US with Canadian parties

[Kali Redcoat]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

alrighty, how'd you make yours?

[S019]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

alrighty, how'd you make yours?

So I'll break it down by region, I started by synthesizing what people said in both this thread and the other one, which gave me a starting point. Notably Liberal WV/Appalachia and northern ME, NDP Iron Range and NW WI, and NDP western Washington.

So I started in the Northeast and I modeled the Bos-Wash corridor as an expanded GTA. This meant the Liberal Party would do incredibly well in the urban and suburban parts of the Northeast. The Conservative seats here I think are pretty self explanatory, and also the Conservatives tended do well in parts of the Mid Atlantic outside of the urban corridors (resembling downstate Ontario). So from there I moved to the South, which felt like the logical choice to put a Southern interests party. Southern metros I guess here function as a Montreal analogue, as well as the obvious notion that this iteration of Bloc would be toxic among black voters, so the southern Liberal vote is a mix of urban/suburban voters and black voters. Florida is kind of its own thing, though I assumed due to the stronger position of the Liberals among "Middle Americans" compared to the Democrats they would do a bit better with "Middle American" transplants in places like Volusia and Pinellas leading to them doing a bit better in I-4 overall, the rest of Florida basically votes as it does in our world. In terms of the Midwest, what became clear pretty quickly is the Tories don't really do well in industrial areas, so the gains we have seen the GOP make in these areas are not really going to be replicated. I gave Pittsburgh to the NDP because of its history in the steel industry and it seemed a valid analogue to Hamilton. In the rest of the Midwest, the NDP seats were often college-town heavy (like the downstate IL college town seat, Madison, and Ann Arbor). I also gave them Tlaib's seat as there does seem to be a bit of NDP vote tied to the auto sector (as seen in Windsor), I assumed black voters would be heavily Liberal explaining the other seat. Also as we moved further west, what becomes clear is the Tories do much better in suburban areas than they do in the Laurentian heartland, this means I gave several normally Democratic suburban seats like IL-14, MI-11, and MI-03 to the Tories. Moving even further east, the country gets bluer as these states really seemed good analogues for the prairie provinces. Omaha/Kansas City in particular seem like good analogues for cities like Regina and Saskatoon and would probably have NDP cores but in both cases they are paired with very Tory suburbs. One place where I disagreed with the consensus was Texas. Rather than seeing Colorado as the equivalent of Alberta, I saw Texas as a natural choice. A state that has a strong connection to the oil industry, as such I saw Hispanics in TX working in the energy sector (both along the border and in eastern Houston) as logical analogues for the conservative leaning minorities often seen in Alberta (such as Indians). Houston in fact felt like a natural analogue for Calgary, and the result is a map of Texas that looks more like from the Bush years and is probably the heart of the Tory party. So now moving further into the mountain west, I guess I modeled this as a mix of western provinces. The NDP wins Boulder and would do well in ski country, but the ski country counties have been split multiple directions, I suspect they come second in CO-07 and CO-03 though, they also win the Native seat in NM. The urban centered seats and the AZ VRA seat vote Liberal, and the suburbs of these cities are all very Tory. So the last part of the country is the West Coast, which I chose to model as similar to BC. The wealthier inner suburbs of Seattle vote Liberal while the outer suburbs vote Tory and the progressive core votes NDP, a similar story in Oregon. California's many metros seemed like the perfect parallels for Vancouver, where in 2021, the Liberals did quite well. San Francisco very much resembles the "too rich to vote NDP, too progressive/left-liberal to vote Liberal" demographic and that is reflected though the Emerald Coast seat votes NDP despite a strong Liberal base in Marin. Further down the coast, the Liberals do well in LA though losing a few deprived minority heavy seats, as deprived seats in Vancouver do tend to lean NDP and does well in San Diego and the IE.

Can't disagree with this too much. I think the Staten Island seat and possibly the Jersey seats vote Liberal though. Also my impression of the South is that the style of conservatism there has more in common with Reform than BQ, though I guess it's the least bad equivalent of BQ possible.

