Suburban gains: Demographic Change or Newly Liberal Whites

[Starry Eyed Jagaloon]

Del Taichi recently made the point that:

Biden will not approach anywhere near 60% in Cobb or North Fulton, even though these counties are both becoming increasingly diverse (South Cobb is essentially an extension of West Atlanta/Douglas, it's over 60% Black.)  He may break 60% in Gwinnett only because it's the poorest, most downscale of Atlanta's northern suburban counties.  There really aren't these Romney-Clinton voters you obsess over as much as there are Romney-Clinton places.  

Also:  how come when talking about Georgia's suburban Romney-Clinton counties no one seems to give any love to South Atlanta's Henry County (even though it's the most Democratic of any of them)?  Might it have something to do with it only being 55% White and 36% college-educated?  Doesn't quite fit your narrative I guess, hmm?

Naturally, minority voters are the base of the Democratic party, particularly in southern metros like Atlanta. However, it is interesting to consider whether the county map of these big metro areas is shifting because of the white vote or the suburbanization of minorities. It's also interesting to address whether or not we should talk about shifts in the political landscape of North Fulton and Cobb in the same way we talk about shifts in Douglass and Henry.

This raises three questions:

1. How much of the 2008-2016 swing in big metro areas comes from increased minority populations and how much comes from whites flipping parties?

2. If whites are flipping parties in big metro areas, is this swing concentrated in favored quarters (like the North Atlanta suburbs) or can you see it in more downscale parts of the metro as well?

3. Regardless of swings between elections, do white voters in favored quarters consistently vote left of white voters in other parts of metro areas?

To answer these, I decided to dig into DRA data for four major metro areas with well-defined favored quarters: Dallas-Fort Worth (North Dallas/Collin/Southeast Denton), Houston (West Houston/Energy Corridor/North Fort Bend), Atlanta (North Fulton/North DeKalb/East Cobb/Forsyth), and Chicago (Streeterville to the North Shore). If I wanted to add more data, I would also take a look at Philadelphia, Seattle, and Washington. In each of these four metro areas, I determined the white vote in the 2008 election and the 2016 election in both the favored quarter and the rest of the metro area. This is what I got:

Dallas-Fort Worth
White vote, 2008, favored quarter: R+58
White vote, 2008, rest of metro: R+66
White vote, 2016, favored quarter: R+30
White vote, 2016, rest of metro: R+52

Houston
White vote, 2008, favored quarter: R+66
White vote, 2008, rest of metro: R+80
White vote, 2016, favored quarter: R+34
White vote, 2016, rest of metro: R+80

Atlanta
White vote, 2008, favored quarter: R+34
White vote, 2008, rest of metro: R+36
White vote, 2016, favored quarter: R+8
White vote, 2016, rest of metro: R+30

Chicago
White vote, 2008, favored quarter: D+48
White vote, 2008, rest of metro: D+22
White vote, 2016, favored quarter: D+60
White vote, 2016, rest of metro: D+14

After you get past the topline numbers (North Atlanta whites are probably going to vote for Biden even though sunbelt whites are absurdly conservative--~80% of Houston whites voting for McCain--wtf!), you notice a very consistent pattern: favored quarter whites ALWAYS vote left of the rest of the metro, metropolitan whites really did swing towards Hillary Clinton, and these swings have mostly happened in favored quarters. In 2008, the average favored quarter white voter was 13 points left of the rest of the metro, while in 2016, the average favored quarter white voter was 34 points left of the rest of the metro.

Metro-by-metro, white swings (2008-2016) were:

Dallas-Fort Worth
Favored quarter whites: D+28
Rest of metro whites: D+14
Relative swing of favored quarter: D+14

Houston
Favored quarter whites: D+32
Rest of metro whites: D+0 (!)
Relative swing of favored quarter: D+32

Atlanta
Favored quarter whites: D+25
Rest of metro whites: D+7
Relative swing of favored quarter: D+18

Chicago
Favored quarter whites: D+12
Rest of metro whites: R+8
Relative swing of favored quarter: D+20

What's the big takeaway?

