Political Reapportionment: Drawing Boundaries with QGIS
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 351 | |
| Author | ||
| License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/69113 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Production Year | 2022 |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
FOSS4G Firenze 2022131 / 351
1
7
13
22
25
31
33
36
39
41
43
44
46
52
53
55
58
59
60
76
80
93
98
104
108
127
128
133
135
141
142
143
150
151
168
173
176
178
190
196
200
201
202
204
211
219
225
226
236
242
251
258
263
270
284
285
292
00:00
Information technology consultingStatisticsThermodynamisches SystemReduction of orderMathematicsOpen sourceOpen sourceMathematicsPlug-in (computing)Thermodynamisches SystemFrequencyAreaRepresentation (politics)Beta functionDistribution (mathematics)Information technology consultingRevision controlState of matterOrder of magnitudeSlide ruleNumerical analysisCASE <Informatik>FeedbackReduction of orderGoodness of fitBitLevel (video gaming)MappingSoftwareComputer animation
03:38
Open sourceClosed setInformationBound statePlug-in (computing)SoftwareThermodynamisches SystemAlgebraData typeField (computer science)Computer fileNumber theoryEquals signEvolutionarily stable strategyPolygonScale (map)Personal digital assistantTime zoneCharacteristic polynomialCodierung <Programmierung>Address spaceNeighbourhood (graph theory)Ideal (ethics)Level (video gaming)Thermodynamisches SystemProduct (business)Open sourceField (computer science)SoftwareEqualiser (mathematics)Core dumpCycle (graph theory)PlanningAttribute grammarCASE <Informatik>System identificationPolygonTable (information)EmailComputer fileMappingClosed setAreaLatent heatSlide ruleRule of inferenceInstance (computer science)1 (number)Web 2.0State of matterSelectivity (electronic)Plug-in (computing)View (database)Term (mathematics)Different (Kate Ryan album)Computer configurationNumerical analysisMultiplication signMathematical analysisLoop (music)Total S.A.Uniform resource locatorWhiteboardDecision theoryInformation technology consultingPower (physics)Goodness of fitWeb applicationMathematicsPresentation of a group2 (number)BitProcess (computing)Menu (computing)Computer programTheoryVotingDesign by contractScaling (geometry)Execution unitInformationReal-time operating systemFlow separationLine (geometry)Vector spacePosition operatorSystem callType theoryOrder (biology)Zoom lensNegative numberOrder of magnitudeGroup actionMorley's categoricity theoremQuicksortGraph coloringOpen setResultantFunctional (mathematics)Software developerBlock (periodic table)Grass (card game)Physical lawStandard deviationSystem administratorLimit (category theory)Chemical equationElectronic mailing listStatisticsGreatest elementLattice (order)Representation (politics)SequelOnline helpPoint (geometry)Computer animationDiagram
Transcript: English(auto-generated)
00:01
Good morning everybody and thank you for coming. I'm John Holden and this is Blake Esselstyn. I am a consultant who developed the QGIS redistricting plug-in and Blake is a consultant who kept sending in most of the feedback and helping getting the plug-in from
00:20
something that was a beta version into something that was used internationally to redraw political boundaries. So before we jump into how the plug-in works, I'd like to talk a little bit about what is redistricting, why is it important, and why is it important that we have open source software to
00:44
actually be able to redraw these boundaries. And for that I'd like to jump over to Blake. Can everyone hear me now?
01:08
Great. So in many places representative democracy requires that districts be drawn so that each representative is representing a geographic area. And these geographic areas in many cases are
01:25
expected to have equal populations. And as this slide indicates, populations change. So this is North Carolina in the United States. As John mentioned, this plug-in has been used internationally. Our examples will focus on the United States, which is where where I worked.
01:43
This is the changes from 2010 to 2020. You can see the the green shades of green indicate growth, shades of yellow and orange indicate population reduction. While the state of North Carolina had exceptional growth, it was that the state grew by about 10% over this 10-year period.