I went with Tory for NJ-04 because Orthodox Jews are a very strongly Tory demographic and I don't think the Liberals would do well enough in the rest of Ocean County to outvote Lakewood. Staten Island was a tough call for sure, I did some research and it seemed that the Tories tended to do better in Italian heavy GTA seats compared to those with fewer Italians, but I suspect it would be pretty marginal. NJ-02 is kind of outside the spheres of influence of both Philadelphia and NYC and the Liberals tend to do poorly outside of metropolitan seats. I feel more confident in the NJ seats than the Staten Island ones.

[Kali Redcoat]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

alrighty, how'd you make yours?

So I'll break it down by region, I started by synthesizing what people said in both this thread and the other one, which gave me a starting point. Notably Liberal WV/Appalachia and northern ME, NDP Iron Range and NW WI, and NDP western Washington.

So I started in the Northeast and I modeled the Bos-Wash corridor as an expanded GTA. This meant the Liberal Party would do incredibly well in the urban and suburban parts of the Northeast. The Conservative seats here I think are pretty self explanatory, and also the Conservatives tended do well in parts of the Mid Atlantic outside of the urban corridors (resembling downstate Ontario). So from there I moved to the South, which felt like the logical choice to put a Southern interests party. Southern metros I guess here function as a Montreal analogue, as well as the obvious notion that this iteration of Bloc would be toxic among black voters, so the southern Liberal vote is a mix of urban/suburban voters and black voters. Florida is kind of its own thing, though I assumed due to the stronger position of the Liberals among "Middle Americans" compared to the Democrats they would do a bit better with "Middle American" transplants in places like Volusia and Pinellas leading to them doing a bit better in I-4 overall, the rest of Florida basically votes as it does in our world. In terms of the Midwest, what became clear pretty quickly is the Tories don't really do well in industrial areas, so the gains we have seen the GOP make in these areas are not really going to be replicated. I gave Pittsburgh to the NDP because of its history in the steel industry and it seemed a valid analogue to Hamilton. In the rest of the Midwest, the NDP seats were often college-town heavy (like the downstate IL college town seat, Madison, and Ann Arbor). I also gave them Tlaib's seat as there does seem to be a bit of NDP vote tied to the auto sector (as seen in Windsor), I assumed black voters would be heavily Liberal explaining the other seat. Also as we moved further west, what becomes clear is the Tories do much better in suburban areas than they do in the Laurentian heartland, this means I gave several normally Democratic suburban seats like IL-14, MI-11, and MI-03 to the Tories. Moving even further east, the country gets bluer as these states really seemed good analogues for the prairie provinces. Omaha/Kansas City in particular seem like good analogues for cities like Regina and Saskatoon and would probably have NDP cores but in both cases they are paired with very Tory suburbs. One place where I disagreed with the consensus was Texas. Rather than seeing Colorado as the equivalent of Alberta, I saw Texas as a natural choice. A state that has a strong connection to the oil industry, as such I saw Hispanics in TX working in the energy sector (both along the border and in eastern Houston) as logical analogues for the conservative leaning minorities often seen in Alberta (such as Indians). Houston in fact felt like a natural analogue for Calgary, and the result is a map of Texas that looks more like from the Bush years and is probably the heart of the Tory party. So now moving further into the mountain west, I guess I modeled this as a mix of western provinces. The NDP wins Boulder and would do well in ski country, but the ski country counties have been split multiple directions, I suspect they come second in CO-07 and CO-03 though, they also win the Native seat in NM. The urban centered seats and the AZ VRA seat vote Liberal, and the suburbs of these cities are all very Tory. So the last part of the country is the West Coast, which I chose to model as similar to BC. The wealthier inner suburbs of Seattle vote Liberal while the outer suburbs vote Tory and the progressive core votes NDP, a similar story in Oregon. California's many metros seemed like the perfect parallels for Vancouver, where in 2021, the Liberals did quite well. San Francisco very much resembles the "too rich to vote NDP, too progressive/left-liberal to vote Liberal" demographic and that is reflected though the Emerald Coast seat votes NDP despite a strong Liberal base in Marin. Further down the coast, the Liberals do well in LA though losing a few deprived minority heavy seats, as deprived seats in Vancouver do tend to lean NDP and does well in San Diego and the IE.

