US-Canada County Maps Thread

[RI]

I like to imagine the US and Canada as one country sometimes. Here are some maps of demographics, etc. (whatever I feel like making) that look at the North American subcontinent at the county/census division level. I think they're quite interesting and beautiful, and they show the continuity between the two countries for comparisons.

Population Density (in people/sq. miles):



Data for Canada is from the 2011 Census. Data for US is from 2012 population estimates.

[RI]

This is one of my favorites: a religion map. The data for Canada is from the 2001 Census, so it might be a bit old, but not too much. The data for the US is the most recent ARDA data with some adjustments I made: I took the adherents data and compared it to the ARIS survey of religious self-identification. I found that the data matched up nationally for Catholics and minor religions but not Protestants due to the about 14% of the population that identifies as generically Christian but doesn't attend any one specific church. So I adjusted for these "invisible Christians" to ARIS levels, and divided the new total adherents by the total adult population and fixed a couple other minor issues to get this map:



It's worth pointing out that "no religion" does not necessarily equal atheist or agnostic.

[RI]

Yes, there are some rather cool maps coming. One in particular is requiring a long methodology but should be quite interesting.

Cool maps. Hopefully you're using a map-making program and not Paint.

This is one of my favorites: a religion map. The data for Canada is from the 2001 Census, so it might be a bit old, but not too much. The data for the US is the most recent ARDA data with some adjustments I made: I took the adherents data and compared it to the ARIS survey of religious self-identification. I found that the data matched up nationally for Catholics and minor religions but not Protestants due to the about 14% of the population that identifies as generically Christian but doesn't attend any one specific church. So I adjusted for these "invisible Christians" to ARIS levels, and divided the new total adherents by the total adult population and fixed a couple other minor issues to get this map:

Would it be possible for you to post an example of your methodology? Maybe with a WA county (King?) so I can better follow along.

I'm surprised Franklin/Adams aren't Catholic

Yes, I can do that later.

Lake and Peninsula Borough in Alaska intrigues me. Why is it so overwhelmingly Orthodox? It's Wikipedia article says nothing on the topic. (And I thought there might be parts of, say, rural Saskatchewan which might be quite Orthodox as well).

Russian Orthodox churches are still relatively strong along the south Alaskan coast from the 1700s when the Russian Empire colonized the region. I was surprised by how strongly the data showed them though.

Awesome template.

One caveat about the Canadian religious figures, I believe StatsCan lumps Mormons into Protestants for some reason. Not that they have a majority in any census division, but it would affect the %'s perhaps in at least one CD in Alberta.

Yeah, that frustrated me, but I haven't been able to find any geographic breakdown of Mormonism in Canada beyond "they're strong in Alberta."

[RI]

Would it be possible for you to post an example of your methodology? Maybe with a WA county (King?) so I can better follow along.

I'm surprised Franklin/Adams aren't Catholic

One of the big reasons, though not the only one, that Franklin/Adams counties aren't Catholic is that they have really strong Mormon presences, which is part of the reason that Romney did so well there in the 2012 caucuses (Adams was Romney's best county) and in the surrounding counties. Since Washington our home state, I'll give the top 10 counties for some of the major groups by my methodology:

Protestants
1. Lincoln County - 58.37%
2. Wahkiakum County - 54.04%
3. Walla Walla County - 51.46% (Adventist haven of College Place)
4. Kitsap County - 50.67%
5. Cowlitz County - 50.49%
6. Columbia County - 50.11%
7. Island County - 50.00%
8. Whatcom County - 49.09% (Dutch Reformed territory near Lynden)
9. Pierce County - 49.08%
10. Chelan County - 48.81%

Catholics - Generally Hispanics and Natives
1. Garfield County - 43.77%
2. Yakima County - 36.19%
3. Douglas County - 29.05%
4. Chelan County - 25.36%
5. Grant County - 23.96%
6. San Juan County - 22.90% ()
7. Skagit County - 21.93%
8. Benton County - 20.75%
9. Ferry County - 19.77%
10. King County - 18.48%

