We do many things at the state level in the United States: We elect senators and governors, pay taxes, pay in-state college tuition at state-defined rates, get married, get driver’s licenses, and follow state speed limits. But it’s not always best to do everything at the state level. Perhaps COVID-19 policies should not always be implemented at the state level.
The COVID-19 pandemic created and exacerbated political fissures within states. For instance, in Georgia in the spring of 2020, Savannah mayor Van Johnson unsuccessfully tried to impose a mask-wearing mandate to keep Savannah’s visitors and residents safe. Atlanta mayor Keisha Lance Bottoms also tried to invoke a mask-wearing mandate in Atlanta, Georgia; and Georgia Governor Kemp sued the city for overreach. Meanwhile, Southwestern Georgia was suffering devastating rates of COVID-19 deaths following a super-spreader funeral that became a nationwide lesson. Local government asked Kemp for intra-state regional groups that could produce their own responses to the virus. But they had no luck with that request.
Geographic theory tells us that in a networked society, states—as units that do not always follow human behavior—may not be the proper unit of analysis for the administration of policies to mitigate infectious diseases. States are heterogeneous in terms of population density, locations of vulnerable populations, numbers of at-risk institutions such as nursing homes, and mobility behaviors. State boundaries are rigid; they are often crossed for commutes, especially when metropolitan areas straddle multiple states. States are also not sufficiently granular for emergency messaging; the National Weather Service issues warnings and watches for storms, hurricanes, tornados, and even wind advisories at the county level. When it comes to health, logistical improvements in zones and regions can save lives: In one forward-thinking example, state zones were scrapped for more ‘sensible’ organ-donation zones.
COVID-19 response and policy coordination should be applied to regions of well-connected counties. Because movement and social-network ties correlate with the spread of COVID-19, we try to define these regions by leveraging natural human networks of commuters, daily trips, domestic migrants, and social networks. Although the concept of bottom-up regions from human behavior is not new—classic studies in regional science have delineated cultural regions and functional regions for decades—we use a larger suite of data sets and newer network modularity methods to create regions.
What We Did
We detected regions using input data of migrants from the U.S. Census American Community Survey, commuters from the U.S. Census LODES-LEHD data set, GPS traces from Safegraph, Twitter co-mentions, and social regions from Facebook. All data is at the U.S. county level for the 48 contiguous U.S. states. The networks are undirected and include self-edges. After testing various modularization (i.e., community-detection) algorithms through R’s igraph package, we used the Louvain clustering method to partition the networks.
Figure: Resulting regions from the community-detection algorithm for five different input networks (and states). We show the numbers of regions in parentheses. These regions are our ‘independent’ variables; they have different numbers of regions and often follow state boundaries.
Our ‘dependent’ variables are COVID-19 cases, case rates (specifically, cases per 1,000 people), and case-rate differences from the New York Times COVID-19 Data. We calculate these variables for pairs of adjacent counties. There are a total of 8150 such pairs. We calculate each case count by computing the sum of the cases in two adjacent counties. We calculate each case rate by computing the sum of the cases divided by the sum of the populations in two adjacent counties. Each case difference is the absolute value of the difference between the combined case rates of the two counties in a pair. We want cases (C), case rates (CR), and case differences (CD) to be larger, larger and smaller, respectively, within these newly devised regions than they are betweenthese regions because this would suggest that these regional boundaries are preventing spillovers of COVID-19 infections. We used median values and an ANOVA test to determine whether these rates are different across and between regions. Because we had five sets of regions and a set of states, we have six different cases. We also use three temporal ‘waves’, in accordance with peaks and troughs of cases, that coincided with the school year.
Figure: This figure walks through the steps of our analysis to create regions from network data and to determine whether these regions’ boundaries serve as natural boundaries for COVID-19 cases.
We evaluated whether the COVID-19 cases (C), case rates (CR), and case-rate differences (CD) are different for pairs of adjacent counties that lie within regions (w) and between (b) regions. The regions that we construct from mobility networks (in our case, GPS traces and commuters) tend to result in well-connected regions with natural divisions between regions. Facebook boundaries perform the worst in this regard, followed by Twitter boundaries and state boundaries. The regions that we constructed using migration data gave mixed results.
Table: Median values of cases (C), case rates (CR), and case-rate differences (CD) for pairs of adjacent counties both within regions (w) and between regions (b). We show this data for our “third wave” (Aug 31 2020—Jul 1 2021).
