Georgia 2020 Redistricting Discussion

[Adam Griffin]

.......

It's possible to draw an 11-3 map on paper, but it's going to potentially be one hell of a dummymander: the best you could hope for would be a 9R-5D split by mid-decade. When you start trying to maximize GOP performance everywhere - even in many rural areas - those districts start getting flimsy. That doesn't even count the ATL metro, which is its own problem.


Perhaps the only tangible requirement are the 3 VRA-protected ATL districts. Bishop's district isn't protected as I understand it, but the GOP drew it as a votesink (remember that Jim Marshall barely lost in 2010: a decent chunk of South GA was competitive then). I highly doubt it'll be eligible for protection in 2021, but I could easily see them drawing it once again, with the below map illustrating why.

Those 4 South GA districts are not particularly strong for the GOP. Obviously the GOP isn't going to draw a bunch of relatively non-contiguous and/or east-to-west abominations that run the risk of being struck down in court post-2022. When you take apart Bishop's district's votesink configuration, you start unleashing a cascading effect across the southern half of the state.

Trump won all 4 of these districts by 10-15 points - which doesn't necessarily make them very competitive (but they're not particularly great performances, either) and I'd say 3 of the 4 could be vulnerable under the right set of circumstances. In particular, CDs 13 & 14 have sizable and growing urban/suburban areas. I'd also point out that in Bishop's real district (CD 2; below, it'd be CD 11), he overperformed Clinton in 2016 by 10 points (Clinton +12 vs Bishop +22) - there's a chance you could still end up losing such a seat, or having to fight tooth and nail for it at minimum. Throw in a bad cycle for the GOP and a turnout differential, and things can get real messy real quick in multiple districts.


For the metro itself, it's fairly easy to shore up the GOP - assuming the kinds of swings from 2016-2018 don't continue unabated (I think it's a fair assumption; demographic shifts will continue but there aren't hordes of white suburbanites remaining to be flipped as has happened over the past 3 years). The 10th district obviously falls either in 2022 or 2024 (it's filled with young non-whites, and I could see the turnout differential in 2022 costing Democrats the seat initially, as what happened in the SD-43 special in 2015, which was an Obama +40 district). District 3 would probably fall somewhere around mid-decade, being the Democrats' 5th guaranteed seat; I think CDs 2, 4 & 6 would hold barring massive growth and demographic shifts.

[Adam Griffin]

At any rate, given this will likely be the last cycle that the GOP gets to draw maps in an uncontested fashion, it makes little sense for them to take on potentially huge risks that could wipe them out prematurely. They can either cushion the fall and make it a much more gradual one - where they maintain a majority of the congressional delegation and/or state legislature for another 8-10 years - or they could roll the dice and potentially cost themselves way more far sooner than they would otherwise lose it (this is especially true in the General Assembly). The VRA of course helps temper their worse temptations in this regard, but I don't see the GAGOP trying to overextend themselves given the clear trajectory of the state at this point.

[Adam Griffin]

BTW, Georgia DRA now has 2016 Presidential & Senatorial results in lieu of the 2012/16 PVI ratings, so you can see both contests' results by precinct.

The map above I initially drew using 2016 block group population estimates and racial data, but sadly when using those, only the 2008 data is displayed. I redrew the map using 2010 VTDs to get the 2016 margins for each district.

[Adam Griffin]

At any rate, given this will likely be the last cycle that the GOP gets to draw maps in an uncontested fashion, it makes little sense for them to take on potentially huge risks that could wipe them out prematurely. They can either cushion the fall and make it a much more gradual one - where they maintain a majority of the congressional delegation and/or state legislature for another 8-10 years - or they could roll the dice and potentially cost themselves way more far sooner than they would otherwise lose it (this is especially true in the General Assembly). The VRA of course helps temper their worse temptations in this regard, but I don't see the GAGOP trying to overextend themselves given the clear trajectory of the state at this point.

Speaking of which, could the control of the lower house be in play in a scenario where Trump loses GA?  At the very least, I'd expect a Dem performance equivalent to VA 2008 would flip it?