Mormons - Eastern Washington mostly
1. Adams County - 19.19%
2. Franklin County - 13.94%
3. Grant County - 12.50%
4. Benton County - 11.19%
5. Columbia County - 8.92%
6. Asotin County - 8.70%
7. Lewis County - 7.58%
8. Spokane County - 6.91%
9. Clark County - 6.70%
10. Wahkiakum County - 6.64%

No Religion - Here be libertarians
1. Skamania County - 45.42%
2. Pend Oreille County - 44.34%
3. Island County - 41.04%
4. Jefferson County - 41.04%
5. Ferry County - 38.66%
6. Columbia County - 38.39%
7. Kitsap County - 38.38%
8. Mason County - 37.53%
9. Stevens County - 37.22%
10. Clallam County - 37.22%

To give a sample of my methodology for King County:
ARDA data
Protestants 316,359
Catholics 278,340
Mormons 56,985
Jews 14,925
Orthodox 12,257

Total Adherents: 726,669
Total Adult Population, 2011: 1,506,075

In order to estimate the number of unaccounted for Christians (as ARIS data, which is more complete, and Pew suggest that the true percentage of Protestants is ~48%, Catholics ~26%, None ~20%) I took (Adult Pop_i - Adherents_i)*((ARIS Protestants_Tot - ARDA Protestants_Tot)/(Adult Pop_Tot - Adherents_Tot)) = ~0.421*(Unallocated_i). Unfortunately, this assumes that all counties have the same proportion of "invisible Protestants", which isn't true, but I couldn't come up with a better way to do it, plus the errors sum out to even in the end for the entire country.

For King County:
Total Unallocated: 779,406
Appx 42% of this: 324,233 "invisible Protestants"

Total King County Protestants: 640,592 = 42.53% of Adult Population

All other groups reported in line with ARIS once this adjustment was made, so they were not adjusted from ARDA numbers.

[RI]

Racial/Ethnic Plurality:

[RI]

What's the island in the upper left inset?  Guam?

Yes.

[RI]

Here's one that I did just for fun and because I thought the topic was interesting. I projected how Canada might have voted in the 2012 US Presidential election based solely upon demographic data. I first ran regressions for Obama and Romney vote shares using US data, created the formulae with the highest R²s (both ended up just over 0.7), then used that formula on the data for Canada's census divisions. No consideration for actual Canadian voting patterns was used. I could probably have gotten more accurate results if there was more of the same data across both countries, but I was somewhat limited in that way. If anyone is interested, I can provide the regressions.

Here is the map:



I ran an additional regression for turnout rates (also had about a 0.7 R²) and used this to tabulate how each province/territory voted in my model:

Alberta: Romney 56.75%, Obama 41.15%
British Columbia: Obama 64.74%, Romney 32.98%
Manitoba: Romney 49.56%, Obama 48.28%
New Brunswick: Obama 52.89%, Romney 45.13%
Newfoundland and Labrador: Romney 68.86%, Obama 28.98%
Northwest Territories: Romney 55.95%, Obama 41.96%
Nova Scotia: Romney 58.48%, Obama 39.40%
Nunavut: Obama 61.67%, Romney 36.30%
Ontario: Obama 55.74%, Romney 42.16%
Prince Edward Island: Romney 60.14%, Obama 37.77%
Quebec: Obama 87.30%, Romney 10.89%
Saskatchewan: Romney 61.60%, Obama 36.27%
Yukon Territory: Romney 60.72%, Obama 37.21%

Canada Total: Obama 62.52%, Romney 35.44%
Canada Excluding Quebec: Obama 52.99%, Romney 44.87%

I also created a secondary model where the French population didn't vote as a block but rather where they voted along their other demographic characteristics. In this version, Quebec voted Obama 50.70%, Romney 47.31% with Canada as a whole voting Obama 52.36%, Romney 45.55%.

