New Tech for the People
Last week in North Carolina, a state court panel ruled that the state’s 2017 house and senate legislative districts were unconstitutional extreme partisan gerrymanders. Yuck. The court took some very pro-democracy steps in their ruling: the General Assembly was given two weeks to come up with new maps with a very specific pro-democracy, pro-constitution, pro-American mandate — no partisan data at all can be used to draw the maps, and the entire process has to be out in the open for public view. Power to the people. Power to opponents of gerrymandering. Power to the users of Dave’s Redistricting App.
Today, Dave’s Redistricting is pleased to introduce a special release of tools, data, and analytics to support the general public’s active involvement in the redistricting process in North Carolina. Dave’s Redistricting App (DRA) was the original free-to-the-public redistricting tool for the 2010 Census. Now rebuilt from scratch as a modern web service, DRA 2020 is available free to redistricting enthusiasts nationwide.
New Tools, Data, and Analytics
There are three new functions that are relevant to anyone who wants to understand proposed district maps in North Carolina:
1. Pre-loaded Data — specific to 2019 NC redistricting, namely Total Population (2010 Census), Voting Tabulation District shapes (VTDs) (2010 Census), Voting Age Population (2010 Census), and 2016 NC election results from the Presidential, US Senate, Governor, and AG races, combined into a Voter Preference Index. Special thanks to Michael McDonald and his team at FEST for the 2016 election data.
2. Block Assignment Files — we now have the ability to load arbitrary Block Assignment Files for North Carolina (e.g. the CSV files published by the NC General Assembly here, here, or here or maps produced by other commercial mapping tools).
3. Analytics Suite — the new Analyze button generates a detailed report specific to the map you have loaded, including any changes you make to it.
New, Free Stuff?
So what should I do with it? We have some ideas:
- Start by creating a free account on Dave’s Redistricting
- Load up a copy of the 2017 House or Senate district maps that were ruled unconstitutional on September 3, 2019 — just so you can get an idea of what an extreme partisan gerrymander looks like in North Carolina. Play around with it, move the mouse around to different districts to view population and partisanship data, play with different display options, zoom in and zoom out.
- Click the Analyze button for the map you’re viewing to check out the redistricting analysis report. (More details, nuances, and caveats about this version of the analytics are detailed below.)
- Visit the NC General Assembly web site regularly to see what proposals are coming out of the House and the Senate. The GA is supposed to pass remedial maps under the directions the court provided them, by September 18. Final proposed maps are due out any day. Will they continue to violate North Carolina’s state constitution? Will they follow the court’s instructions? What do the analytics say about these maps?
- Now you’re ready to dive in for real. Check out the steps below for digging in deep, making your own maps, and responding to maps proposed by the NC House or Senate. Your voice matters. They’re even taking public comments on NCLeg.gov.
How to Draw New House and Senate Maps
To draw or evaluate and NC House or State Senate map, you first need to create a new blank NC map with the correct number of districts, 120 for the House and 50 for the Senate. These are the steps:
- Login to DRA 2020
- Click New Map
- Give your map a Name
- Select North Carolina as the State
- Note: Leave the data source as is (defaults to 2010)
- Change the District Count to 120 for House or 50 for a Senate map
- Scroll to the bottom of the window and click Apply
If you are going to start from scratch you are ready to go. Be sure to click Help (?) for help on creating districts by coloring VTDs, view options and more.
We highly recommend customizing the District Panel (tall panel on the left showing Census and election data), to view only the sections you want.
If you want to load a block assignment file (such as the current NC House file), do the following:
- Click the gear icon (⚙) in the upper right corner
- Under Color from a .CSV file, click Choose File, and select the block assignment CSV, which should give you feedback on the number of blocks, etc.
- Click Apply
How to Evaluate Proposed House and Senate Maps
To evaluate a proposed map, or a map you get from somewhere else, get the block assignment file for the map. (The new NC maps should be published here.) Then follow the steps in the section above to create a new blank map, and load the block assignment file using the “Color from a .CSV file” feature in Settings.
What Data We’ve Pulled Together For You
When you create maps in this special release of Dave’s Redistricting for North Carolina, we use four datasets:
- 2010 VTD shapes — VTDs are the geographic shapes you assign to form districts.
- 2010 Census data — We use the Census’s total population by VTD, to compute population deviation by district and to determine whether a set of districts has roughly equal populations
- 2010 Census Voting-Age Population (VAP) demographic breakdowns — These support analyses of the opportunity for minority representation. Here we use “lone or in part” demographic categories that may overlap, as opposed to single-race, mutually-exclusive ones.
