Does British Columbia really have an urban-rural divide?
Crunching the numbers on #bcelxn2017
A quick note before we start: I am not a data journalist nor trained in the fine arts of number crunching. This is less an official report and more a back-of-the-napkin set of calculations I made to satisfy my own curiousity, and should be viewed that way. I have sourced all my numbers and information and encourage anyone with a stronger grasp on this sort of thing to take them and run with them, though I would ask that you let me know so I can see what you come up with!
Last night I was walking between the different election parties in downtown Prince George, talking to the winners (the Liberals) and losers (NDP and Greens) about what had happened.
One theme that stood out after speaking to the Liberal winners was the notion that there is an “urban-rural” divide in the province — that Vancouver and Vancouver Island folks are in cities with different concerns than the rest of the province.
I’ve heard this notion echoed by a variety of pundits and politicians over the last 24 hours and so I wanted to dig in to find out if it was true.
Why I care
In the wake of the United States election, there have been any number of think-pieces and mea culpas exploring the “divided States of America.” There’s a notion floating in the air that people in urban areas on the U.S. coasts are out of touch with the needs and reality of Middle America and vice versa, with people in rural areas failing to grasp the new economic realities and diversity found in the east and west.
I’ve been trying to pay attention to whether the same is possible in Canada — if urban and rural areas are heading in divergent directions. An election is as good an opportunity as any to find out if that divide, or even a hint of it, exists.
I live in a city of roughly 80,000 people, an eight hour drive from the next biggest urban centre. By many definitions, this would be a “rural” area. Yet my experience living here is not what I would think of as rural. I bike and bus to get around, eat at a cluster of downtown restaurants, and have never had to leave town for education or medical treatment.
One thing I’ve learned reporting for an audience of communities similar or smaller than this one is that people can have wildly divergent views on what “rural” is. For some, it’s anywhere outside a major metropolitan area. For others, so long as you have neighbours and a grocery store, you’ve got an urban experience.
So I started Googling and discovered Statistics Canada defines a rural area as:
Areas with fewer than 1,000 inhabitants and a population density below 400 people per square kilometre.
This is as good a definition as any, but doesn’t really work when trying to find a rural-urban divide in an election.
The problem is that B.C.’s electoral boundaries commission tries to keep every riding roughly the same size, population-wise, or at least within a similar range. 1,000 people would be too small. So to adjust for that, they clump sparsely populated areas together, often with urban spots, in order to balance things out. The result is a cluster of tiny geographic ridings in densely populated areas to the south and huge spread out ones in the north.
However, if you were to go based on population density alone, you can start to get a better picture. Using the 400 people per square kilometer threshold, 48 of B.C.’s 87 ridings are rural (based on population and geographic data from the Electoral Boundaries Commission 2015 report).
However, I still don’t feel like that fully captures things. You can see that Prince George, Kamloops, Kelowna, Nanaimo and even parts of the Greater Vancouver area would be considered “rural” under this definition, all because their dense populations are being dispersed amongst a larger geographic area that goes outside their population centres.
So I looked up what Stats Canada considers non-rural; ie: urban. It turns out their are two names: census metropolitian areas and census agglomeration.
Census metropolitan areas are, essentially, metro areas: populations of 100,000, with at least half clustered around what could be called a core area. In B.C. this includes what we might generally refer to as Greater Vancouver, Greater Victoria, Greater Kelowna, and Abbotsford-Mission.
Just below that level are census agglomerations which, colloquially, would be what we consider cities: areas with a population of at least 10,000 centred around a core area. B.C.’s census agglomerations, as of 2016, are Campbell River, Chilliwack, Courtenay, Cranbrook, Dawson Creek, Duncan, Fort St. John, Kamloops, Nanaimo, Nelson, Parksville, Penticton, Port Alberni, Prince George, Prince Rupert, Quesnel, Salmon Arm, Squamish, Terrace, Vernon and Williams Lake (full list here).
Crucially, a census agglomeration does not have to follow political boundaries. Instead it takes note of the way people behave so that if an outlying area is strongly integrated with a nearby city it is included- and if a city has a huge gap so that it may as well not be connected, it is not. You can read more about the definitions here.
Dictionary, Census of Population, 2016 - Census metropolitan area (CMA) and census agglomeration…
Definitions of variables used in the 2016 Census
Using these as a starting point I came up with a definition that I feel works for accurately reflecting the nature of different political ridings in the province. But before we get there…
Mapping it out
In order to see what these urban areas would look like overlaid on B.C.’s political boundaries I grabbed two sets of data to put into a single map. From Elections B.C. I took the Electoral Boundaries kmz file and from Statistics Canada I downloaded the census metropolitan and census agglomeration digital boundary file. I then uploaded them both onto a newly MyGeoData.cloud account to see what it would look like.