But are these differences significant? An ANOVA test shows that there are significant differences across regions for case rates between regions—indicating that some borders (specifically, the commutes and the trips, followed by state boundaries) statistically perform better than others. There is little evidence for variation in regions for cases, and there is no difference for case-rate differences. We also calculated a quantity that normalizes cases by the potential for case crossings. This is the ratio of cases between and within regions (Cb/Cw) divided by the potential for a set of adjacent counties between or within regions (Eb/Ew). This quantity ((Cb/Cw)/(Eb/Ew)) reflects the effectiveness of each region at “containing” cases and is smallest (and hence best) for commutes and trips.
We used network-partitioning methods to build regions from the “bottom up” and found that those that we constructed from movement data can be used to delineate COVID-19 core regions more effectively than states. These resulting smart regions are tailor-made to reflect the behaviors that transmit disease. We hope these findings can help policy-makers develop county-level pacts (i.e. regions), and apply homogenous policies to these regions.
A note on replicability
For the input networks, the migration and commuter data sets are public information from the U.S. Census. The Facebook regions, Twitter data, and Safegraph traces must be acquired by contacting the owners. However, all R and python code is open source and will be made available for use upon publication of the manuscript that is associated with this project. We have also created an online tool to allow users to create their own regions by weighting input network data.
The purpose of this website, https://cybergisxhub.cigi.illinois.edu/
is to provide information to You related to “CyberGISX Project” (“Project”). This Project aims to enable open access and sharing of computation- and data-intensive geospatial analytics, under Apache 2.0 license agreement.
The initial version of the Project (“Original Project”), was developed by the CyberGIS Center for Advanced Digital and Spatial Studies (CyberGIS Center) and is solely owned by The Board of Trustees of the Illinois of Illinois.
The Site and its content is made available by the Board of Trustees of the Illinois of Illinois (“Illinois”, “we”, “our”, “us”), on behalf of the CyberGIS Center. The Site is owned by the Illinois.
or otherwise accessing the Site. Please read these Terms carefully. The Terms govern Your use of the Site and any incorporated content (such as Illinois or third party software, text, data, information, or graphics) made available through the Site.
By continuing to use this site, You accept these Terms. If You are not willing to be bound by these Terms, please stop using this Site.
1. Governing Terms:
The Terms are governed by Illinois’s web privacy notices and policy links presented here:
You hereby accept that all intellectual property, including copyrights, and other proprietary rights in or related to Original Project and Site are, and will remain, the exclusive property of Illinois, whether or not specifically recognized or perfected under applicable law. As End User, You will not take any action that jeopardizes Illinois’s proprietary rights.
3. Credit and Attribution.
You agree to credit and attribute the authors and creators of Project that You use with the copyright notice or statement of credit/attribution as customarily acceptable in industry.
In no event shall Illinois or its employees be liable for any damages (including, without limitation, damages for loss of data or profit, or due to business interruption) arising out of the use or inability to use the content on Site, even if Illinois or a Illinois authorized representative has been notified orally or in writing of the possibility of such damage. Because some jurisdictions do not allow limitations on implied warranties, or limitations of liability for consequential or incidental damages, these limitations may not apply to you.
5. Modification of Site and Terms.
Illinois reserves the right to modify the Site and may discontinue, temporarily or permanently, the Site for any reason, at its sole discretion, with or without notice to You. Illinois may likewise change the Terms from time to time with or without notice to you. You agree to review the Terms periodically to ensure that you are aware of any modifications. Your continued access to the Site after the modifications have become effective shall be deemed your conclusive acceptance of the modified Terms.
If you violate or fail to comply with any of these Terms or in Illinois’ sole discretion restrict or inhibit any other user from using or enjoying the Site, Illinois may terminate your access to the Site account without notice to You.
CyberGISX Project Contributor License Agreement
Version 1.0 February 9, 2021
Thank you for your interest in contributing to the CyberGISX Project. In order to contribute, you will need to provide your name and contact information and sign this CyberGISX Project Contributor License Agreement, which sets the terms and conditions of the intellectual property license granted with your contributions.