First, there's massive internal variation in the white vote across big metro areas. Second, less upscale Romney-Clinton areas (like Henry County, GA) are moving left overwhelmingly because of the suburbanization of minorities. However, upscale wedges of metros (like the GA-400 corridor) are mostly moving left because of shifts in the white vote (some combination of the mythical Romney-Clinton voters, generational turnover, and coastal transplants). If these trends continue into the 2020s, we should be unsurprised by favored quarter counties like Collin, TX or Orange, CA voting to the left of non-favored quarter counties like Tarrant, TX or San Bernardino, CA.

So to address the original Del Taichi quote that prompted all this, North Atlanta is a Romney-Clinton place driven by Romney-Clinton voters while Henry, GA is a Romney-Clinton place driven by demographic change.

Hopefully you all find this interesting. I hope to update this after the 2020 election.

[Starry Eyed Jagaloon]

What is your definition of "favored quarter" since you place San Bernardino in it but not Orange County?

He said SB is not a favored quarter and neither is Tarrant.Parts of Tarrant are favored quarter although most of the swing in Tarrant from 2008 >2016 is mostly Demographical and the white vote in Tarrant was probably pretty neutral.

Actually I mistyped that and edited my original post. But you're correct here.

To be fair, Orange isn't the best example of a favored quarter because greater Los Angeles (and greater New York) are too big to fit such a topology, and if you had to pick a favored quarter in LA it would be from Downtown going west towards Malibu.

[Starry Eyed Jagaloon]

Probably a bit inaccurate, I doubt white votes in rest of Metro were 90% R, that seems a bit too far even for a deep southern city but good point. How did you estimate white vote? Im guessing you probably overestimated hispanic vote share ?
The rest of metro for Atlanta 08 seems a bit too D too IMO.

I could be off by a few points, but the error should apply to the swings equally and that's the really useful data. Basically I took the minority CVAP percentage of the DRA region, adjusted it for turnout and margin based on that year's exit polls, and subtracted it out of the Dem/GOP vote share with the remaining voted being the raw white votes. I did take into account TX Hispanic vote share but I was looking at statewide data and Harris could be different.

Also for atlanta I would exclude the actual precincts with the 400 road itself as those are quite low income IIRC.

For Chicago I feel like your definition is far too narrow for the "favored" quarter as there are plenty of more rich areas like most of Dupage or northern Will etc.

You'd be right if I'm going specifically for rich and educated precincts, but my thesis here is that regardless of these other demographic indicators, whites in these "favored quarters"--which correspond to very specific wedges of their cities--vote left of other whites. I'd speculate it's because these areas are where most transplants go, they're where most Class A office space (and the corresponding jobs) are, and they're culturally more liberal--which encourages people on the fence to go a specific direction.

[Starry Eyed Jagaloon]

What is your definition of "favored quarter"

Favored quarters are basically the wedge of a city radiating out from it's core that's the wealthiest. Usually one side of Downtown had the rich people to start with so "big ticket" growth all went that direction and employment centers followed. Today, they are usually places where the Fortune 500s, out of state transplants, and cultural attractions end up. For example, in Atlanta, the favored quarter goes north of Downtown which is why you have all the secondary CBDs in a line (Midtown, Buckhead, Perimeter Center, increasingly Alpharetta) going that way while you don't see the same development in other directions. The same applies to Philadelphia going west (Downtown, University City, King of Prussia/Radnor) or Houston going west (Downtown, West Loop neighborhoods, Uptown, Energy Corridor), etc. You know them when you see them.

[Starry Eyed Jagaloon]

Another interesting test for Georgia(is it Demographics vs white vote swing) is to check the 2016  senate(yes senate) vs 2008 presidential within Atlanta. If Barksdale did worse than Obama it was white vote swing but if he did better it was demographical swing. For example East cobb was +30 Mccain but +40 Johnny Iskason. On the other hand all Gwinett was McCain +10 and Iskason +4.

That's very interesting, and it basically reinforces my favored quarter idea. Gwinnett clearly lies outside it despite it being a broadly upscale (and diverse) county and it's shift being attributable to demographic change (in a way that doesn't apply to North Fulton, North DeKalb, East Cobb, and South Forsyth) is another strong data point in favor of this idea.