Perhaps - I'm not sure of exactly where along the sliding scale the dam would break, but the General Assembly performance lagged the gubernatorial performance by roughly 7.5 points in PV alone, and the congressional PV by 4 points.

Contest	GOP	DEM	GOP	DEM	Margin
Governor	50.2	48.8	1978408	1923685	R+1.4
US House	52.3	47.7	1987191	1814469	R+4.6
State House	54.3	45.6	1884211	1582161	R+8.7
State Senate	54.4	45.6	1914110	1603472	R+8.8

Some of this is certainly due to a lack of credible (or any) challengers on the part of Democrats in state legislative districts - even just a couple of cycles ago, Democrats didn't even field the 91 candidates needed on paper in the State House to theoretically take the majority. I wish I knew how much could be attributed to that off the top of my head (I've calculated it before I believe), but alas.

[Adam Griffin]

At any rate, given this will likely be the last cycle that the GOP gets to draw maps in an uncontested fashion, it makes little sense for them to take on potentially huge risks that could wipe them out prematurely. They can either cushion the fall and make it a much more gradual one - where they maintain a majority of the congressional delegation and/or state legislature for another 8-10 years - or they could roll the dice and potentially cost themselves way more far sooner than they would otherwise lose it (this is especially true in the General Assembly). The VRA of course helps temper their worse temptations in this regard, but I don't see the GAGOP trying to overextend themselves given the clear trajectory of the state at this point.

Speaking of which, could the control of the lower house be in play in a scenario where Trump loses GA?  At the very least, I'd expect a Dem performance equivalent to VA 2008 would flip it?

Perhaps - I'm not sure of exactly where along the sliding scale the dam would break, but the General Assembly performance lagged the gubernatorial performance by roughly 7.5 points in PV alone, and the congressional PV by 4 points.

Contest	GOP	DEM	GOP	DEM	Margin
Governor	50.2	48.8	1978408	1923685	R+1.4
US House	52.3	47.7	1987191	1814469	R+4.6
State House	54.3	45.6	1884211	1582161	R+8.7
State Senate	54.4	45.6	1914110	1603472	R+8.8

I have to imagine there were a bunch of outer ATL seats Abrams ended up winning that legislative Dems didn't seriously contest?

Late edit above, but there were some districts where Democrats could have won if they had ran credible candidates, and some where Abrams won despite our legislative candidates losing.

In 2018 in the State House, Democrats fielded 118 candidates compared to the GOP's 127: the difference is equivalent to 5% of the seats in the lower chamber. I'm not sure about 2018 specifically, but I wouldn't be surprised if the Democrat stands to receive a greater share of the vote in an uncontested GOP district (say, 30-35%) than vice-versa (20-25%).

I'm guessing right now, but I figure the difference between the 127 vs 118 & the 30-35% vs 20-25% probably would account for around 1 point or so.

[Adam Griffin]

Well thank y'all for your kind 8 seat maps, but the GA GOP doesn't even need to go close to that

https://imgur.com/a/cpbG9t6

Here's a safe R 11-3 map.

Green ATL District - 22%-75%, Clinton +53

Purple ATL District - 15%-83%, Clinton +68

Orange ATL District - 13%-85%, Clinton +72

Yellow ATL District - 57%-38%, Trump +19

Greenish-Blue ATL District - 57%-39%, Trump +18

Grey ATL District - 58%-38%, Trump +20

Purple District that includes Athens (to the right of ATL) - 60%-37%, Trump +23

Illuminated Blue district along SC border - 61-37%, Trump +24

Pink District that includes Savannah - 59-38%, Trump +21

Light Green South GA district - 60-38%, Trump +22

North to Central GA light blue district - 70-28%, Trump +42

Brown District South of ATL - 57-40%, Trump+17

Alabama Border District - 64-34, Trump +30

Weird Central GA Orange District - 62-36, Trump +26

It ain't pretty, but it gives you 11 safe R Trump +17 or higher districts. There's no need to come even close to an 8 seat map - drawing an 11 seat map isn't even too hard, though a 10 seater might be better to avoid VRA trouble.

