ArticlePDF Available
Full Terms & Conditions of access and use can be found at
Journal of Maps
ISSN: (Print) (Online) Journal homepage:
Normalizing the pandemic: exploring the
cartographic issues in state government COVID-19
Aaron M. Adams, Xiang Chen, Weidong Li & Chuanrong Zhang
To cite this article: Aaron M. Adams, Xiang Chen, Weidong Li & Chuanrong Zhang (2023)
Normalizing the pandemic: exploring the cartographic issues in state government COVID-19
dashboards, Journal of Maps, 19:1, 1-9, DOI: 10.1080/17445647.2023.2235385
To link to this article:
© 2023 The Author(s). Published by Informa
UK Limited, trading as Taylor & Francis
Group on behalf of Journal of Maps
Published online: 27 Jul 2023.
Submit your article to this journal
Article views: 121
View related articles
View Crossmark data
Normalizing the pandemic: exploring the cartographic issues in state
government COVID-19 dashboards
Aaron M. Adams
, Xiang Chen
, Weidong Li
and Chuanrong Zhang
Department of Geography, University of Connecticut, Storrs, CT, USA;
UConn GIS Health Lab at Connecticut Childrens, Hartford, CT, USA;
Institute for Collaboration on Health, Intervention, and Policy (InCHIP), University of Connecticut, Storrs, CT, USA;
Center of
Environmental Sciences and Engineering, University of Connecticut, Storrs, CT, USA
Government agencies have utilized Web Geographic Information Systems (GIS) dashboards to
collect and disseminate spatial information on COVID-19. However, not all maps on these
dashboards adhere to established cartographic principles. This article explores the extent of
the cartographic issues by surveying state governmentsocial COVID-19 websites in the
United States on February 11, 2021. The results indicate that out of the fty states, thirty-
one (62.0%) incorrectly used unnormalized data in choropleth maps, sixteen (32.0%) used
normalized data, and three (6.0%) did not employ choropleth maps. Among states using
normalized data correctly, we identied other cartographic problems, including
inappropriate data class divisions and suboptimal enumeration units. As dashboards serve
as authoritative sources for health information, issues in map creation can inuence public
perception of the health crisis. These ndings underscore the need for map standards to
ensure the accuracy and reliability of health information in the Web GIS era.
Received 6 November 2022
Revised 28 April 2023
Accepted 26 June 2023
COVID-19; choropleth; web
GIS; dashboard; infodemic
1. Introduction
The coronavirus disease 2019 (COVID-19) pandemic
has disrupted nearly every aspect of our society and is
responsible for an unprecedented public health cala-
mity in the United States (Centers for Disease Control
and Prevention, 2021). Since the disease information
is organized spatially, governments have employed
geospatial information technology to monitor and
respond to the virus spread in close to real-time
(Kamel Boulos & Geraghty, 2020). With Web Geo-
graphic Information Systems (GIS) that automate
map creation and distribution, public users and auth-
orities can quickly develop, share, and update health
information using web maps quickly enough to make
actionable decisions (Richards, 1999;Zhang et al.,
2015;Zhang & Li, 2005). One of the most widely used
tools to host web maps is the dashboard, a hosted web
service that facilitates interactive visualization of spatial
and non-spatial data (Dong et al., 2020;Grin, 2020).
After the initialization of the COVID-19 dashboard by
Johns Hopkins University (Dong et al., 2020), all fty
United States state governments created their own
dashboards to enhance pandemic surveillance and
facilitate health communication. Dashboards became
the default method for communicating relevant infor-
mation involving the pandemic with the general public,
and each state made dierent design decisions
(Geraghty & Artz, 2022).
