If this was a conventional election, it would appear First-Past-the-Post (FPTP) will prevail over Proportional Representation (PR). Is it a conventional election? Let’s take a look.
It appears, according to public polls, that it’s a competitive race between the two options. Various polls have it in the 50-50 range. There is also a consensus among public polls that PR is strongly favoured among young people and FPTP is strongly favoured among older people.
There are regional variations. Areas where PR appears strong are Vancouver Island (where the Greens are strongest) and the City of Vancouver. FPTP appears strongest in BC’s Interior.
Ethnicity and language also appear to play a role. Public polls do not show breakdowns by ethnicity (and I wonder if they are properly represented in sampling), but we can see that turnout has been lower in ridings that have lower English-first language populations, such as Richmond and Surrey.
There has been some interesting analysis conducted by Star Metro reporter David Ball and Langara College professor Bryan Breguet (an unabashed PR supporter) who has a blog called “Too Close to Call”.
Ball tracked the rate-of-return of the mail-in-ballots and compared it to the HST and Translink referendums. The seemingly slow rate of response at the outset of the process was typical of these processes.
Brequet ran regressions and other analysis to determine who might be voting. He’s also conducted Google polls, that show the race very tight. I’m a bit more primitive in my approach so take the following as you will.
Here’s what we know:
1,361,000 ballots were received, and Elections BC has published the riding-by-riding totals for over 92% of those ballots, providing a fairly clear picture of the regional breakdowns.
On the spectrum of turnout for province-wide votes over the past seven years, the PR referendum was low. Take into account population growth and the PR turnout was even weaker when compared to the HST referendum.
One of the wrinkles with looking at these results by region is to look at raw votes rather than ridings. There are 24 seats in the Interior and only 15 on the Island/Sunshine Coast. Yet the average riding population on the Coast is higher. In the final analysis, there are about as many votes from the Island/Sunshine Coast as there are from the Interior.
The Interior received ballots earlier, and returned them earlier. Over the course of the balloting period, the Interior’s share of the overall pile of votes diminished to about its share of registered voters. The FPTP advocates may have hoped that the Interior would punch above its weight in terms of turnout. The region that did over-perform its share of the electoral pie was Vancouver Island / Sunshine Coast. By the time the final 8% of ballots are allocated to ridings, the Island will likely surpass the Interior in terms of overall votes, and will have over 20% more influence on the process than compared to its share of registered voters. The Lower Mainland will have about 8% less influence on the process that its share of registered votes.
We can also track where the votes came by held-seat. In the HST referendum, every NDP seat voted against the HST, while half of the BC Liberal seats voted for the HST. Who represents a seat can be an indicator of support – not because of the MLA, but because of the underlying attitudes that got that party and MLA elected in the first place. One could reasonably assume that BC Liberal seats will lean FPTP and NDP/Green seats will lean PR. That will probably be the case writ-large, though there will be exceptions. (For the purposes of this analysis, I am basing it on held seats as of election night, 2017).
What does this tell us? The three Green seats over-performed on turnout, while the NDP under-performed. NDP held-seats in Surrey, for example, had very low voter turnout. Apparently, voters in Cowichan, Oak Bay, and, Saanich North & the Islands were more eager to receive their ballots than to have a visit from Santa. But there may be other reasons for that, stay tuned (see below). The BC Liberal seats held their own, but no advantage in turnout. Again, those 20 of 24 seats in the Interior held by the BC Liberals did not generate as much in terms of turnout percentage as I would have thought.
In the Lower Mainland, it was very close in terms of raw vote turnout between the NDP-held seats and the BC Liberal-held seats. The NDP’s 26 Lower Mainland seats account for about 50.5% of all votes in the Lower Mainland processed already versus 49.5% for the 22 BC Liberal-held seats. There are no Green seats in the Lower Mainland. So, in the Lower Mainland, the BC Liberal seats did better relative to the NDP.