- 2016 Presidential, US Senate, and Governor, and AG election results — These are combined into a single statewide Voter Preference Index (VPI), to help you evaluate the partisan lean of districts as well as individual VTDs.
NOTE: if you are a Legislative Defendant, as defined in the court’s ruling in Common Cause vs. Lewis, then you are legally prohibited from using partisan information in your redistricting efforts — even if it comes from Dave’s Redistricting. As for the rest of us, we’ll be watching.
What You’ll See When You’re Looking at a Map
When you are viewing a map you will see several panels on the left side, along with the map itself taking the larger portion of your screen. The leftmost panel (Overview) lists all of the districts you are building (120 for NC House, 50 for the NC Senate, or 13 for NC US House). This panel shows the total population you’ve put into each district and the deviation from the target population for the district. You can click the radio buttons in this panel to change which district you want to Color.
The next panel (District) shows various sections of Census data and election data for whichever district is selected in the Overview panel. You can show fewer sections by clicking the Edit icon (little pencil) and choosing which you want to hide.
You’ll notice when you hover the mouse over the map another panel (Hover) appears on the left. This shows Census and election data for the precinct over which you are hovering. You can adjust the Hover panel’s position on your screen by clicking the Show dropdown and selecting Adjust. Then check Adjust Hover Detail Position, click the Move icon on the panel and click again where you want the panel.
Check out View Options to show the background map, county lines and more.
What You’ll See When You Analyze a Map
When you press the “Analyze” button, the analysis is divided into two parts: some metrics for the overall plan and a table of district statistics.
This section analyzes the map or plan overall. Right now, there are two subsections.
The “Map Validations” subsection tells you whether the plan is complete, contiguous, and free of holes, and whether districts have roughly equal population. You can also determine many of these from the district statistics table described below.
The second subsection reports on two metrics that gauge “Traditional Districting Principles:”
- Population Deviation — This is reported both as a percentage, e.g., 0.24% and a number between [0–100], e.g., 93 / 100. If the deviation is greater than the allowable threshold — 10% for state legislative districts — the score is 0. If the deviation is zero, then the score is 100. For deviations in between, the score is scaled between 0 and 100. Like all our [0–100] normalized scores, bigger is better. The maximum population deviation between districts in terms of numbers of people is also reported.
- Compactness — The geometric compactness of districts is assessed using both Reock and Polsby-Popper metrics. Each is calculated and reported in raw form, e.g., 0.4191, as well as a score between [0–100], e.g., 68 / 100. The raw numbers are scaled to the [0–100] range, using historical data.
- The two scores are combined to produce a single overall [0–100] compactness score.
In the district statistics table, you get row for each district 1–N and for each several columns of statistics. The row for dummy “unassigned” district at the top that holds VTDs that haven’t been assigned to districts yet, and there is a summary row for statewide statistics at the bottom described below.
The statistics for each district are:
- Total — The total population from the Census dataset noted above. Use this to gauge the degree to which districts have equal populations.
- Δ% — The population deviation for that district from the target district size, which is the average population of a district (total population divided by the number of districts).
- OK? — Yes or No, depending on whether the population deviation for the district is within the generally accepted threshold for legislative districts (which is 10% or less).
- * — Yes, if the district is both contiguous and not fully embedded within any other district; No if it is either not contiguous or is embedded within another district (i.e., like a donut hole).
- Dem — The Democratic share (%) of the two-party (not total) vote, using the election composite noted above.
- Rep — The Republican share (%) of the two-party vote. This is simply 100% minus the Democratic share, of course.
- White — The white-only (i.e., non-Hispanic) voting-age population (VAP) of the district as a percentage of the total VAP for the district.
- Minority — The total minority VAP of the district, as a percentage of the total VAP for the district, where minority is the total VAP minus the white-only VAP.
- Black — The Black VAP as a percentage of the total VAP for the district, including all people who reported being Black alone or in combination.
- Hispanic — The Hispanic VAP as a percentage of the total VAP for the district, including all people who reported being Hispanic alone or in combination.
- Pacific — The “Native Hawaiian and Other Pacific Islander” VAP as a percentage of the total VAP for the district, including all people who reported being that alone or in combination.
- Asian — The Asian VAP as a percentage of the total VAP for the district, including all people who reported being Asian alone or in combination.