[Starry Eyed Jagaloon]

This is really interesting stuff thanks, sorry if this has already been asked but is the significantly larger swing in Dallas compared to "rest of metro" attributable to a greater younger of young white transplants? Alternatively I've heard that there's been a significant degree of white flight from Dallas so maybe it's the result of conservative whites being the ones that are more likely to leave.

Honestly, I'm not quite sure (although the biggest relative swing is actually in Houston. I'd speculate it might have to do with the sectors people work in. Dallas is more tech and finance heavy compared to the others.

[Starry Eyed Jagaloon]

OP, would you mind providing maps for each of these "favored quarter" regions? I have some quibbles with your Chicago-area definitions, but I'll withhold them until I know exactly which parts of the city (including parts of the city is one such issue)/suburbs you're referring to.

Sure.

Dallas:


Houston:


Atlanta:


Chicago: (Indiana suburbs also calculated in)

[Starry Eyed Jagaloon]

Let's start here, as I want to put this unequivocally on the record:

What're the demographic breakdowns of your "favored quarters" using 2018 CVAP vs 2010 Census?  Total population changes?

Atlanta, favored quarter:
2010 Census:

White	645,737
Hispanic	118,162
Black	135,253
Asian	99,342
Other	9,539

2018 CVAP:

White	533,214
Hispanic	37,648
Black	108,974
Asian	57,055
Other	4,727

Atlanta, rest of metro:
2010 Census:

White	1,333,514
Hispanic	379,932
Black	1,499,116
Asian	176,023
Other	36,995

2018 CVAP:

White	1,045,723
Hispanic	139,577
Black	1,178,547
Asian	102,590
Other	16,781

Chicago, favored quarter:
2010 Census:

White	809,399
Hispanic	241,176
Black	139,935
Asian	153,587
Other	14,337

2018 CVAP:

White	533,214
Hispanic	114,810
Black	91,766
Asian	97,558
Other	5,308

Chicago, rest of metro:
2010 Census:

White	3,821,483
Hispanic	1,648,994
Black	1,541,827
Asian	437,610
Other	68,403

2018 CVAP:

White	2,873,352
Hispanic	806,758
Black	1,069,561
Asian	264,000
Other	20,603

Dallas, favored quarter:
2010 VAP:

White	594,049
Hispanic	206,183
Black	102,969
Asian	93,813
Other	12,095

2018 CVAP:

White	662,045
Hispanic	128,615
Black	126,528
Asian	86,264
Other	11,815

Dallas, rest of metro:
2010 VAP:

White	1,409,330
Hispanic	795,274
Black	556,734
Asian	173,149
Other	42,833

2018 CVAP:

White	1,413,419
Hispanic	582,349
Black	619,850
Asian	148,490
Other	28,594

Houston, favored quarter:
2010 VAP:

White	390,220
Hispanic	251,596
Black	148,057
Asian	125,113
Other	11,391

2018 CVAP:

White	396,320
Hispanic	156,624
Black	142,942
Asian	107,193
Other	5,347

Houston, rest of metro:
2010 VAP:

White	1,169,254
Hispanic	1,020,722
Black	560,033
Asian	186,212
Other	36,965

2018 CVAP:

White	1,212,509
Hispanic	797,699
Black	632,174
Asian	172,927
Other	19,861

[Starry Eyed Jagaloon]

Do you not have citizen VAP data from the 2010 Census? That would be more useful than just the census count or just the total VAP. It’s hard to accurately gauge changes in the eligible voting population if we’re comparing apples to oranges.

Unfortunately I don't.

[Starry Eyed Jagaloon]

Do you not have citizen VAP data from the 2010 Census? That would be more useful than just the census count or just the total VAP. It’s hard to accurately gauge changes in the eligible voting population if we’re comparing apples to oranges.

Unfortunately I don't.

Doesn't DRA have 2010 VAP data, at least?

Depends on the state. I gave it for Texas but it wouldn't load in Georgia and Illinois/Indiana.