Alberta Election 2019

[Smid]

I did some work last year to come up with new vote margins following the redistribution, if anyone's interested.

[Smid]

I did some work last year to come up with new vote margins following the redistribution, if anyone's interested.

Definitely interested.

I had to resize it because I'd forgotten the pixel dimension constraints, so the writing needs fixing up, and some of the boundaries aren't as tidy as I'd like, but I might do that down the track.

*Edit* I've tidied it up, so it's a bit neater and more like the maps I normally post (cleaned up the pixels along the boundaries). There is also a blank map version that I've uploaded to the gallery.

The shapefile looks like it's in Eastings and Northings instead of GPS coordinates, which seems odd, but I don't think that created any weirdness... keep an eye out for errors, though, just in case.

The estimated results took anything that is a poll and allocated it to the redistributed riding. I then tallied all votes cast in non-polls (Advance/Mobile/Postal/etc) and proportioned them based on the party's strength in each poll (party's vote in the poll compared to party's vote across the original riding, and then allocated the same proportion of declaration votes to that poll). This is slightly less accurate for the Advance Polls which relate to specific polls, but should be close enough.

There were a few polls that were split between two ridings. For rural polls that were thus split, I typically looked on Google Maps, identified the main township, and allocated the poll to wherever that township was transferred, but in some cases, I guessed what proportion of the poll was transferred into which riding (a few of these were split 50-50 between Calgary-Mountain View and Calgary-Klein).

DC - I can send you the .CSV of estimated party votes via Facebook Messenger, if you'd like. I can also send you a spreadsheet showing the poll transfers (2015 Riding Name, 2015 Poll Number, 2015 Poll Name, 2015 Enrolment, 2017 Riding Name, Transfer Weighting).

[Smid]

Sure that would be great.

Okay, I've just sent that through to you.

[Smid]

I did some work last year to come up with new vote margins following the redistribution, if anyone's interested.

Definitely interested.

Amazing blank map, thank you. Do you know if there's a redistricted blank version of PEI?

Thank you - I haven't been on here in a while, but I used to upload blank maps for Canadian and Australian elections after each redistribution.

Because it's been a few years, I'm not sure of the provinces where I'm behind. I can do PEI, no worries. Presumably they have GIS shapefiles? If not, it will be a little longer, but I suspect they have the shapefiles available for download.

[Smid]

I can't seem to find the redistributed results from 2015, can anyone pass it along to me?

Give me half an hour or so, I'll email you or send it via Messenger.

[Smid]

Amazing blank map, thank you. Do you know if there's a redistricted blank version of PEI?

Done. Blank map in the Gallery, copy displayed in the relevant thread.

[Smid]

I think Nenshi also benefits from low turnout in the more conservative suburbs, although that's more an assumption than an analysis... He also is very charismatic and plain-speaking, just says what he thinks, regardless of whether popular or not/politically correct (I'm thinking specifically of this story). You know, sort of joking about Darwin Awards is not something the typical politician would do, but he's really just saying what people are thinking, so it's kind of refreshing that he's setting aside political correctness. Now that I've typed that, it sounds a bit like a certain US President, however he has a very different tone, obviously.

[Smid]

Looks like Mainstreet has the NDP leading in both Edmonton and Calgary: Alberta is looking more and more like BC; NDP (unified left) vs UCP (unified right).

Edmonton
NDP - 46.4%
UCP - 37.8%
AP - 7.3%
FCP - 3.9%
Green - 2.4%
ALP - 2.2%

Calgary
NDP - 45%
UCP - 43.9%
AP - 3.9%
ALP - 3.9%
FCP - 1.6%
Green - 1.6%

Rest of Alberta
UCP - 65%
NDP - 23.3%
AP - 5.9%
FCP - 2.8%
ALP - 2%

I presume someone here runs leantossup.ca - how does this poll look through the lens of your model?

I can imagine that in Calgary, the NDP will run up their margins in the inner seats (Buffalo, Mountain View, Varsity, Klein, Currie), and the UCP will run up their margins in the Western suburbs (especially South of the Bow, but also in the North-West). The South-East will probably also go UCP, but less strongly so, and then the North-East goes NDP?

Edit: Have any of our number mapped Nenshi's latest victory? I think I did the previous one.