Top Obama Census Divisions
1. Quebec, Quebec: Obama 96.58%, Romney 1.68%
2. Montreal, Quebec: Obama 94.72%, Romney 3.27%
3. Longueuil, Quebec: Obama 93.45%, Romney 4.71%
4. Sherbrooke, Quebec: Obama 93.25%, Romney 4.98%
5. Levis, Quebec: Obama 90.76%, Romney 7.54%
6. Francheville, Quebec: Obama 90.57%, Romney 7.69%
7. Joliette, Quebec: Obama 89.31%, Romney 8.95%
8. Gatineau, Quebec: Obama 88.75%, Romney 9.44%
9. La Rivière-du-Nord, Quebec: Obama 88.74%, Romney 9.52%
10. L'Assomption, Quebec: Obama 87.98%, Romney 10.31%

Top Non-Quebec Obama Census Divisions
1. Greater Vancouver, British Columbia: Obama 82.57%, Romney 15.08%
2. Toronto, Ontario: Obama 81.66%, Romney 16.15%
3. Madawaska County, New Brunswick: Obama 80.84%, Romney 17.39%
4. Gloucester County, New Brunswick: Obama 77.61%, Romney 20.59%
5. North East, Manitoba: Obama 73.55%, Romney 24.60%
6. Kent County, New Brunswick: Obama 70.84%, Romney 27.26%
7. Keewatin, Nunavut: Obama 70.59%, Romney 27.47%
8. Prescott-Russell County, Ontario: Obama 69.26%, Romney 28.92%
9. Peel, Ontario: Obama 68.79%, Romney 29.06%
10. Kitikmeot, Nunavut: Obama 66.26%, Romney 31.77%

Top French-minority White-majority Obama Census Divisions
1. Greater Vancouver, British Columbia: Obama 82.57%, Romney 15.08%
2. Toronto, Ontario: Obama 81.66%, Romney 16.15%
3. Ottawa, Ontario: Obama 61.73%, Romney 36.25%
4. Westmorland County, New Brunswick: Obama 60.69%, Romney 37.38%
5. Winnipeg, Ontario: Obama 60.66%, Romney 37.17%
6. York, Ontario: Obama 59.44%, Romney 38.44%
7. Capital, British Columbia: Obama 57.42%, Romney 40.42%
8. Victoria County, New Brunswick: Obama 54.69%, Romney 43.31%
9. Greater Sudbury, Ontario: Obama 53.82%, Romney 44.19%
10. Nanaimo, British Columbia: Obama 51.39%, Romney 46.45%

Top Romney Census Divisions
1. Division 16, Alberta: Romney 95.04%, Obama 2.93%
2. Division 10, Newfoundland and Labrador: Romney 85.17%, Obama 12.42%
3. Division 3, Newfoundland and Labrador: Romney 81.80%, Obama 15.94%
4. Division 9, Newfoundland and Labrador: Romney 80.00%, Obama 17.76%
5. Division 4, Alberta: Romney 79.45%, Obama 18.60%
6. Division 8, Newfoundland and Labrador: Romney 77.62%, Obama 20.17%
7. Northern Rockies, British Columbia: Romney 77.20%, Obama 20.48%
8. Division 7, Newfoundland and Labrador: Romney 77.08%, Obama 20.73%
9. Division 14, Alberta: Romney 75.17%, Obama 22.73%
10. Division 4, Saskatchewan: Romney 74.72%, Obama 23.19%

Top Romney Census Divisions with Over 50,000 votes
1. Thunder Bay District, Ontario: Romney 66.88%, Obama 30.99%
2. Division 2, Alberta: Romney 65.62%, Obama 32.36%
3. Division 1, Newfoundland and Labrador: Romney 62.64%, Obama 35.26%
4. Hastings County, Ontario: Romney 61.73%, Obama 36.20%
5. Simcoe County, Ontario: Romney 59.59%, Obama 38.36%
6. Division 6, Saskatchewan: Romney 58.61%, Obama 39.26%
7. Thompson-Nicola, British Columbia: Romney 57.05%, Obama 40.71%
8. Durham, Ontario: Romney 56.92%, Obama 41.03%
9. Wellington County, Ontario: Romney 56.30%, Obama 41.63%
10. Division 11, Saskatchewan: Romney 56.22%, Obama 41.65%

[RI]

If I can find industry data, I'll update the model. The data needs to be the same for the US and Canada for it to work though. Atlantic Canada did seem odd, but the model had it being so Romney friendly due to low education, low incomes, low French population, low population densities, and a dearth of minorities.

[RI]

If I can find industry data, I'll update the model. The data needs to be the same for the US and Canada for it to work though. Atlantic Canada did seem odd, but the model had it being so Romney friendly due to low education, low incomes, low French population, low population densities, and a dearth of minorities.