In the Lower Mainland, the turnout was highest in Vancouver-North Shore and the Fraser Valley, relative to Burnaby to Mission corridor and Richmond to Surrey corridor.
Older people vote
It is a consistent truth in Canadian elections that turnout is higher among older people than younger people. In the 2015 federal election, there was a bigger turnout among all age groups, but the old geezers still outpaced the Millennials.
Let’s look at turnout-by-age in the 2017 provincial general election. Work with me here.
The 18-24s represent 11.1% of the eligible voters (the red bar). However, they represented only 6.3% of those who actually voted (the blue bar). Meanwhile, the 65-74s are busy voting on the first day of the advance polls following their coffee and muffin at Tim Horton’s. This group represents 12.7% of eligible voters (red), but 17.6% of those who voted (blue). There are only about 10% more 65-74s than 18-24s in terms of eligible voters, but almost a 3X difference between them in terms of who voted.
Chart 1: Turnout by Age, 2017 BC provincial election (source: Elections BC)
So, the question is, given that public polls indicate a strong preference among seniors for FPTP, did old voters rule the roost in the PR referendum too?
Bryan Brequet from Too Close to Call has undertaken some analysis and he sees a trend toward more younger voters. I think what he means is that the gap between younger and older voters in the PR referendum may not be as pronounced as the 2017 general election.
I’m not so sure. When you have a lower turnout, the age discrepancy is usually bigger. When you have a higher turnout, more younger people (and other less-likely voters) are showing up to the polls.
So here’s what I did…
I ranked the ridings 1-87 in terms of turnout in the PR referendum. Parksville-Qualicum is #1, at 46% turnout. Surrey Whalley is #87, at 18%. (This is, as of, 92% of ballots being screened. Turnout by riding will increase but the ranking of ridings 1-87 will probably not change that much).
Then I took BC Stats data and looked at the 18-44 population per riding. I ranked the ridings 1-87 in terms of their proportion of 18-44s relative to the overall adult population of the riding.
The green shading indicates the ridings that are in the lowest quartile of 18-44s (the oldest) ,and the red shading indicates ridings that are in the top quartile of 18-44s (the youngest).
What did I find? The top 8 ridings in terms of voter turnout in this referendum are also in the oldest quartile when it comes to age by population. Six of the 8 lowest turnout ridings are in the youngest quartile. In fact, as you go down the list from highest turnout to lowest turnout, you will see 19 oldest-quartile ridings before you hit the first youngest-quartile riding.
The Rosedeer Decision Desk calls it: Older people vote more.
Older People vote more … but what does that mean?
If I’m right, and a similar pattern exists in this referendum as it does in most general elections, then it’s good news for FPTP. But there are definitely some mixed messages.
One of the oldest ridings is Saanich North & the Islands represented by the Greens. Will older people deliver this riding for FPTP or will the Greens deliver it for PR? I think it’s both. I’m sure the Greens will have worked hard to deliver votes there (they get elected there for a reason), but age will be a bit of a mitigating factor. This could be the story of the Island – an inexorable pull to PR by the Greens (and to a lesser extent, the NDP), restrained by a sizeable population of old geezers.
In older, BC Liberal-held Lower Mainland ridings like Delta South, West Vancouver-Capilano and White Rock, you might see some of the largest margins for FPTP.
Even if there are more older people voting than younger people voting, the question is, where is support at for FPTP and PR?
If it is 50-50 in the polls, and if public polls are correct in so much as older people favour FPTP and younger people favour PR, and IF an age turnout factor is present is as above, then 50-50 becomes 47-53 or even 45-55 in favour of FPTP.
However, if PR has burst through, and if it has weakened opposition among older people down the home stretch, and made a breakthrough with 35-54s, then overall support of 55% among eligible voters may translate to just enough (50% +1) among those who actually voted.