Although the COVID-19 dashboard is considered
the most striking cultural artifactof the pandemic
(Everts, 2020), considerable issues arise as they do
not always follow cartographic principles (Adams
et al., 2020). In this study, we observed that the domi-
nant map type used in dashboards is the choropleth
map, which is a thematic map type using the intensity
of colors to correspond to data values within spatial
enumeration units (Dent, 1990;Tobler, 1973). Choro-
pleth maps are one of the most popular thematic map
types, and some cartographic principles have been well
established to ensure their proper use, such as data
normalization and map symbology principles (Brewer
& Pickle, 2002;Harrower & Brewer, 2003;Jenks, 1963;
Jenks & Caspall, 1971). However, dashboard develo-
pers, including those employed by an authoritative
agency, may not have the essential training to comply
with these principles (Harrower & Brewer, 2003;Juer-
gens, 2020;Lan et al., 2021;Plewe, 2007).
One prevailing cartographic issue in choropleth-
based dashboards is the failure to use normalized data
for mapping that is, using a relative value (e.g. infec-
tion rates) rather than an absolute value (e.g. cases of
infection) (Adams et al., 2020;Engel et al., 2022;
© 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of Journal of Maps
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (, which
permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been
published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
CONTACT Xiang Chen Department of Geography, University of Connecticut, Storrs, CT 06269, USA; UConn GIS Health Lab
at Connecticut Childrens, Hartford, CT 06106, USA
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
2023, VOL. 19, NO. 1, 2235385
Kronenfelda & Yoo, 2020). Violating this principle may
disguise spatial patterns due to comparing absolute
values in dierent sizes of enumeration units (Dent,
1990;Krygier & Wood, 2005;Monmonier, 2018;Rezk
& Hendawy, 2023). In the case of the COVID-19 pan-
demic, research has shown that choropleth maps that
show only totals may change peoples perception of
the pandemic (Engel et al., 2022). Therefore, it is rec-
ommended not to use totals in a choropleth map
(Adams et al., 2020). If totals need to be depicted,
other map symbolizations, such as proportional sym-
bols, area cartograms, or dot density, may be used
(Brewer, 2016;Dent, 1990;Zhang, 2020). While it is
possible to nd examples in the literature where a chor-
opleth map showing totals is appropriate, that same lit-
erature recommends considering graduated symbols in
these cases (Krygier & Wood, 2005).
Figure 1, created from available data on the Centers
for Disease Control and Prevention (CDC)s COVID-
19 website on May 28, 2022, demonstrates why this mat-
ters with two choropleth maps displaying the samecase
data (Centers for Disease Control and Prevention,
2022). Figure 1(a) shows the total cases of infection by
state, and Figure 1(b) shows infection rates per
100,000 population. As shown in Figure 1(a), Califor-
nia, Florida, New York, and Texas, all have the most
overall cases. There is no coincidence that these four
states also have the largest population in the US, and
Figure 1(a) reects this pattern (Census Bureau,
2022). Figure 1(b) shows the infection rate per
100,000 people, which better facilitates comparisons
among states with small and large populations. Thus,
using an unnormalized map may mislead the public
or decision-makers about the severity of the pandemic;
for example, North Dakotas high infection rate is only
unveiled in the normalized map (Figure 1(b)).
While this cartographic bias has been documented
in the literature, it has not been well recognized by
most public health professionals and map readers.
For example, while the CDC has switched to publish-
ing normalized case data, it still uses choropleth maps
to show total cases well into the COVID-19 pandemic
(Centers for Disease Control and Prevention, 2021,
2022). While other reviews of dashboards have looked
at the issue and others, including visualizations
besides maps and the frequency of updates (Clarkson,
2023;Fareed et al., 2021;Kronenfelda & Yoo, 2020),
they may not contain data for the entire duration of
the pandemic due to the dynamic nature of web maps.
To this end, this paper aims to articulate the carto-
graphic issues in the COVID-19 dashboards published
by all fty state governments in the United States.
