Saturday, March 30, 2019

Is the UCP really doing that well in Edmonton?

No, and I hope to show why.

First off let me present to you my Alberta Map

It is not done. I need to, for example, add a colour for the Alberta Party, as the Greens are not going to take Calgary Elbow. And it is not showing a projection yet; it is simply showing the basic results from last time. These results have been slightly modified however and don't perfectly match the combined WR+PC vote for the UCP. Sadly when my HDD died in October I lost all my data, but found a datasource I'd made in July. I thus need to reconstruct what I was working on and try to remember where I was in the process.

Regardless I want to introduce you, quickly, to Census Metropolitan Areas.

This is what statistics canada considers a "city". These definitions can seem nonsensical to the observer.

For example the Toronto CMA is often thought of as equal to the "Greater Toronto Area". Despite that, the Toronto CMA does not include Whitby, or Burlington, two areas often included in the GTA, while it does include Orangeville, an area often thought of as too far away and rural to be part of the GTA.

The same is true in Alberta for Edmonton and Calgary. To help, I've added darker boundaries to the map to show where the CMA boundaries are. Note: these are not where the exact boundaries are, the boundaries cut ridings in half often; instead, these are ridings where half or more of the voters are within the CMA.

Calgary does not look so bad but Edmonton is a mess!

Lets get some numbers. Now I caution you again that my numbers are modified already. The PC+WR vote will not equal the UCP vote; but it will come close, within a few percentage points.

For the actual municipalities of Edmonton and Calgary the popular vote balance between the top two parties is 53.6%-37.0% and 40.8%-49.5% respectively. For the CMAs however, we get 50.8%-41.7% ad 39.5%-51.5%. Both become more UCP friendly, Edmonton by a larger margin.

If the UCP and NDP were tied in the Edmonton CMA, the NDP would still have a 7 point lead within the municipality of Edmonton.

Lastly, I want to note the population shares.

In 2016 (the last census) Alberta had a population of 4.067 million.
The Edmonton CMA had 1.321 million and the Calgary CMA had 1.392 million. This left 1.354 million in the "rural areas".

In the same year, Edmonton the municipality had 0.933 million while Calgary had 1.239 million, leaving 1.895 million in "rural areas"

If you use rounding, you get ratios of 1.3-1.4-1.4 for the CMAs, and 0.9-1.2-1.9 for the Municipalities.

So, now that we've established that any poll of the Edmonton CMA will have the UCP higher than in any poll of Edmonton as a Municipality, lets find out if the pollsters are using CMA or Municipal boundaries for their poll "regions". We will look at these 4 polls.

Only ThinkHQ explicitly say, so, how are we going to find out?


Angus Reid has the weights, of 201-214-390, which when rounded, are 2-2-4. This is much closer to our Municipal ratio. Angus Reid has a strong NDP lead in Edmonton.

Ipsos has the weights of 296-311-293, or rounded to 3-3-3, which is much closer to our CMA number, Ipsos has the UCP and NDP neck and neck in Edmonton.

Mainstreet has the weights of 338-474-348 which rounds to 3-5-3, which is... weird to say the least, but more closely matches our CMAs, have Edmonton neck and neck.

ThinkHQ which claims to use the CMAs has weighting of 377-417-401 which rounds to 4-4-4 which does match our CMAs; has the NDP with a strong lead in the Edmonton CMA.

TLDR: The NDP is leading in "Edmonton" if "Edmonton" means "The City" and not "The CMA"

1 comment:

  1. reminder to follow my twitter where you get content like this but more frequently. (also I swear sometimes)