Prior to the suburban collapse, you could easily draw a 12-2 or even a 13-1 under the right circumstances (the former may still be possible), but they're just as realistic as the one above: no way would that hold up in court, even if just on the splicing alone.

You've packed black voters into 2 60%+ black VAP districts and proceeded to crack the 3rd required district (which can't be much more than 40% black VAP from the looks of it); not a single one of those would be Section 2 VRA-compatible. It would be hilarious if the GAGOP tried this, it somehow managed to stand for a single cycle, and a Democratic Governor was elected in 2022 who would then force the maps to be drawn by the courts.

Required reading: the last congressional map passed by Democrats in GA; the Barnes Dummymander

[Adam Griffin]

https://imgur.com/a/pFAldhX

Here's a better GA map, probably legal, much cleaner and more compact, safe 10-4.

Once again, you've made a critical mistake: the 3 black-majority/plurality districts in Metro ATL are required - without packing or cracking - and you can't substitute one of those by drawing an east-to-west abomination connecting Albany, Columbus, Macon & Augusta. Looks like the 2 black-majority districts you've included in the metro are severely packed (>60% black VAP). The gray one almost certainly wouldn't pass muster (majority-minority I'm sure, but that's not the same thing; thanks, Gingles!)

I'll make it easy for you: draw 3 districts in metro ATL where black voters are the largest group in each (between 48-55% VAP; ideally based on 2016 estimates) before doing anything else. Then see how far you can extend the dominance of the GOP without creating multiple vulnerabilities and/or drawing absolutely atrocious lines. There's a reason why the GOP maps of 2011 were the first since the passage of the VRA that didn't undergo massive court challenges: they aren't the type to take risks (try being out of power for 150+ years and then gambling your fleeting decade-long majority with some bad maps).

[Adam Griffin]

What's the case for 55% VAP being the upper limit? Given that the Supreme Court isn't exactly keen on defending the VRA, surely there's an incentive to push the envelope there? More generally, it looks like it shouldn't be that hard to remove largely white precincts from the 3 VRA districts and replace them with more Hispanic/Asian precincts from GA-6 and GA-7.

The east to west abomination won't satisfy the Gingles criteria, but it might be an effective way of packing GA-2 and freeing up space for other districts to bacon-slice the Atlanta metro.

It's somewhat subjective, but around that number is when VRA packing cases tend to start being quite successful. There are of course a number of factors that can impact such, but for a congressional map in GA, I'm comfortable in saying that you don't want to exceed that figure if your maps making it through the decade are a priority.

[Adam Griffin]

Would this satisfy the VRA for GA 2? This packs more of Atlanta to be safer for the GA GOp but still has 4 black majority districts(actually all around 49% VAP)

Hilariously enough, reducing all 3 black VRA districts to pluralities could possibly open up arguments for both packing and cracking here. There is some subjectivity to it, but on one hand, this vaguely resembles what Democrats might try to do (spread black VAP as thin as possible to maximize seats) and what the GOP would pursue (cramming various non-compact black communities together). Even if cracking doesn't benefit the GOP necessarily, that doesn't mean it can't be used as an argument against the maps.

Drawing a 4th black-plurality district also doesn't offset any inadequacies with the 3 required districts: it's not a matter of "balancing things out" in this case.

Remember that GA-2 was 49.5% BVAP as drawn in 2011, and still isn't VRA-protected. Honestly, I think it'd be better to slightly exceed the 55% BVAP threshold (after all, the GOP did this by a tad in 2011: the 3 black-majority districts were 53.9%, 56.4% and 57.6% VAP) than leaving all of them as plurality-black VAP districts, if surviving legal challenges is the goal.

[Adam Griffin]

Georgia
Perhaps the best place to start with is the “Known Knowns.” First, Georgia has to have at least four African American seats.