Can't disagree with this too much. I think the Staten Island seat and possibly the Jersey seats vote Liberal though. Also my impression of the South is that the style of conservatism there has more in common with Reform than BQ, though I guess it's the least bad equivalent of BQ possible.

I went with Tory for NJ-04 because Orthodox Jews are a very strongly Tory demographic and I don't think the Liberals would do well enough in the rest of Ocean County to outvote Lakewood. Staten Island was a tough call for sure, I did some research and it seemed that the Tories tended to do better in Italian heavy GTA seats compared to those with fewer Italians, but I suspect it would be pretty marginal. NJ-02 is kind of outside the spheres of influence of both Philadelphia and NYC and the Liberals tend to do poorly outside of metropolitan seats. I feel more confident in the NJ seats than the Staten Island ones.

So when I ran my statistical model, I actually saw a *slightly* negative correlation between industrial areas and voting NDP and a negligible relationship with college education. The Tories had a positive relationship with the vote in industrial areas and the Liberals had a negative one, while the NDP was also negligible. There was indeed a correlation between Jewish populations and voting for the Tories but the trend-line was almost flat. Mind you, this is a multiple regression, so all of the variables are impacted by every other variable. For instance, there was a negative correlation between identifying as Christian and voting Tory in my map until I added the variable of identifying as Catholic as well, in which case the negative correlation for Christians in general became positive while becoming more strongly negative with Catholics. And mind you this model excluded every Quebec riding. I neglected to add a variable for French speakers so maybe this correlation has more to do with French Catholics than Anglo Catholics. I don't quite know why my model consistently showed a negative correlation between the NDP popular vote and the vote in industrial areas, but it was actually one of the strongest correlations in my analysis, though it was so confusing to me that I assumed there was something wrong and removed it when doing the analysis for the US since I assumed the NDP would be a good fit for the Rust Belt.

Now of course, if I was enough of a masochist, I would have split up the U.S. into similar regions to Canada and done a similar analysis to you. The prairie provinces correlating with the prairie states and Midwest, the Laurentian Heartland with the Northeast Corridor, West Coast with BC, and then maybe the South with the Maritimes (while I think it's funny, there probably wouldn't be a "Southern Block" like the Bloc Quebecois like with Quebec). My model was a relatively good fit for the Canadian provinces (r-squared of 0.7 for the major parties) but the residuals showed very obvious signs of regional factors overpowering things. The Liberals overperformed in the Maritimes, the Conservatives in Western Canada, etc. Meanwhile the fit for the GOP and Dems were both something like 0.85. As I said already, there is a flaw in my model that using this reverse model on Canada had the GOP win in a landslide, but I still believe that it's still more accurate vice versa.

EDIT: As someone who has experience with Staten Island, I think it'll be a Tory voting area. PS what's the seat count for each party?

[Hatman 🍁]

The problem with trying to do a demographic model is that it ignores the fact that partisan support in Canada is very regionalized.  If there's a negative correlation between industry and NDP vote, it might be because certain regions of the country are bringing those numbers down.

You really have to conduct this exercise more in a qualitative sense.

Anyway, changing the subject to the Bloc here, I think a better analogy would be to have a Latino Bloc party. The political and cultural similarities between French Canadians and Hispanics is much greater than the Quebec and the South.

[King of Kensington]

Yes there's no American "Calgary" for example.  The federal Conservatives get about 10 points higher there than the provincial UCP.  Alberta would probably vote more like Texas than Wyoming or the Dakotas in a US election.

[Continential]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

I really can't see PA-14 voting Liberal.

For PA-14, the suburban areas that exist in them (Peters, Canonsburg) are very Trumpy and attract the type of people who want to live in an area with low taxes. Most of the rest of Washington County is fairly exurban/suburban. The areas around the Monongahela are fairly economically depressed areas which were once Democratic a decade ago but are Republican and the area is fairly economically depressed. The parts of Westmoreland (and Indiana) in the district, aside from Greensburg/Latrobe/Indiana, are rural with parts being exurban.

[S019]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

I really can't see PA-14 voting Liberal.