[Smid]

Janet Brown Opinion research shows UCP with a 19 point lead.  It is a CATI which is what Nanos uses who has a good track record, but since it is out of whack with other pollsters I tend to think it is a bit closer, but would be nice if more polls were in the field.  Numbers in Edmonton and rest of Alberta are inline with other pollsters, main difference is in Calgary they show UCP solidly ahead rather than a competitive race.  Also CATI might skew towards older voters as all polls show UCP well ahead amongst older voters but NDP ahead amongst millennials so perhaps that is part of it.  Indeed millennial turnout will be key.  If they show up in droves it should be competitive and Notley might even win, but if they don't I expect Kenney to coast to a pretty easy victory.

UCP 53%
NDP 34%
Alberta Party 8%
Liberals 4%

I think turnout generally will be key... Turnout is frequently low in Alberta, I suspect because there are so few ridings that could go either way.  A by-product of this string of close polling results could be that it motivates suburban conservatives to go and vote because the result is in doubt.

[Smid]

Our internal polls do not show the Liberals winning a seat BTW. The NDP are winning Mountain View, perhaps their only pickup of the night.

My calculations on the redistribution tipped it into the NDP column on 2015 results, because of NDP polls to the North that were transferred in, and Liberal polls in the West that were transferred to Varsity. Of course, if Swann had run in those areas in 2015, it probably would be different, but without him on the ballot, I was guessing the NDP would be favoured to win the seat.

Edit: Actually, my mistake, it's NDP if you treat the PC/WRP separately, however if you amalgamate their votes, it becomes a UCP seat on the redistribution, and therefore classing it as a "pickup" is fair (even though we know that 1+1 =/= 2 in political mergers, treating it that way is probably the fairest way of working out which seats are gains/losses).

[Smid]

Some (most?) of you may have seen these already, but I thought I'd best upload them here as well. These are some poll by poll maps taken from the 2019 Alberta Provincial election results.  

Vote share of winners in each poll, based on the preliminary (end of election night) count. Given that the out-of-riding ballots aren't linked to specific polls, they shouldn't affect these, although any recounts might.

As always, bigger versions in the Gallery (and even bigger versions are available).

Calgary:


Edmonton:


Lethbridge:


Medicine Hat:


High River and Okotoks:


Red Deer:


Grande Prairie:


Fort McMurray:



I'm also working on Election Day Turnout maps (bearing in mind that Advance Poll voters won't show up on that map, except by their absence). There are some weird instances, however, so I don't think I should upload them without closer inspection.

Firstly, some polls have greater than 100% election day turnout (highest is St Albert poll 75, where the 51 voters cast 114 ballots, if I remember correctly - and those are numbers on the Elections Alberta website), there were a handful of others, too, with above 100% turnout. Some of these might be in growth areas, and perhaps the enrollment figures were recorded at the time of the redistribution or something... that could explain it.

Secondly, I think the Fort Mac seats often have very low turnout, whereas it was suggesting quite high turnout. Perhaps that's not wrong, but I want to check more closely first.

I also want to review my handling of split and combined polls, to check I'm not double-counting anything.


Edit: Here is a teaser, but bear in mind that I haven't QA'd these yet, so errors are possible.

Calgary:



Edmonton:



I'll probably also work on some vote by party versions, too, similar to Earl's Alberta Party map above, although I don't think I'll be able to achieve a swing map because that would require me to match pre- and post-redistribution polls to one another.

[Smid]

Polls with more than 100% turnout are usually explained by people registering to vote on election day.

Anyway, looking at the Calgary map, one could easily gerrymander the city for 1-2 more NDP ridings, with a nicer looking map.

Yeah, St Albert poll 75 is up in one corner - Google Maps satellite view had a cluster of small houses or condos - looked like maybe a retirement village or something, just very small houses, and the rest of the poll was mostly streets and fairly recently cleared dirt, so presumably a new housing estate. A similar thing could also be said of the poll the NDP won in High River, as I was discussing with Miles yesterday, so yeah, I was figuring that the high turnout polls are mostly going to be in growth areas where people move in and register on the day (similar things happen here, despite compulsory voting, etc - lowest turnout is in high "churn" electorates near the inner city, where people may have not updated their enrollment after they move out, while highest turnout is in outer suburban areas with high development, because of the large number of people moving in but few people moving out.

Just looking at Calgary-Varsity, it looks like voters in the most conservative part of the riding were most likely to turn out on Election Day, but whether that's because they had higher turnout, or whether it was because the other parts were more likely to vote in Advance polls, I couldn't say.