This CyberGISX Project Contributor License Agreement (“Agreement”) is by and between you, any person or entity (“You” or “Your”) and the Board of Trustees of the University of Illinois, a body corporate and politic of the State of Illinois (“Illinois”). Please read this document carefully before signing and keep a copy for your records. By signing this Agreement or making a “Contribution” to the “CyberGISX Project” as defined below (even if You do not sign), You agree to the following:
1. “CyberGISX Project” is an open-source software project that aims to enable open access and sharing of computation- and data-intensive geospatial analytics with computational reproducibility and transparency support available under Apache 2.0 license agreement: https://www.apache.org/licenses/LICENSE-2.0
2. “Contribution” means all of Your contributions of object code, source code, content, and documentation, and any modifications thereof, to the CyberGISX Project.
3. “Licensed Patents” mean patent claims licensable by You which are necessarily infringed by the making, using, selling, offering for sale, having made, import, or transfer of either Your Contribution alone or when combined with the CyberGISX Project.
4. “Representation.” You represent the following to the best of your knowledge:
a. You are at least 18 years of age and have full power and authority to enter into this Agreement and to grant the rights in and to the Contribution as set forth herein (individuals who are under 18 years of age and who wish to contribute to the CyberGISX project may not enter into this Agreement, but may contact CyberGISX at firstname.lastname@example.org to explore alternatives);
b. If your employer has rights to intellectual property that You create as part of the Contribution, You represent that You have obtained permission from Your employer to make the Contribution on behalf of that employer, or Your employer waived any rights in and to Your Contribution, or Your employer authorizes the Contribution and agrees to be bound by the terms herein by signing as an entity below;
c. That either:
1. all documentation and code in the Contribution is Your original work and includes complete details of any third-party license and any other restriction (including, but not limited to related patents and trademarks) of which You are personally aware and which are associated with any part of Your Contributions; or
2. any part of the Contribution that is not Your original creation is submitted to CyberGISX separately from any original Contribution, includes the complete details of its source and any corresponding license and any other restriction (including, but not limited to related patents, trademarks, and license agreements) of which You are personally aware, and is conspicuously marked as "Submitted on behalf of a third party: [named here]".
d. That Your Contribution does not include any viruses, worms, Trojan horses, malicious code or other harmful or destructive content;
e. The CyberGISX Project delivered under this Agreement may be subject to U.S. export control laws and may be subject to export or import regulations in other countries. You agree to comply strictly with all such laws and regulations and acknowledge that You have the responsibility, at Your own expense, to obtain such licenses to export, re-export, or import as may be required; and
f. Your Contribution does not include any encryption technology and no government license or permission is required for the export, import, transfer or use of the Contribution.
5. Notification. Your representations are accurate, and You agree to notify Illinois of any facts or circumstances of which You become aware that would make any of Your representations inaccurate in any respect.
6. License grant to Contributions. You grant to Illinois and to recipients of the CyberGISX software (collectively, “Recipients”), a perpetual, irrevocable, non-exclusive, worldwide, royalty-free unrestricted license to use, reproduce, prepare derivative works of, publicly display, publicly perform, distribute, and sublicense the Contribution, and such derivative works, in source code and object code forms.
7. License grant to Patents.
8. No support. Except for the rights granted to Recipients above, You reserve all right, title, and interest in and to Your Contribution. You are not expected to provide support for your Contributions.
9. No Warranties. Subject to Your representations above, Your Contributions are provided on an “AS-IS” basis WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND (express or implied), including, without limitation, any implied warranty of merchantability and fitness for a particular purpose, and any warranty of non-infringement.
10. No Liability. Illinois, its trustees, directors, officers, employees, students, and agents have no liability for any infringement of any copyright, patent, or other right of third parties in connection with any Contributions, the CyberGISX Project, or CyberGISX software, and are not liable for any direct, indirect, punitive, special, incidental, consequential, or exemplary damages arising in connection with any Contribution, the CyberGISX Project, or CyberGISX software.
11. Indemnity. You agree to indemnify, defend, and hold Illinois and its subsidiaries, affiliates, officers, agents, employees, partners, licensees, and licensors harmless from any claim or demand, including but not limited to reasonable attorneys’ fees, made by any third party due to or arising out of Contributions You submit, post, transmit or otherwise make available through the CyberGISX Project, as well as Your use of the CyberGISX software; connection to the CyberGISX Project; or violation of any rights of another.
12. You agree that Illinois may assign this Agreement and CyberGISX Project to any third party without notice or consent.
13. Illinois is under no obligation to accept or include Your Contribution to the CyberGISX Project.
14. This Agreement is governed by the laws of the State of Illinois, excluding its conflict of laws provisions.