Unfortunately I lost the area names doing this, but what I got was the darker-red urban areas on a map of electoral ridings. I would have to reference the original maps in order to check some things, but now I could see which areas were completely rural, which were completely urban, and which fell somewhere in between.
My definition is this…
Before I started classifying each riding as either urban or rural I had to decide what that would mean. Here’s what I came up with:
An urban riding is one which is either completely or very nearly completely inside of a census metropolitan or census agglomeration area. If you look at the ridings in the Lower Mainland, for example, you’ll see they are all very much inside this more highly populated areas and are, therefore, urban. 57 of B.C.’s 87 ridings are urban and they are all in the Lower Mainland, southern Vancouver Island and Kelowna- the census metropolitan areas.
A rural riding is pretty much the opposite. These are the ridings that don’t contain a single urban area- no portion of the riding has a population cluster of 10,000+ people. There are actually only four of these in the province: Stikine, Nechako Lakes, Columbia River-Revelstoke and Kootenay West.
The remaining 26 ridings were neither fully urban nor fully rural. Instead they are places that contain one or sometimes two urban areas in a wider geographic area that is more sparsely populated.
I decided I would divide these into either mostly urban or mostly rural. Here’s how I made the decision about which was which. Let’s take a look at B.C.’s two northeast ridings, Peace River North and Peace River South.
At first glance, these look to be fully rural ridings. However, if you zoom in…
You can see that they each have an urban area: Fort St. John and Dawson Creek, respectively. The Fort St. John urban area is more geographically spread out that Dawson Creek’s, encompassing the nearby community of Taylor, as well.
So now I take a look at the population of Peace River North and Peace River South using the data I collected from the Electoral Boundaries Commission: Peace River North is 43,263 and Peace River South is 28,104.
Then I go back to Statistics Canada’s list of census metropolitan areas and census agglomerations to find out how much of these two ridings is in an urban area. According to Stats Canada the Fort St. John census agglomeration has a population of 28,396 while the Dawson Creek one has a population of 12,178.
Since Dawson Creek’s 12,178 is less than half or Peace River South’s 28,104, that riding is rural primary. And because Fort St. John’s 28,396 people are more than half Peace River North’s 43,263 people, it is urban primary.
I use these two as an illustration of one of the weaknesses- or, possibly, strengths- of this overall strategy. Arguably, Peace River North, stretching up to the Yukon border is far more rural than Peace River South, which is closer to larger centers in Prince George and Edmonton. On the other hand, the majority of people living in Peace River North actually live in an urban area, something not reflected if we simply go by population density or geographic position of the riding as a whole.
You could get more granular with this, reflecting how urban and how rural each riding is based on a sliding scale, but I don’t precise enough information about how some of these areas are divided- like the ones that split Prince George and Kamloops- to get into that. So this is what we’ll go with.
Having gathered all that, I plugged the ridings into a Google Spreadsheet with columns for population, population density, the winning party and the riding’s classification as rural, rural primary, urban primary or urban. You can look at the full thing here or just read some of what I found.
Looking at the purely rural ridings, there’s an even split. The NDP have Stikine and Kootenay West and the Liberals have Nechako Lakes and Columbia River-Revelstoke.
What’s interesting to me about this is each party has one rural riding in the northwest and each has one in the Kootenays- no obvious geographic ideological divide.
Expanding out to include the rural primary ridings the Liberals pick up an additional five while the NDP grab three more.
So the idea that the Liberals appeal to rural ridings bears out, but only just and with a rather small set of data.
Here’s where the bulk of the province- both population and political power- sits. In all, 75 ridings are either urban or urban primary.
57 ridings are purely urban. Here’s how they are divided:
18 ridings are urban primary. They are split thusly:
And if you combine urban and urban primary the divide is:
Conclusion and next steps
While it’s true that the purely urban ridings skew NDP, it’s not as if the Liberals have no inroads there. And on the rural end, it’s an even split, albeit between just four ridings.
Where the Liberals come up victorious is the murky middle- the ridings that are neither purely urban nor purely rural, but instead a mix of the two. Of the mixed ridings, the Liberals have 18, the NDP have seven and the Greens have one.
At some point individual voting station data will be released and someone (probably not me) might be able to delve into this further, finding out if urban areas of mixed ridings skewed one way while the rural parts skewed another. One could also try taking my data and digging a little deeper — taking the margins of victory in different ridings, for example, and finding out if the big wins skew urban or rural; or attempting to add more precise categories between “urban primary” and “rural primary” (“strongly urban” for areas that are 70 per cent urban, for example).
In the meantime, I’d point out two things when it comes to any divide in the province.
One- both major political parties got votes in every part of the province, and there are only three ridings where one of them cleared 60 per cent of the popular vote, so everyone is living alongside people who voted differently.
Two if you’re looking for easy explanations about why people vote the way they do, proximity to ocean water seems to be as good a predictor of ideological leanings as whether a riding is rural or not.