02:00
But you can see that more than half of the 100 counties, actually 51 of the counties, actually had a reduction in population. So over this 10-year period any districts covering the state would have to be redrawn so that they would each each representative would be representing an approximately equal population. The shades here represent the percent change in population. In
02:22
this slide the the height of the extruded prisms represents the actual numeric magnitude of change. So you can see there are two counties here. This is Mecklenburg County and Wake County. These are the counties where the cities of Charlotte and Raleigh are located, had exceptional growth, over 20% growth. And
02:43
the the distribution of population change very unequal. So this brings about the challenge of redrawing districts so that the the area contains, each area contains the same number of people. This was the previous congressional map. This is in the north in, sorry, in the United States. This is the
03:01
redistricting realm that gets the most attention, statewide congressional district mapping. And in the elections that were held in 2020, this was the map that was used for 13 congressional districts. In 2020, because of that population growth, North Carolina actually got a 14th congressional district. And these are the districts that will be used in the upcoming 2022 elections in North Carolina.
03:25
So this is just an example to show how widely, how dramatically the districts can change. I alluded earlier to the city of Raleigh, which is in Wake County here. Other examples will be focused on that county. As I said, it had more than 20% growth.
03:42
Each of the polygons here has a roughly equal population. These are all voting districts. And you can see that it's also unevenly distributed across the county. Again, the colors represent percentages, but the height of these prisms represent magnitudes. So working for Wake County, I had the challenge of taking their existing nine districts for a county-wide board.
04:06
And if you look at this, this is just another way of representing how uneven the populations were. This solid line here is the ideal population, basically the population of the whole county divided by nine. And the dotted line and the dashed line represent the upper limit and lower limit, plus or minus 5%. And we had to redraw the
04:25
districts to get all those populations within that plus or minus 5% acceptable deviation. So that's just a preview to kind of how this works, and then John is going to talk more about the software.
04:44
So I identified a need for good open source redistricting software back over a decade ago in 2010 when I was in law school in the United States, and we used open source online software to win a redistricting competition. And
05:01
unfortunately, it was web-based, and I was a grass GIS user. I also used QGIS at the time, but it was still up and coming at that point. And in 2017, I was working in Australia, and I was talking with a consultant, and he said, we love QGIS. We use it all the time,
05:22
but there's one thing that we still use map info for, and that's to redraw the political boundaries. And I thought, well, there's, you know, there's PyQGIS, like there's ways of being able to implement this in QGIS so that you don't have to use map info anymore, and you don't have to keep maintaining a very old piece of software just for one specific use case. And so
05:45
also being 2017, the main United States redistricting cycles happen pretty much every 10 years. With litigation, these boundaries are going to get redrawn pretty much consistently over the next 10 years as well, but the main data dumps happen with each U.S. census every 10 years. And so in 2017,
06:05
I thought, well, if I can make this work for Australia, it will also be able to work for the United States in a couple of years. And one of the biggest problems with redistricting in terms of software issues was that the main
06:22
pieces of software that people use were Maptitude, which is desktop software, and Map Info, also desktop software. And the open source software was typically designed specifically for United States users and specifically for the web, including District Builder, Districtor, some of the ones that were used in this cycle. And those are very good pieces of software, and there's no reason
06:46
to use any of the closed source software, although people do like Dave's redistricting for the United States, which is also a web-based piece of software. But as a consultant, I really wanted something that would work on the desktop, and I wanted something that would work within QGIS,
07:05
and I wanted something robust that could work in any country in the world that produced polygons that included population data that would then be used to redraw district boundaries. And another thing that was important, and that's been important for everybody who's developed the open source software, is citizen participation.
07:23
So, especially in the United States, this is a process that consultants with specific subject matter expertise would be drawing the maps for, and that's one of the reasons why gerrymandering takes place so much is because they're typically employed by a political party, and it doesn't matter which political party it is, unfortunately. There's no good political party.
07:42
Everybody is trying to maximize the amount of seats they can win without having to campaign, and being able to open up the process to citizens to submit their own maps should, in theory, lead to better quality maps and more competitive districts. So, the way that this
08:00
redistricting software works, it's actually very simple in concept. So, if there's any sequel programmers here, it's basically a select some population group by district. So, you've got a categorical district column, you've got a numerical population column that's typically provided to you in the vector files provided by the government, and
08:24
then you have to allow the user to categorize those polygons really easily and instantly display the population changes. And as simple as it is, it was an interesting piece of software to develop. And so, the way that you would actually use the software is you would download the plug-in in QGIS.