For PA-14, the suburban areas that exist in them (Peters, Canonsburg) are very Trumpy and attract the type of people who want to live in an area with low taxes. Most of the rest of Washington County is fairly exurban/suburban. The areas around the Monongahela are fairly economically depressed areas which were once Democratic a decade ago but are Republican and the area is fairly economically depressed. The parts of Westmoreland (and Indiana) in the district, aside from Greensburg/Latrobe/Indiana, are rural with parts being exurban.

I gave PA-14 to them on the basis that the Liberals would likely do well in Appalachia and many of the Pittsburgh exurb counties did vote Democratic as recently as 2004, but it is possible the area has materially changed since then, so I could be wrong. I had always imagined the stereotypical well off suburb of Pittsburgh being Butler County but it’s entirely possible Westmoreland and the rurals could outvote Greene, Washington, etc.

The problem with trying to do a demographic model is that it ignores the fact that partisan support in Canada is very regionalized.  If there's a negative correlation between industry and NDP vote, it might be because certain regions of the country are bringing those numbers down.

You really have to conduct this exercise more in a qualitative sense.

Anyway, changing the subject to the Bloc here, I think a better analogy would be to have a Latino Bloc party. The political and cultural similarities between French Canadians and Hispanics is much greater than the Quebec and the South.

The best idea imo would just be for the Bloc to not exist and the Tories to have a similar coalition in the Deep South to the GOP IRL, but I found the idea of a regional party to be interesting and US History means it really only works in the South, but obviously there is nowhere in the US that speaks a totally different language with a culture totally alien to the rest of the country,

[Kali Redcoat]

The problem with trying to do a demographic model is that it ignores the fact that partisan support in Canada is very regionalized.  If there's a negative correlation between industry and NDP vote, it might be because certain regions of the country are bringing those numbers down.

You really have to conduct this exercise more in a qualitative sense.

Anyway, changing the subject to the Bloc here, I think a better analogy would be to have a Latino Bloc party. The political and cultural similarities between French Canadians and Hispanics is much greater than the Quebec and the South.

Yep I almost immediately recognized that the residuals between the predicted values and the actual values of my model in the Canadian dataset were almost entirely correlated with regional areas, i.e. demographic factors showed that both the NDP and Conservatives should be doing better in the Maritimes than they actually would, while downtown Calgary is not the Liberal stronghold my model predicts it to be. Given the regionalized nature of Canadian politics, which I could viscerally see from looking at where my model had diverged from reality, I think a more realistic result would come from either the method used in the other map of coloring in parties based on gut instinct, or what I suggested before with splitting the US into regions which correspond better to regions of Canada. An alternative would also be to look at opinion polls of how Canadians would vote in the 2020 election, with crosstabs by party, and then using that to reverse-engineer the results. Though given that a poll I found seems to show that 80% of Canadians would've voted for Biden, this would imply the Tories having a landslide that blows out my map. So possibly no dice there.

[King of Kensington]

If you matched neighboring provinces/regions with US divisions you'd get something like this:

New England : Atlantic
Mid-Atlantic : Quebec
East North Central : Ontario
West North Central: Manitoba/Saskatchewan
Mountain : Alberta
Pacific : British Columbia

The lineup seems reasonable (or at least somewhat reasonable) in Western Canada, but not so much in Eastern Canada.  The Northeast - an old, low-growth, high-wealth region lines up with old, low-growth, low-wealth regions (not to mention Quebec is a nation within a nation).  Ontario meanwhile is the closest analogue to Great Lakes region, yet plays an outsized role in Canada and makes up 40% of the population.

[Kali Redcoat]

If you matched neighboring provinces/regions with US divisions you'd get something like this:

New England : Atlantic
Mid-Atlantic : Quebec
East North Central : Ontario
West North Central: Manitoba/Saskatchewan
Mountain : Alberta
Pacific : British Columbia

The lineup seems reasonable (or at least somewhat reasonable) in Western Canada, but not so much in Eastern Canada.  The Northeast - an old, low-growth, high-wealth region lines up with old, low-growth, low-wealth regions (not to mention Quebec is a nation within a nation).  Ontario meanwhile is the closest analogue to Great Lakes region, yet plays an outsized role in Canada and makes up 40% of the population.