Specically, on February 11, 2021, we examined
whether state governments employed a choropleth
map or another map form for publishing COVID-19
case data. Then, we identied if these ocial web
maps followed fundamental cartographic principles,
including data normalization, number of classes, color
schemes, and enumeration unit selection. On April 8,
2023, following the United States Senate vote to end
the pandemic, we returned to each Uniform Resource
Locator (URL) to determine the number of states still
hosted the COVID-19 dashboards. This review can
serve as a partial record of how the state governments
of the United States portrayed spatial data during the
COVID-19 pandemic for future researchers. Even-
tually, we hope eorts can be made to improve the con-
sistency and accuracy of health information delivery.
2. Methods and results
To evaluate the extent of the cartographic issues in
COVID-19 maps, we identied all fty statesocial
websites where case data were published in terms of
a dashboard or other types of web maps (Table 2).
We excluded Puerto Rico, Guam, Washington D.C.,
Figure 1. (a) Total infections by state and (b) infection rates in terms of cases per 100,000 people by state. Case data were derived
from the CDC as of May 28, 2022. Class breaks were created using Jenks natural breaks in ArcMap, and labels are rounded to two
signicant digits for simplicity.
and other parts of the United States not within state
boundaries to limit the study to comparable adminis-
trative units. We then recorded the types of thematic
maps used on the websites, mainly choropleth maps
and proportional symbol maps (i.e. maps using dier-
ent sizes of circles). If a choropleth map was used, we
checked if the cartographers employed normalized
data (e.g. infection rates) or inappropriately used
unnormalized data (e.g. total cases). We recorded
dashboard URLs along with the data collection,
which took place on February 11, 2021.
As Table 1 illustrates, all states employed certain
forms of COVID-19 dashboards to visualize case
data. Forty-seven states (94.0%) used choropleth
maps, while two states (4.0%), Texas and Wyoming,
used proportional symbol maps. The state of Nebraska
had a general reference map but lacked any form of
thematic map related to COVID-19 cases, with its
website indicating that a map was under production.
Colorado was the only state employing both choro-
pleth and proportional symbol maps.
Figure 2 visualizes the ndings by state. We found
that thirty-one states (62.0%) used unnormalized data
in their choropleth maps in at least one map on their
dashboard, sixteen states (32.0%) rigorously stuck to
normalized data in their choropleth maps, and three
states (6.0%) did not employ choropleth maps. We
observed that at least eleven states, including Alaska,
Alabama, Arizona, Hawaii, Indiana, Michigan,
Mississippi, North Carolina, North Dakota, West Vir-
ginia, and Wisconsin, employed choropleth maps
showing both normalized and unnormalized cases,
which we recorded as using unnormalized choropleth
maps. We also found that the CDC had both normal-
ized and unnormalized data on its web maps (Centers
for Disease Control and Prevention, 2021). We
acknowledge that as dashboards often have multiple
visualizations, we may have missed a map that was
either normalized or unnormalized. Such an omission
would increase the number of states using both but
not decrease the scale of the normalization issue.
These ndings demonstrate that the problem of mis-
using choropleth maps to visualize total COVID-19
cases was widespread.
To investigate further, we separately evaluated the
three categories of normalized choropleth, not choro-
pleth, and unnormalized choropleth. For the sixteen
states that mapped normalized data only, we further
examined their normalization methods. This follow-
up evaluation identied if other cartographic prin-
ciples regarding map symbology in a thematic map
were followed, such as the number of classes, color
schemes, and enumeration units. Table 2 shows the
results: (1) the most popular normalization method
was a ratio of cases in an enumeration unit (n= 14,
87.5%); only two states (12.5%) employed dierent
normalization methods: California was based on a 7-
day risk level, and South Dakota was based on the
Figure 2. Data normalization status on state COVID-19 dashboards by state.
Table 1. Thematic map types and data normalization status on state COVID-19 dashboards.