Incorrect: 3 Metro ATL plurality/majority-black districts will be all that's required barring major changes. Bishop's seat is not protected and was instead designed as a vote-sink in the 2011 map to ameliorate concerns about another Jim Marshall winning in what would certainly be a nominally-Republican South GA district at the time (given South GA's Democratic tilt for many offices then) in any set of maps that weren't obviously gerrymanders (the GOP didn't want to take risks with DoJ approval, plus they were pretty confident about Barrow falling in 2012). Additionally, I'm fairly confident the black voting age population percentage in his district (which was a plurality in 2011) has only decreased since then. McBath's seat also obviously isn't protected given the race of the candidate means nothing and it's a majority-white district.

As far as I know, simple statewide demography is not a guarantee for proportional or near-proportional congressional representation under the VRA. If so, then AL and LA both should have 2 majority-black Democratic districts. If every county in GA had 30% black population, then you would likely have no black-majority districts mandated under the VRA.

[Adam Griffin]

What would a McBath vote sink look like?

While this uses 2010 population & 2016 presidential numbers (I might try to replicate the district using 2016 block groups, as I imagine it would be a bit smaller geographically and blacker now), this 50% BVAP would kill two birds with one stone: shove McBath into Scott's territory and consolidate the district. McBath lives in that broader peninsula in the NE Cobb portion of the district.

Here's a fun observation, though: Scott isn't registered to vote in GA-13 - at least according to the voter file. His address is clearly in GA-5, in Little Five Points. I don't know if this is a recent change or what, or if he's been running in a district he hasn't lived in for ages...?

EDIT: this has been the case since at least 2017:

There’s also Rep. John Lewis (D-Ga.), who represents the district. But a review of vote registration records by The Washington Post suggests that Ossoff would be the third member of Congress to make his home in the 5th. According to voter data provided to The Post by the political data firm L2, Rep. David Scott (D-Ga.) is also registered to vote in the district, instead of the 13th District that he represents.

[Adam Griffin]

also besides the obvious implications for Congressional redistricting, note that each State House district would have a population of ~58,442 using this estimate

among many other observations to be made, note that the metro counties that voted for Hillary are collectively gaining like eight house seats here

Used your map to try and break the state into various clusters of growth (or loss):



Among the red cluster of counties above that have lost population, Clinton counties were responsible for 15,183 out of the 21,348 lost. However and overall, since 2010:

Clinton Counties: +488830
Trump Counties: +342992

[Adam Griffin]

Do you have a map for this with 2016 data? Apologies if I missed this!

Actually you may have meant his purported 10-4 map, but in case not:

[Adam Griffin]

Yeah, it looks like I was informed about GA-2's status under the VRA (and not due to it being plurality-black VAP, of course). My initial understanding was that - after all the chaotic rounds of redistricting between 2001-2006 - the district was not required by the VRA when it was drawn in 2011.

For what it's worth and has already been said, population loss across the region isn't necessarily making it more difficult to redraw the broader 2011 boundaries: it's pretty easy (surprisingly) to draw a comparably geographically compact GA-2 using 2016 data, especially if you include parts of Houston County and all of Bibb...even though basically everything save for Houston (+16k), Muscogee (+3k) and Lee (+1k) have lost population (second map below; red = growth & blue = loss). The district below as a whole has only lost 7k people since 2010.

The only questions are 1) how much change will occur between 2016-2020 in population estimates and 2) how much of a difference exists between BPOP & BVAP (it is worth noting that the district below went from 49% black to 51% black between '10-'16). I figure that a new GA-2 is probably going to want to include parts of Houston unless the GOP wants to aim for a mid-decade win here (while Bishop overperforms, he won't be around forever: Clinton only won the district below by 9 and Isakson came within a half-point of carrying it in 2016).

 

[Adam Griffin]

There's also a chance that Congress expands the House by 100 districts and implements MMDs, but that's not the world as it exists today - nor is it even likely based on current events and reality. If/once the courts begin deliberating on tangible changes to the VRA, then you might have a point to make in such a discussion (and believe me, there's plenty regarding the VRA that I think is either outdated or insufficient). However, this is about what happens in 2021 - the likelihood of a major change in either redistricting policies or the VRA between now and next year is small enough to completely dismiss such discussion presently.

[Adam Griffin]

Given how no one has drawn a white Dem district (is one even possible?)