08:43
You would download a population file. I have downloaded the state of Oregon's census block file. This is the most granular vector file that you can download, and it's all everybody's done a very good job of trying to make this data as publicly accessible this cycle as possible in the United States. And
09:00
if you look at the attributes file, your attributes table, the it's got the identification, it's got the county, and it's got different columns, and the header of the column represents the population as determined by a lookup table for the US Census. So, for this
09:20
example, I'm using the P0010001 column, which I, if I remember correctly, is the total amount of people in each individual polygon. So, you would create a new district plan, and you'd select a name, you select the layer, you select the number of districts that you're trying to create, and you select the equal population tolerance,
09:44
which is, in the US, is down to one person, and in other countries, and in state redistricting, there's a little bit more tolerance, and you would select that. And then, I had the feature request to add a second optional population field, because, in some countries, depending on the rules for redistricting, not only do you have to take into account
10:03
how many people are living in the district at the time, but they also have to take into account future population growth, to make sure that, by the time that it's time to draw the next political boundaries, there's not going to be one district that has way too many people in another district that is completely underpopulated. And
10:21
then, you would select the district field, and then any other fields that you would like to sum, and if you're using another, especially a country-specific piece of redistricting software, this will be, most of the other data fields will automatically be shown, and that's sort of one of the negatives and positives of the QGIS tool,
10:41
is that you are completely dependent on the data that you are importing to the file, and then you have to define all of the columns that you want on your own. And this is so that it can work as broadly as possible, in as many countries as possible, with as many files as possible. But at the same time, you might lose some
11:00
specific information about, say, partisan makeup that would automatically be shown to you by some of the other pieces of software. Also, there's a little checkbox, and if you check it, it will automatically style your map, which was, that was actually one of the, it didn't take very long, but that was one of the more fun pieces to code.
11:21
So, an example of how you categorize the districts. So, I have zoomed in, this is North Portland, Oregon, and this is, I can't do a product demonstration because of the rules of FOSS 4G. So, imagine as if I have selected a specific area, and
11:41
this is the slide demonstrating that I have now selected all of these polygons, and then I would click the update selected polygons button with the active district, and then I would assign all of those polygons to, for in this case, District 1, and the attributes would automatically update, saying that there are
12:00
47,000 people living in this currently assigned district, and that there's, we're still 93% short of creating a population, or polygon, or district with an equal population file. And you would repeat this loop as many times as you needed, and there's also a number of different features that helps you
12:23
draw your polygons. For instance, if you have a geographic selection, or geographic information on your polygon, so for instance, this has counties, you can select an entire county at once by selecting the column that contains that information, and then clicking select by geography, and it will automatically select all of those.
12:41
So you can create a map very, very quickly using this, and because it's in QGIS, you've got the full functionality of QGIS. So for instance, there's an XYZ layer that with containing OpenStreetMap, and if typically when I'm working on this, I would have a little bit of transparency on my layer,
13:00
so I could see exactly where I am assigning the districts. You can also preview the proposed changes so that you don't have to commit until you're happy with the changes. Previewing is especially important if you're trying to get down to that last equal population by person. You can see how the district is going to change before you actually
13:25
commit those changes to the district. So it's a very simple, but very powerful piece of software, and it's been used in a number of different places. It's been used to reapportion several Australian states. I've received support requests from Canada because they are currently in the process of
13:46
redrawing their districts. I believe they have a contract with Esri, so I think a lot of the requests I've been giving are from individual citizens, and I've used it to consult in the United States as has Blake, and you can use this at any different scale from the school board redistricting all the way up to federal congressional redistricting.
14:05
Some of the plans, as I have said, for instance in Australia, they have been adopted. I have used them to help lawyers figure out exactly what a specific plan would look like and then redraw the plan in real time to see how the changes would work, typically on a Zoom call because of the pandemic.
14:25
And then I've also seen citizens download and use the plug-in in order to try to create a plan that they would want to present to try to make that politically palatable to adopt. And as an example use case, I'll turn it back over to Blake to talk about how this was used in North Carolina.
