most polling stations, allowing them to count almost all absentee votes on election night.

In a brief phone chat, Andrew Watson also told me that, in the past, all mail-in ballots were held until the final count. But this year all mail-in ballots that arrived before the last day of advance voting were included in the election-night results.

With so few votes left to count, overcoming even a small gap in the election-night count would require a big win in those few remaining votes.

Let's take Surrey-Guildford as an example — perhaps the NDP's best shot at grabbing another seat, given the deficit there is just 103 votes. Each party won about 47% of the vote in that riding on election night.

If there are 500 votes left to count in Surrey-Guildford (it could be more, it could be less), that means the NDP would need to get about 57% of the remaining votes and the Conservatives only 37% for the NDP to pick up 104 more votes (287 vs. 183).  That's a tall order: It would require the NDP to do 10 percentage points better in the final count, and the Conservatives 10 percentage points worse than they did on an election night. That's a lot bigger gap than the NDP's usual advantage in the final count. There have been some ridings where the swing in the final count is that big, but they're pretty rare.

(UPDATE: Now that Elections BC says there are more like 65,000 votes remaining to be counted, about 700 per riding, the math changes a bit. In Surrey-Guildford, that would mean the NDP would only need about 54% of the remaining votes to the Conservatives 40% to pick up another 104 votes (381 vs. 277). That's 7 percentage points better for the NDP and 7 points worse for the Conservatives. Or, put another way, just a 14-point gap rather than a 20-point one. That's still significantly higher than the average shift we've seen in final count. But it gets more in the realm of the possible.)

The Conservatives have a smaller hurdle to clear to pick up an extra seat: Just 20 votes in Juan de Fuca - Malahat. But given that the right-of-centre BC Liberals consistently underperformed in the final vote count in past elections, it's reasonable to expect the Conservatives will, too.

I think the most likely outcome is that the NDP will narrow the gap in some of these close ridings but that none will change hands after the final count is complete.

But there is one (small) glimmer of hope for the NDP.

Given the new technology Elections BC used in 2024, there are likely going to be very few absentee ballots in the final vote count. Indeed, Elections BC says the only ones will basically be from what it calls "non-technology voting places": voting places so remote they weren't able to check the master list of voters online on election night.

That means almost all of those 49,000 uncounted ballots are mail-in ballots that arrived in the last few days — after the advance voting closed but before 8 p.m. on Oct. 19. 

Interestingly, that suggests that while there are a lot fewer mail-in ballots in 2024 than in the 2020 pandemic election, there are considerably more mail-in ballots than there were in 2013 or 2017, when there were only about 6,000. Indeed, in an email, Election BC's Andrew Watson told me the agency received roughly 112,000 requests for mail ballots in 2024.

And that 49,000 figure doesn't even include those mail-in ballots that came in before the last day of advance voting and so were included in the election-night count.

Why such a huge increase in mail-in ballots this year? It's hard to say. But I suspect at least some voters who voted by mail during the 2020 pandemic election found it more convenient than they expected and so decided to vote the same way this time. I know at least one relative who fits that description.

If that's the case, that could be good news for the NDP. Here's a comparison of election night vs. mail-in ballots over the past few elections:

The NDP did a lot better in mail-in ballots in 2020 than any previous election — 53%, compared to just 45% of votes on election night. In contrast, the BC Liberals only won 30% of the mail-in ballots compared to 35% of those counted on election night. As I mentioned above, this is probably because whether people were willing to vote in-person or not in 2020 may have also correlated with which party they were likely to support.

Those particular pandemic circumstances no longer exist in 2024, obviously.

But it is possible that the 2020 election might have made some voters more comfortable with voting by mail — behaviour they then carried over to how they decided to vote in the 2024 election. And there were significantly more NDP voters who had that mail-in voting experience in 2020 than supporters of other parties.

Could that mean mail-in voters in 2024 are more likely to look like the heavily NDP mail-in voters of 2020 than those in 2017 or 2013? It's possible. Would there be enough more of them to close the gap in places like Surrey-Guildford and Kelowna Centre? I'm doubtful. But if you're an NDP supporter, it's the one place you might look for hope of an upset.

P.S. One thing this post doesn't address is the recounts that will take place in at least a couple of ridings. One reason for that is I that I have trouble believing a recount could change the result. Recounts are usually done on the assumption that there was some sort of error in the original count — and a slower, more careful recount might catch that mistake. But the counts this year were done by machine. The machines might not be perfect but I have to imagine that Elections BC did a lot of testing on them (or reviewed the testing of others) before adopting them. My gut says those machines are probably more accurate than pretty much any hand count. Now, weirdly, the recounts by law have to be hand counts, not just a do-over of the machine count. This makes me think if the results do change, it's actually more likely it will because the recount was less accurate than the election-night machine count. How that would all play out is an open question. But my bottom line assumption here is that the hand recounts will come up with basically the same result in ridings as the election-night machine counts.