Using 2010 figures (which I hate to do, but I wanted easy access to new election data), drawing 2 3 majority-white VAP Dem (or Dem-leaning) districts anchored around the metro is simple enough - even with 3 VRA districts adjoining them. I'm sure the lines could be tidied up a bit on subsequent revisions, but this is more proof of concept than anything else.

Not only that, but drawing a fourth majority-white VAP lean Dem district between Augusta and Savannah is fairly easily done as well.

https://davesredistricting.org/join/acc4198d-2ac6-4d68-98e4-46f28e4c138d

CD	White VAP	Black VAP	2016 Pres	2016 Sen	2012-16 Composite
1	56%	21%	Clinton +30	Barksdale +9	Democrat +15
2	56%	26%	Clinton +19	Barksdale +7	Democrat +10
3	36%	53%	Clinton +40	Barksdale +29	Democrat +34
4	39%	54%	Clinton +31	Barksdale +25	Democrat +28
5	39%	49%	Clinton +24	Barksdale +16	Democrat +20
6	55%	37%	Clinton +3	Isakson +7	Republican +1
--	--	--	--	--	--
8	56%	17%	Clinton +2	Isakson +15	Republican +11

Caveats:
1) CD 5 is 49% BVAP by 2010 standards. I wouldn't usually draw an ATL VRA district as plurality-black, but such a configuration should be majority-black VAP today.
2) CD 2 might actually be close to plurality-white by now, given the huge shifts in Gwinnett County.

[Adam Griffin]

For what it's worth, it's completely possible to draw a broader metro set of districts under 2018 population estimates that include:

• 3 VRA districts
• A (very meaningfully VAP) majority-white ATL district
• Another majority-white (VAP) district
• 2 GOP-leaning districts

The lines are by no means clean and neat, but that's never been a real issue in Georgia redistricting anyway. By default, this would produce a 8-6 map (though the 2 GOP-leaning districts' fates throughout the decade would always be in question).

CD	White %	Black %	Other %	Margin
1	59	19	22	Clinton +30
2	43	17	40	Trump +5
3	61	21	18	Trump +9
4	49	33	18	Clinton +5
5	40	53	7	Clinton +15
6	31	57	12	Clinton +35
7	27	65	8	Clinton +48

At this point, drawing a "fair" 7-7 is child's play when factoring in Bishop's district plus the ease of drawing an Augusta-Savannah district to boot (which can neatly be done under '18 population, with a Clinton win of 6-9 points).

[Adam Griffin]

CVAP is also another consideration: there seems to be some debate (at least historically) on whether to use VAP or CVAP, but if CVAP is the litmus test, it's potentially possible to draw 4 majority-black CVAP districts in the Atlanta metro now.

Based on 2018 raw population estimates, these districts (1-4) are 53%, 56%, 45% and 54% black, respectively.

Given that CD-3's combined Latino/Asian population is around 30%, it's almost certain that it's majority-black CVAP as well (~51-52%).

[Adam Griffin]

Got bored, so I wanted to make a map that maximized Democratic House seats without creating either an overly-risky situation or one that blatantly would violate the VRA.

I also had an impulse to expand the existing ATL majority-black districts as far out into rural areas as possible, which is why their Democratic vote share more closely resembles their black vote share; obviously the GOP's "enforcement" of the VRA whereupon they aim to create 55% black districts that vote 80% Democratic is pure gerrymandering. It has the side-effects of avoiding that common, unsightly L-shaped CD that forms in between ATL & Macon/Columbus (or further south - CD 4 here doesn't look anywhere nearly as bad by comparison); also, being rep'd by Democrats makes voters less hostile to them over time.

Additionally, creating the oft-overlooked opportunity for a compact, majority-white Augusta-Savannah CD featured into the plans (one that can easily be made >Clinton +5 without this map specifically that forgoes everything north of Augusta while including all of Chatham & Liberty).

While this map might trigger a VRA lawsuit - it creates 5 black-majority CDs based on 2018 population, but 3 of them are 51-52% black (meaning they're probably 49-50% BVAP) - it basically hands Democrats 7-8 seats depending on the climate, with the likely prospect of a 9th seat later in the decade.