Industry might work, but you could also do comparison of demographics within identifiable regions. The demographics of Atlantic Canada are basically identical to those of Maine (and similar demographics elsewhere in New England also vote solidly Democratic).

I did this already, though with the Northeast not New England specifically.

Also, I noticed your map excludes CD #11 (Nunatsiavut) in NL, which is basically just Inuit- which I suppose you'd have going Obama. So, there should be some red in Labrador (not all of it though)

Unfortunately Division 11 didn't exist in 2001 when the latest religion data is from.

[RI]

So I know exactly what I'm looking for, what are the key industries in Atlantic Canada, Ontario, etc., that you believe would impact the projections?

[RI]

The first map might be a good indication of how Canada might vote had it been annexed by the USA long ago.

Yes, that's my goal. I don't think a blood red Canada map would be super interesting.

[RI]

It would actually be very interesting to see a map like this of the United States based on how it would have voted according to a national demographic calibration. The differences from the actual county results in the same election would indicate local factors that aren't captured by whatever broad demographic categories were being used.

That would basically be mapping my residuals. I'm adding in some industry data, then we'll see what I get out of that.

[RI]

I'd be interested in seeing a map of the (much more competitive) Quebec results when the Francophone factor is removed from the model.

I'm planning on doing this.

[RI]

I reran my models with NAICS industry data, and I'm afraid that most of your objections to my projections were not impacted greatly, though my R²s increased across the board by about 0.05. Obama's vote share rose in some places and fell in others, and overall Romney actually gained. Obama's vote shares in a lot of the major cities fell (I did feel they were rather high before), though he gained in some other areas.

New province numbers
Alberta: Romney 59.78%, Obama 38.12% (Romney +3.03%)
British Columbia: Obama 60.60%, Romney 37.10% (Romney +4.12%)
Manitoba: Romney 52.10%, Obama 45.72% (Romney +2.54%)
New Brunswick: Obama 53.62%, Romney 44.39% (Obama +0.73%)
Newfoundland and Labrador: Romney 66.54%, Obama 31.31% (Obama +2.33%)
Northwest Territories: Romney 52.46%, Obama 45.45% (Obama +3.49%)
Nova Scotia: Romney 57.59%, Obama 40.31% (Obama +0.91%)
Nunavut: Obama 66.30%, Romney 31.62% (Obama +4.63%)
Ontario: Obama 53.28%, Romney 44.66% (Romney +2.50%)
Prince Edward Island: Romney 58.92%, Obama 39.00% (Obama +1.23%)
Quebec: Obama 85.78%, Romney 12.38% (Romney +1.49%)
Saskatchewan: Romney 64.38%, Obama 33.47% (Romney +2.78%)
Yukon Territory: Romney 59.40%, Obama 38.52% (Obama +1.31%)

I'll have maps later.

[RI]

These maps are absolutely fantastic!!!!

Does anyone have the blank map? It could be extremely useful to me.

I made the map myself:

[RI]

Updated election projection:

[RI]

Here's the projection of Quebec from the updated model without the French block vote. This map is actually a very slight Romney victory, 49.70%-48.28%.

[RI]

I have a couple more variables I can try, but I would rather not include actual election data. However, if you have election results by census division (actual percentages), I am interested in potentially doing a reverse projection: how the US might have voted if it were annexed by Canada long ago.

[RI]

Maybe it would be better to use riding data for a projection onto the US?

[RI]

If anyone is wondering, I'm currently working on taking my second projection and applying it to federal ridings, and shortly thereafter I'll do the reverse projection.

[RI]

If anyone is wondering, I'm currently working on taking my second projection and applying it to federal ridings, and shortly thereafter I'll do the reverse projection.

What data will you be using?

All of the same Census data I used for my census division projections is available for ridings.

[RI]

Projection by riding (with industry data):



Obama wins 174 (99 outside of Quebec), Romney wins 134 (all outside of Quebec).

[RI]

It shouldn't come as much shock what the outcome of my 2011 Canada election projection onto the US is -- a Conservative rout, though mostly by vote splitting. The national popular vote would be 41.12% Conservative, 28.83% Liberal, 25.69% NDP, and 4.36% Green, assuming no BQ vote. The Conservatives would win ~90% of counties nationally.

