Read on Twitter
I'm starting something new! Join me here on Fridays for a deep dive into redistricting topics as they pertain to NC district plans. For this week's #FairMapsFriday, I look at how the GOP used incumbency protection in the 2019 redraw of #NC8 and #NC9 to subvert democracy. #ncpol
In case you're new here: in 2019, the GOP-controlled NCGA redrew the state's entire Congressional map after it was challenged in court as an extreme partisan gerrymander. This is after the original 2011 plan was found to be a RACIAL gerrymander and redrawn by court-order in 2016.
In what is baffling to me as a non-lawyer, the NCGA redrew these maps in 2019 at the *suggestion* of the court. There was no court order telling them what was allowed and not allowed. And so the GOP adopted their own criteria. They seemed reasonable at first glance...
Equal population and contiguity are the most basic criteria. The acceptable population deviance for Congressional districts in NC is ONE person. Contiguity just means that districts can't have multiple disconnected pieces.
The Data section has some tricky language. It says "Data identifying the race of *individuals* or *voters* shall not be used..." But they don't explicitly forbid the use of VTD-level racial data, which they used in 2011.
In the compactness section, they say they shall make a "reasonable effort" to create districts that are as or more compact than the 2016 districts, but they clearly don't have an interest in creating the most compact districts possible. This is a "Get Out of Jail FREE" card.
They explicitly forbid "partisan considerations" and the use of election results data in map construction, though I would argue that incumbency protection *is* a partisan consideration. More on that later...
OK finally - Double Bunking. Even though Congressional candidates are not actually required to live in the district they represent in NC, the NCGA set out to make "reasonable efforts" not to pair incumbents in the same district (or "double bunk" them).
Even though the legislature used "reasonable efforts" twice in their criteria, they don't set boundaries on what they consider to be "reasonable." Without a clear definition, *anything* that doesn't break the hard boundaries of contiguity and population size becomes reasonable.
This means that you can stretch districts out a third of the way across the state in the name of incumbency protection, even when far more compact options are available. This is what was done with #NC8 and #NC9 to protect incumbents Richard Hudson and Dan Bishop.
At the time, I complained about Districts 8 and 9 a LOT, to no avail. My main issues were 1) you could draw districts that were twice as compact, but would double bunk the incumbents and 2) there was too much overlap between the old and new districts. https://twitter.com/EJSbrocco/status/1195313626212700160?s=20
I wrote more about population overlap in this thread: https://twitter.com/EJSbrocco/status/1197152199677227009?s=20
I also had major issues with the process. The online public comment portal closed before the final map was selected, and there were 18 maps up for consideration at the in-person comment session! It was hard to know what to focus on, but that was the point. https://twitter.com/EJSbrocco/status/1195333118821126146?s=20
These maps were rushed through the redistricting process in what I have called "legislative blitzkrieg." This not only prevented the public from offering meaningful input but also kept data scientists like me (and others) from fully evaluating each plan before a vote was taken.
My sincere hope is that by doing a post-mortem on this plan, I can help the public and legislators recognize the warning signs of an unfair map so that they know what to object to next time. It makes sense to start with #NC8 and #NC9 here for a few reasons.
First - these were the last two districts to be drawn during the process in 2019. That means any alternative plan that I can come up with that only changes Districts 8 and 9 is an alternative that was available to legislators in 2019.
Second - District 9 was the tightest Congressional race in the state in 2018. Is it reasonable to draw a plan that so closely resembles the previous partisan gerrymander in the name of incumbency protection? Surely there will be a partisan impact in such a tight race!
We can eyeball these districts and see that there are more compact ways to draw them, but we can't identify the full range of possibilities in compactness scores when we pair or don't pair incumbents without simulating random plans and seeing if the enacted plan is an outlier.
So that's what I've done. Using the criteria laid out by the NCGA, I drew 944 random, unique district plans for NC8 & 9 in order to evaluate how plans that paired incumbents compared to those that didn't with respect to compactness, electoral outcomes, and racial composition.
(I will not get into deep methodological details here, but I am happy to share if you send me a DM or email.)
First, I found that it is a lot easier to draw plans that pair incumbents than it is to draw plans that don't. 900 of the 944 plans (95%) placed Hudson and Bishop in the same district. And plans that pair incumbents in the same district are far more compact than those that don't.
The stars in the figure above mark the compactness values for the 2019 enacted plan. 82% (based on the Polsby-Popper score) or 100% (based on the Reock score) of plans that pair incumbents (purple bars) are more compact than the enacted plan (stars).
More generally, 64% (Polsby-Popper) or 94% (Reock) of plans that pair incumbents (purple bars) are more compact than the *most compact* random plan that doesn't (orange bars).
Finally, the most compact plan that pairs incumbents is 1.5x more compact than the most compact plan that doesn't based on the Polsby-Popper score, and is more than twice as compact based on the Reock score.
Moving on to demographics: I looked at the % of the total voting-age population that is black (Black Voting Age Population, or BVAP) for both districts in each plan to see how the distribution of BVAP compares in plans that pair or don't pair incumbents.
As before, the stars mark values for the 2019 enacted plan, orange bars show the distribution of plans that don't pair incumbents, and purple bars show plans that do pair incumbents. The dotted verticle line is the % BVAP required to elect a Democrat (VRA analysis not shown).
Two things stand out in this chart. First, plans that don't pair incumbents have elevated values in the district with the lowest BVAP and depressed values in the district with the highest BVAP relative to plans that pair incumbents. This is a classic signature of "cracking."
"Cracking" is a gerrymandering tactic that dilutes the voting power of a group by splitting it among two or more districts. This keeps the group from electing a candidate of their choice in any district.
Second: In this part of the state, I found that a Black Voting Age Population of ~29.5% or greater consistently elected a Democrat. Plans that didn't pair incumbents never reached this threshold, whereas plans that did pair incumbents exceeded it 82% of the time in one district.
By prioritizing the protection of incumbents, NC state legislators completely eliminated the opportunity for black voters in these districts to elect a candidate of their choice. Incumbency protection, in this case, resulted in a racial gerrymander.
We can also look at election data. Here I looked at the election results for the 2016 Presidential contest, but other contests yielded similar results. As above, the asterisk shows the enacted plan and orange bars show the plans that don't pair incumbents in the same district.
As with BVAP, the Democratic voteshare in plans that didn't pair incumbents were elevated in the least Democratic district, and depressed in the most Democratic district. Again, this is a classic signal of "cracking" of Democratic voters so that they can't achieve a majority.
Plans that don't pair incumbents NEVER gave Democrats a win, while plans that do pair incumbents give Democrats a win in one district 82% of the time.
By prioritizing incumbency protection at all costs, the GOP-controlled legislature in NC has completely eliminated the possibility of Democrats winning a Congressional seat in these districts. Incumbency protection, in this case, results in a partisan gerrymander.
I started this thread by saying that without a clear definition of "reasonable," anything goes. Is making a district slightly less compact in order to accommodate incumbents a reasonable burden to place on voters? Sure. But when compactness is cut in half? The line is less clear.
When protecting incumbents of one party completely eliminates the possibility of electing members of the other party, it is clear that the line has been crossed! Extreme racial and partisan gerrymanders should never be viewed as reasonable burdens to place on voters.
That's it for now. Feel free to DM me for more info! And stay tuned next week for another edition of #FairMapsFriday when I'll continue this tour of gerrymandering masquerading as incumbency protection in NC Congressional Districts.