14:43
Okay, and I'm gonna try, I'll stay and use this mic and ask John to advance the slides. So this is a slide we saw before. Again, this is Wake County, a population of about 1.1 million people. And as the bar chart shows, when the census data arrived, we could see that the districts were
15:03
significantly out of balance. And the next slide shows, and don't worry, I'm not going to expect everyone to read all this, this is just an idea of some of the criteria that we were expected to consider. So the board in this case, the nine representatives in public meetings, decided what the criteria that should be considered would be as we were drawing these districts. And again,
15:24
I'm not going to go into detail on all of these, but if you look at the bottom of the list, number eight is taking into account expected population growth. That's something that John mentioned. Number nine, consideration of other administrative areas. And number 10, consideration of the locations of schools.
15:41
So this was a Board of Education that was very focused on education-related geography. And I will repeat another thing John said, that there are great other open source products out there, like District Builder, which was talked about in a session at nine o'clock this morning, and Districter.
16:00
But as someone who works a ton with QGIS and wanted to be able to customize my views and bring in other layers and do other analysis within the same tool, looking at these multiple criteria, it was invaluable to be able to work in QGIS and then be going back and forth with the redistricting options.
16:21
This slide just shows various ways of trying to anticipate what the future population growth would look like, so I brought those layers in. Next slide. This was another way of just trying to understand the administrative boundaries, existing administrative boundaries in the areas that would be covered by different superintendents. Next slide.
16:42
This again, trying to harmonize. These are the precinct boundaries and other administrative boundaries and constantly people talk about, wouldn't it be better to do redistricting algorithmically? But in my experience, it takes a lot of human judgment and just familiarity with
17:00
what the considerations are and trying to make value judgments. This is a screenshot of one of the while I was working on this. This was kind of in process. You can see some of the different layers there. I found it helpful to label the individual units, in this case the voting precincts,
17:20
not by their population, but by their percentage of the ideal population. So as I was shifting one from another, rather than seeing that it had, you know, 5,211 people, I could see the percentage of the ideal district and that made it easier for me to conceptualize as I was trying to balance those. So having this desktop tool, again, that I could
17:42
customize in any number of ways really helped with my process. And I think the next slide is just showing the final result. So these are the districts that will be used in the upcoming election. Some of the boundaries may look quirky, but afterwards, if you want to talk, I can explain why some of the the boundaries aren't as
18:01
compact as they might otherwise be. And I'll hand it back over to John. So one of the important things in deciding to develop something for QGIS that would let people redraw the political boundaries was opening up the amount of access that people had. In 2010, there was lots of technology that was enabling lots of people to redraw boundaries very quickly,
18:26
but it was all very expensive and it was not accessible. And one of the challenges now that we have many different types of open source tools that can let you redistrict is actually still the political process. And even though there have been
18:41
many, many citizen maps developed in both open source and closed source software for the United States redistricting cycle, the redistricting process is still out of the hands of citizens and is going to take a political change in order to become less gerrymandered. So just having the technological capability to redraw the boundaries does not necessarily mean that
19:04
we are going to get better outcomes. However, I believe it's a very important step in that process because we now know exactly what better maps look like. Maps that aren't gerrymandered, maps that contain competitive geographically compact districts, that keep communities together as opposed to some of
19:21
the maps that are drawn by the politicians in the United States. And for instance in other countries in Australia, it's still a very political process where political parties will submit maps to a centralized commission that redraws the boundaries. Nobody really seems to have that many problems with the political boundaries in Australia.
19:41
But again, it's still not necessarily a process that's open to citizens. And so the next step in opening up, now that we have the tools to be able to draw these boundaries in an open source manner, the next step is now not a technological step, but instead a political step to try to make sure that not only do people know that these tools exist and can draw the
20:03
maps with these tools, but that the maps that are drawn can be taken seriously by the people who make the decisions on what the final maps will be. So if you're interested in playing around with the plugin, it's available through the QGIS plugins menu. It's the Stata Redistrictor, and there is a tutorial available at QGIS redistricting.com.
20:25
If you like it, please give it a few stars and a good review. And if you have any questions, feel free to ask either me or Blake, and thank you very much for your attendance. We greatly appreciate it.
20:49
Thank you very much for your presentation, both of you.