State map:

Best Liberal states:
1. Connecticut 37.83%
2. Massachusetts 37.55%
3. New Jersey 36.49%
4. Illinois 34.86%
5. California 34.20%
6. Rhode Island 34.13%
7. New York 33.13%
8. Michigan 30.60%
9. Delaware 30.36%
10. Maryland 29.68%

Best NDP states:
1. Maine 36.65%
2. Vermont 34.97%
3. West Virginia 34.57%
4. New Hampshire 32.33%
5. Oregon 31.31%
6. Michigan 30.81%
7. Indiana 30.19%
8. Ohio 29.85%
9. Kentucky 28.95%
10. Colorado 28.36%

Best Green states:
1. Washington 7.24%
2. Oregon 7.14%
3. Alaska 6.46%
4. Maine 6.21%
5. Arizona 6.13%
6. Colorado 6.10%
7. New Hampshire 5.78%
8. Vermont 5.65%
9. Nevada 5.47%
10. Montana 5.46%

Most evenly divided (3-way) states:
1. Michigan (C 34.6, N 30.8, L 30.6)
2. Georgia (C 38.8, L 29.1, N 28.2)
3. New York (C 36.7, L 33.1, N 26.2)
4. Rhode Island (C 34.6, L 34.1, N 26.9)
5. South Carolina (C 40.8, L 27.9, N 27.

[RI]

Thanks for sharing. What makes WA >40% and OR only >30%? Or is it a tiny difference (like 40.2 and 39.7 or something)?

If you know 90% of counties go for the Conservatives, does that mean you've already made a county map...? Given the Conservatives wins of almost everything due to vote-splitting, strength maps for each party could be interesting.

I'll make up some county maps tonight when I have time, including maps for each party; I have all the county results projected already.

WA and OR are projected nearly identically, except that the NDP is about 3% stronger in Oregon, leading to Oregon being 3% less Conservative (WA is 41.96% Con, OR is 38.98% Con). I think some of the difference is in the high Hispanic population of eastern Washington; my model isn't kind to the NDP in Latino areas. Also, Multnomah County, OR is projected in NDP hands whereas King County, WA is projected for the Conservatives.

[RI]

Top counties for each party:

Conservative:
1. Blaine, Nebraska - 95.79%
2. Carter, Nebraska - 94.66%
3. Wheeler, Nebraska - 94.35%
4. Rock, Nebraska - 94.35%
5. Liberty, Montana - 92.11%
6. Grant, Nebraska - 91.80%
7. Petroleum, Montana - 91.66%
8. Loup, Nebraska - 91.41%
9. Keya Paha, Nebraska - 91.12%
10. Powder River, Montana - 89.54%

Liberal:
1. Buffalo, South Dakota - 54.67%
2. Wilcox, Alabama - 53.70%
3. Ziebach, South Dakota - 53.25%
4. Wade Hampton, Alaska - 52.39%
5. New York, New York - 49.27%
6. Clay, Kentucky - 48.89%
7. Allendale, South Carolina - 47.94%
8. Fairfield, Connecticut - 47.07%
9. Hudson, New Jersey - 47.04%
10. Yukon-Koyukuk, Alaska - 46.42%

NDP:
1. Elliott, Kentucky - 64.17%
2. Carter, Kentucky - 59.79%
3. Potter, Pennsylvania - 55.98%
4. Morgan, Kentucky - 55.85%
5. Douglas, Missouri - 55.29%
6. Culberson, Texas - 54.42%
7. Quitman, Georgia - 54.19%
8. Tazewell, Virginia - 53.41%
9. Fulton, New York - 52.25%
10. Johnson, Kentucky - 51.59%

Green:
1. Skagway, Alaska - 19.08%
2. San Juan, Washington - 15.31%
3. Haines, Alaska - 14.15%
4. Clatsop, Oregon - 12.84%
5. Sierra, California - 11.98%
6. Lincoln, Oregon - 11.25%
7. Curry, Oregon - 10.96%
8. Park, Colorado - 9.95%
9. Ouray, Colorado - 9.90%
10. Multnomah, Oregon - 9.83%