I feel like you could split Ontario further. Separate out Northern Ontario and then maybe the Greater Toronto area. Obviously is that there is no easy way to make an American analogy to Ontario but maybe you could approximate that by splitting things up further.

[Hatman 🍁]

Yep.

I don't think you can pair Quebec with anywhere (except with Latinos, if you want to go down that route).

And, Ontario has to be split up.

GTA = NYC or Chicago
SW / Niagara = Midwest
Ottawa = DC
Northern Ontario = Upper Great Lakes

[King of Kensington]

So Ontario is Michigan + Chicago + NOVA

[Hatman 🍁]

So Ontario is Michigan + Chicago + NOVA

Well, Ottawa is nowhere near as wealthy as NOVA. But, probably just as educated, if not more so. It doesn't have the poverty of DC either, though. Is there anywhere in the DC area that is fairly middle class and has very little income disparity?

[King of Kensington]

Silver Spring?

[Kali Redcoat]

Yep.

I don't think you can pair Quebec with anywhere (except with Latinos, if you want to go down that route).

And, Ontario has to be split up.

GTA = NYC or Chicago
SW / Niagara = Midwest
Ottawa = DC
Northern Ontario = Upper Great Lakes

Will take this into consideration if I redo my model. Also, I have a way to treat provinces as a factor of their own but it assumes that the voter bases for the parties in each state are the same (basically taking all of the other factors and adding or subtracting a set amount based on the province). Is this a correct assumption? Or should I manually create different databases for each province or region?

Also I'm considering making the GTA an equivalent to every urban district in the Northeast corridor.

[King of Kensington]

Toronto is one of the two Great Lakes "alpha" cities (along with Chicago) but is also one of the three super-diverse NA metros (along with NYC and L.A.)

[Pouring Rain and Blairing Music]

While the politics would probably be flipped, I feel like Texas or the rest of the South would be a decent analogue for Quebec.

Phoenix and Salt Lake City are probably similar enough to Calgary, with Denver being like Winnipeg. Most of the prairie states could be modeled like Saskatchewan.

California would definitely be like British Columbia, with the Bay Area being like Vancouver Island and LA being Vancouver (City).

The Pacific Northwest and upper New England both have similarities to Atlantic Canada, imo. Patches of NDP strength and a lot of declining mineral extraction/forestry.

I’m tempted to post my own map based on vibes (TM) when I have the chance.

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

This is probably the closest to my thoughts so far.

[Hatman 🍁]

Denver = Winnipeg is nuts. Winnipeg is more similar to Minneapolis. 

I agree with others that Houston is a good analogue for Calgary. SLC is too Mormon influenced.

BC is more like the PNW than anywhere in California, except maybe the northern coast.

[King of Kensington]

Yeah I don't see any connection between Denver and Winnipeg.  Denver and Calgary are much more similar than Denver and Winnipeg.

[Kamala's side hoe]

Since people are posting their maps, I might as well post mine, I am prepared to defend most choices:

This is probably the closest to my thoughts so far.

[King of Kensington]

Educational attainment by province:

https://talkelections.org/FORUM/index.php?topic=573924.0

[Kali Redcoat]

The problem with trying to do a demographic model is that it ignores the fact that partisan support in Canada is very regionalized.  If there's a negative correlation between industry and NDP vote, it might be because certain regions of the country are bringing those numbers down.

You really have to conduct this exercise more in a qualitative sense.

Anyway, changing the subject to the Bloc here, I think a better analogy would be to have a Latino Bloc party. The political and cultural similarities between French Canadians and Hispanics is much greater than the Quebec and the South.

I have started revisiting this scenario, and I actually ran an analysis of the Canadian Election Survey to check whether the connection between the NDP and industry was popular wisdom that no longer held. So I went to the two columns representing whether a person was a union member or not and also what political party the voter voted for in the 2021 election. Out of the survey results where both questions were answered, 17.9% of non-union voters went for the NDP, vs 23.2% of union voters. So at least looking at that one variable, that's a 30% increase. That isn't nothing but it does suggest that there is something with regionalization if the correlation with industry was negative. Maybe. Statistics is hard.

[SnowLabrador]

The Tories would win in a landslide.

[Kali Redcoat]

The Tories would win in a landslide.

Yeah, as in my map.