- Native — The “American Indian and Alaska Native” VAP as a percentage of the total VAP for the district, including all people who reported being that alone or in combination.
Use the first four columns to check whether the plan is valid:
- Is it complete? — Are all VTDs assigned to a district? The total unassigned population should be zero, and each district should have at least one VTD assigned to it (i.e., population greater than zero).
- Is it contiguous and free of holes? — Are all districts contiguous? And are none embedded within other districts?
- Do districts have roughly equal population?
Use the two partisan shares — Dem and Rep — to analyze how districts are likely to split, Democratic or Republican.
Use the six racial and ethnic breakdowns — Black, Hispanic, Pacific, Asian, Native, and total Minority — to analyze the opportunity for minority representation.
The last row of the table includes additional useful state-level information:
- Total — The target district size.
- Δ% — The overall population deviation percentage. For legislative districts, this generally needs to be 10% or less.
- OK? — Yes or No, depending on whether the overall population deviation is within threshold.
- Dem and Rep — The statewide Democratic and Republican shares (%) of the two-party vote.
- Minority, Black, Hispanic, Pacific, Asian, and Native — The statewide proportions of these demographic categories as percentages of the total statewide VAP.
We significantly accelerated our original 2020-oriented launch plan so that the public would have access for North Carolina’s 2019 redistricting process, and so we offer these capabilities with some caveats:
- Analytics features are only enabled for North Carolina, and only for state legislative districts.
- We hard-wired the district population equality checker to only check for state standards (10%). Until this is changed, don’t believe it if you’re drawing Congressional maps for NC (0.75%).
- Our metrics currently focus on map validations, traditional redistricting principles, and partisan and racial/ethnic composition. In the future we plan to cover partisan fairness, county splitting, the opportunity for minority representation, and more.
- And finally, while it’s on our roadmap to support (fine-grained) census blocks, today we only support (the chunkier) precincts. If you plan to begin your redistricting work with someone else’s maps, it’s important that you understand the Important Note below. If you’re making your own maps from scratch, then it’s not so important.
DRA 2020 does not yet support census blocks. It only supports precincts (aka voting districts or VTDs), which are larger than census blocks. As a result, if you make a map from a block assignment file (such as one provided by the NC General Assembly) in DRA 2020, it should be considered an approximation, because it probably includes precincts that are split between multiple House or Senate districts (some might claim this is an indicator of partisan intent.) When assigning a precinct to a district, we generally pick the district from the first census block we see when loading the block assignment file.
This approximation can impact the detailed shapes of some districts in DRA 2020, as well as influence some validation tests, giving false alarms on district compactness or contiguity. In the case of the 2017 NC House maps, of the 2,692 precincts, 49 were carved up by the General Assembly, and so our approximation methodology potentially impacts all of the other census blocks that were in those 49 precincts. That’s a total of 2,261 census blocks, out of nearly 290,000 census blocks in NC or <1%. But it’s still a difference you should be aware of.
Note that maps that are drawn entirely in DRA 2020 do not have this approximation problem, since our current support for precincts prevents splitting precincts between House or Senate districts.
Dave’s Redistricting helps the public engage proactively in congressional and legislative redistricting in all 50 states, with free mapping tools, data, and analytics.
Here at Dave’s Redistricting, we think gerrymandering sucks. It simply makes it too hard for even the most earnest politician to hear what’s important to we, the people. That’s why we’re introducing a whole new suite of world-class mapping tools, data, and analytics — to empower the public to proactively engage in the redistricting process for the first time, on a nationwide basis. You see, no matter how partisan or ideological you are, only 5% of Americans support gerrymandering. It’s an affront to democracy. At Dave’s, we have a clear ideology. We’re in favor of democracy. We’re in favor of free and fair redistricting. And we hope you are too.
We are big believers in transparency. When the public understands what’s going on in redistricting, before it all goes down, we think the voice of the people will be heard loud and clear. After all, when politicians gerrymander, they screw all of us, in a big way. It’s not Democrats vs Republicans. It’s power versus the people. And the people won’t stand still in the social media era. Armed with insight, and data, and algorithms, the people will rally to protect the American family from those who seek to destroy our hallowed democracy. Gerrymandering is cheating. It undermines the constitutional foundations of our freedom. Simply put, leaders who support gerrymandering are not public servants, they are self-servants.
Dave’s Redistricting is a Washington State non-profit, with plans to file for 501(c)3 status.