It would be possible to deliver more Calgary ridings to the NDP by changing the boundaries, but remember that Calgary-Falconridge was quite close, changing those boundaries would make the NDP vote less efficient if the party were to do slightly better in another election (would be accused of packing). Moving the (predominantly NDP) polls North of the Bow River from Calgary-Bow to Calgary-Varsity would probably make sense, though, and may have allowed the NDP to win Varsity (especially if the safe Conservative polls in the NW corner were to drop out to accommodate the new polls from Bow).

[Smid]

I would move McCall and Falconridge  up, as I think it makes more sense to have the Skyview ranch area in McCall.

And yeah, it doesn't make sense that the Montgomery neighbourhood is in Bow and not Varsity.

I'm inclined to agree with you about McCall and Falconridge - I think those changes make sense, however in a good year for the NDP, the current version might give three Calgary ridings (McCall, Falconridge, and perhaps North East, although perhaps the margin in North East is too large for that to make a difference), while in a bad year for the NDP, it only yields one NDP riding (McCall). Given the closeness of Falconridge currently, it probably typically would give two ridings. Moving McCall and Falconridge North probably gives two solid NDP ridings, regardless of how well the NDP performs, reducing NDP vote efficiency in a good year. Some on the left might accuse such riding boundaries of packing. All that said, I think that those changes make sense.

That whole North-Eastern area was split across McCall and Falconridge prior to the redistribution, so I've remapped the poll results with the old boundaries overlaid as the riding boundary layer.



And to complete the set of poll results with different boundaries overlaid, here are Calgary and Edmonton with the current federal riding boundaries.

Calgary:



Edmonton:


and federally:



By the way, I realised that there was a slight error in the maps I posted last night - mapping the polygons didn't map the holes within the polygon, so it over-plotted polls 10 and 11 in Banff-Kananaskis. I've edited the post above to correct the map accordingly. My humble and sincere apologies for this error - I've been mapping in R for a while, but only very basic stuff, I'm just starting to learn a few of the slightly more complex functions and techniques (I learnt a few functions I haven't used before creating these maps over the past couple of days).

[Smid]

How about a map of Banff, Lake Louise, Canmore, and Jasper.  Those are fairly touristy towns so wouldn't be surprised if they went NDP even though UCP won the ridings.  Sort of akin of Blaine County in Idaho and Teton County in Wyoming and sometimes Summit County in Utah.  All very red states and surrounding areas deeply Republican but those usually vote Democrat for that reason.

Those towns tend to be quite non-Conservative (see various federal and provincial elections on election-atlas.ca). I have some family stuff planned, but can do them in a bit.

[Smid]

How about a map of Banff, Lake Louise, Canmore, and Jasper.  Those are fairly touristy towns so wouldn't be surprised if they went NDP even though UCP won the ridings.  Sort of akin of Blaine County in Idaho and Teton County in Wyoming and sometimes Summit County in Utah.  All very red states and surrounding areas deeply Republican but those usually vote Democrat for that reason.

Excellent comparison points.

Here is Banff, Canmore, Cochrane, plus the nearby reservations, but also the rural parts of the riding bordering on Calgary:



Jasper will take me a little longer yet - there are a few towns but large spaces between them. I might need to do a few insets instead, because they aren't especially clear. I'll play around with it and try to get something for you in the next 24 hours or so.

[Smid]

Here is the whole-of-province map (bigger version in gallery, even bigger version available). Obviously less detailed in parts, but happy to zoom in and do detailed maps of any towns of interest (yes, I realise I still haven't done Jasper yet, for Miles). This is now the Official Results, whereas the other maps were all preliminary results. I think Elections Alberta has corrected the two data entry errors in Calgary-East, where the Liberals were "winning" two polls, which were obviously actually NDP votes. Although I think this looks pretty, if you want better detail, obviously head on over to election-atlas.ca, which I think is managed by another poster on here.

 

Also, Miles, that poll in High River we were discussing - looks like that might also have been a typographical error in the preliminary results, as it seems to no longer be a win for the NDP.

[Smid]

Don't know if there is one but I'd be interested to see the results for 2019 on the 2015 boundaries (mostly to see Calgary)

That one's on the previous page (it's listed as 2010 boundaries, as that was when the redistribution occurred, this election was on the 2017 boundaries). It's easy enough to do, though, if you want other areas. I've edited my previous post to include Edmonton on 2010 boundaries.

I've also completed maps for 2015, both "true value" and "merged PC + WRP", with both boundaries from the 2010 redistribution and 2017 redistribution (so basically four sets). Happy to upload them if anyone is interested - if so, would mods prefer I post them here or go all necromancer on the 2015 thread?

By the way, how's your three-member districts map coming along?