CD	BLK %	WHT %	OTH %	'16 P MAR	'16 S MAR	Specs
1	29	53	18	Clinton +1.5	Isakson +12	Cobb
2	51	41	8	Clinton +17	Barksdale +9	VRA
3	10	77	13	Trump +53	Barksdale +54
4	52	36	12	Clinton +17	Barksdale +10	VRA
5	51	43	6	Clinton +13	Barksdale +7	VRA
6	55	34	11	Clinton +60	Barksdale +46	VRA
7	24	38	38	Clinton +20	Barksdale +5	Maj-Min; Gwinnett
8	17	58	25	Trump +7	Isakson +25	Future R-to-D
9	14	70	16	Trump +39	Isakson +43
10	5	81	14	Trump +54	Isakson +58
11	42	50	8	Clinton +2	Isakson +7	Swing District Trending D
12	28	63	9	Trump +31	Isakson +38
13	51	41	8	Clinton +11	Barksdale +2	VRA
14	31	60	9	Trump +36	Isakson +40

[Adam Griffin]

[Adam Griffin]

At any rate, yeah, with the 2018 Census figures, it's going to be tough for the GOP to not draw 5 VRA-protected districts. That still leaves a huge chunk of the metro that's majority-white and majority-minority that's strongly D as well. From there, they have to decide whether they want to hack and splice into oblivion (potentially risking more lawsuits and/or a late-decade dummymander), or give a 6th seat to Democrats in the northern metro and be safely guaranteed their 8 for the rest of the decade.

[Adam Griffin]

Not a huge fan of putting Walton into the Black majority district.

I am. The white residents of Walton County refer to their county seat (Monroe) as "Monkeyrow". Seems like a fitting "punishment" (from their perspective).

[Adam Griffin]

Wanted to draw a map that was proportionate in partisan terms while also trying to adhere to VRA and minimize splits in smaller counties & excessive carving in larger ones. In the end, I ended up with a 7D-7R or 7D-6R-1 map - depending on your perspective - last night. Of course, with the presidential data now added, things change a bit (the swing district is a Kemp-Biden district).

In trying to minimize hacking and splicing while also avoiding the dreaded "hook" that often emerges from LaGrange and either ropes around Macon and to the south or keeps extending eastward, I definitely stretched black majority districts fairly thin in the metro (while of course ending up with 1 >60% BVAP district as the remainder).

I also despise how so few are willing to entertain a Savannah-Augusta district, so I always make sure to include it in any 7D-7R or 8D-6R proposal.

https://davesredistricting.org/join/5a1ca72f-69e6-4466-9147-3473ecf6be60

[Adam Griffin]

I will say this: I would be very surprised if non-metro NWGA gets split. Just like in 2011 when a new district (GA-14) was added, the NWGAGOP delegation will organize, lobby and insist upon the region being kept together for regional representation purposes (and it has enough GOP legislative representation to pose a threat to reapportionment plans counter to that). Perhaps the only exception to this rule is trying to dislodge MTG, but even in that case, I doubt it. She'll just carpetbag to wherever she has to carpetbag like she did to run in the 14th (never forget: she's a Forsyth-Cobb creation!).

Given the current GA-14 has (more or less) kept up with population growth, something like this wouldn't surprise me:

[Adam Griffin]

Are there any reasons Republicans might prefer an Augusta-Macon-Albany district to a Macon-Columbus-Albany district, or are we pretty certain that the configuration of GA-02 isn't going to change significantly (barring the GOP stupidly trying to destroy it entirely)?

That would almost certainly make the district more Dem-friendly than the current iteration of GA-2 (and by a considerable amount, as it would take in more white voters who aren't going 90-10 in favor of Rs) and might even force some R weakening of other downstate districts (for instance, I could easily see a sub-Augusta, coastal district being only 5-10 points in favor of Trump '20).

SW GA as a whole is becoming more favorable to Rs with its population collapse (while it is becoming blacker, the whites that remain have become much more R). It's not unfeasible for the GOP to draw some similar incarnation of the current GA-2 that could have a chance at flipping by the end of the decade or whenever Bishop retires.