As mentioned, some of the Lower Mainland ridings have the lowest turnout of any seats in the province. The Lower Mainland also has the highest share of non-English (first-language) households in the province as well.
Similar to age, there is a correlation between turnout and English-language skills. This table, like the previous, is ranked by turnout. The green shading indicates ridings in the lowest quartile when it comes to non-English households. The red shading indicates ridings in the highest quartile of non-English households.
What does this tell us? We can hypothesize that election materials were not accessible to some voters, or was not debated as extensively in their language (via media) compared to English-speaking media. Unlike turnout-by-age data supplied by Elections BC, we do not have comparable data for ethnicity or language. I can merely point out the correlation. Anecdotally, the HST referendum appeared to have had a high level of engagement in the Chinese community, owing in part to mobilization of Chinese restaurateurs who opposed the tax. It’s fair to say, I think, that the PR referendum did not hold the attention of the Chinese media (or Punjabi media) like the HST issue.
While Chinese voters tend to favour the BC Liberals, historically, and voters from Punjabi-speaking households have leaned more to the NDP over time, the Greens are clearly weaker in these communities relative to support they have elsewhere. Therefore, one could argue that lower turnout in parts of Surrey, Richmond, and other Lower Mainland areas might be good news for PR, if you think you are more likely to find PR voters where the Green Party finds more fertile territory, generally.
The voting is done. I made my case for FPTP (“I was a Teenage Vote Splitter“) and have also pointed out how PR advocates should be careful what they are wishing for.
I think the final result will be close. I think there was momentum toward PR in the final weeks which helped close a (perceived) gap. For PR to win, it will have had to have done quite well with 35-54s (assuming young and old cancel each other), and have won the Lower Mainland (assuming Island and Interior cancel each other).
If PR does win, then it will have likely done so with less than 22 to 23% of registered voters. We can debate the legitimacy of that if it happens. Gordon Campbell had set a 60% threshold which seemed to be a reasonable threshold for a fundamental change to the electoral system. The current government obviously thinks otherwise!
I do not think 42% turnout is enough for PR to win with a majority. I think they needed more voters to flood the polls. I expect FPTP to win by a few whiskers, a few grey whiskers. We will know soon enough.
Merry Christmas and Happy Holidays to PR and FPTP supporters alike !
Well thought out. I agree with the conclusion: FPTP by a grey whisker or two.
Sent from Mail for Windows 10
Great analysis Mike. I’ve done some similar analysis in Excel using the interim ballot counts published by EBC. Using the breakdown of ballots recieved per-riding, I tried to estimate how many of those ballots were cast by people voting yes or no. To do that, I combined partisan vote preference (ex what chance will a liberal voter vote no) as reported in the Angus Reid poll released a couple months ago with the turnout data from the 2017 general election. For the sake of easy numbers, here’s an example of how it looked:
Say a riding in 2017 was won by the liberals with a L/N/G split of 50/40/10. Now let’s assume that the same riding has had 10,000 referendum ballots received. If the proportion of voters remained constant (a huge if I recognize), that means the ballots cast in the referendum would approximate 5000 liberal voters, 4000 NDP voters, and 1000 Green voters. The Angus Reid survey data breaks down the yes/no preference of voters by 2017 vote intention (ex. Liberal voters are approximately 60% no, rising to over 70% if you assume undecideds break according to the ratio of yes/no voters for the same party). This implies that of the 5000 referendum ballots cast by liberal voters, approximately 70% of those will be cast for no and 30% for yes. Doing the same calculations for all three major parties creats estimated yes/no ballot counts for each party in each riding, which can be summed up to estimate total yes/no ballots.
Like any model, I’ve used what I think are logical assumptions but ones that could ultimately (and likely will be) be proven wrong. A major thing I don’t consider is the motivational impact on turnout, which your analysis on age for example does a great job elaborating on. By my model, at 38.4% ballots reported, I had yes squeaking out a win.
Thanks for your analysis Mark!