State Map type
Using unnormalized
choropleth maps Variable(s)
Number of
Color schemes
(low to high) Enumeration unit
Choropleth Yes Total cases 7 Blue County
Choropleth Yes Total cases Continuous Red Borough/Census
Choropleth Yes Total cases No data Red County
Arkansas Choropleth Yes Total Cases Continuous Blue County
California Choropleth No Risk level (based on 7-day
positivity rate)
4 Yellow to purple County
Colorado Choropleth/
No Cases per 100,000 population Continuous Blue County
Connecticut Choropleth No Cases per 100,000 population 2 Grey to red Town
Delaware Choropleth No 7-day cases per 100,000
6 Blue Zip Code
Florida Choropleth Yes Total cases No data Yellow to Blue County
Georgia Choropleth No 14-day cases per 100,000
Continuous Yellow to Red County
Choropleth Yes 14-day total cases 5 White to orange.
One no-data
Zip Code
Idaho Choropleth Yes Total cases No data White to orange County
Illinois Choropleth Yes Total cases No data Blue Zip code
Choropleth Yes Total cases No data White to orange County
Iowa Choropleth No 14-day cases per 100,000
population / 7-day cases per
100,000 population
5 Blue County
Kansas Choropleth Yes Total cases No data Blue County
Choropleth No 7-day cases per 100,000
4 Green to red County
Louisiana Choropleth Yes Total cases 4 Blue Census Tract
Maine Choropleth Yes Total cases 4 Pink Zip Code
Maryland Choropleth Yes Total cases 3 Blue County
Massachusetts Choropleth Yes Total cases 5 Blue County
Choropleth Yes Total cases 5 Blue County
Minnesota Choropleth Yes Total cases 7 Blue County
Choropleth Yes 7-day cases per 100,000
5 Blue County
Missouri Choropleth No Cases per 100,000/7-day cases
per 100,000 population
Continuous Red to orange Jurisdiction
(Appears to be
Montana Choropleth Yes Total active cases 4 Blue County
Nebraska Reference map N/A N/A N/A N/A County
Nevada Choropleth No Cases per 100,000 population 6 Blue County
Choropleth Yes Total cases 6 Red to orange.
One no-data
New Jersey Choropleth Yes New daily cases 7 Blue County
New Mexico Choropleth Yes Total cases Continuous Green County
New York Choropleth Yes Total cases 7 Orange County
Choropleth Yes Total cases 5 Blue County and Zip
North Dakota
Choropleth Yes Total cases No data Red County
Ohio Choropleth Yes Total cases No data Blue County
Oklahoma Choropleth Yes Total cases/total deaths/total
Continuous Yellow Red
Oregon Choropleth No Cumulative cases divided by
5 Blue County
Pennsylvania Choropleth No Cases per 100,000 population 5 Red County
Rhode Island Choropleth No Cases per 100,000 population 6 Yellow to blue Geographic Area or
South Carolina Choropleth No Cases per 100,000 population 5 Blue County
South Dakota Choropleth No Community spread (totals) 3 Blue County
Tennessee Choropleth No Positive tests per 100,000
Continuous Blue and red County
Texas Proportional
N/A Total Cases Continuous N/A County
Utah Choropleth No 14-day case rate per 100,000
7 Yellow to red County
Vermont Choropleth Yes Total cases in the past 14 days No Data blue County
Virginia Choropleth Yes Total cases 6 blue County
Washington Choropleth Yes Total cases 6 Blue County
West Virginia
Choropleth Yes Total cases in past 7 days No data Blue County
Choropleth Yes Total cases 4 Grey County
Wyoming Proportional
NA Total cases Continuous N/A County
Indicates a state where we observed both normalized and unnormalized choropleth maps.
community spread rates (2) We found that the
majority of states with normalized choropleths on
their dashboards (n= 12, 75.0%) employed discrete
class breaks, while four of these states (25.0%) used a
continuous color scheme. Choosing map symbology
is more exible than the need for data normalization
but still demands scrutiny from a cartographic per-
spective (Monmonier, 2018). Generally, the literature
suggests using discrete class breaks over continuous
color schemes for making a thematic map, as it is
easier to discern the dierence between data values
(Brewer & Pickle, 2002;Dobson, 1973;Krygier &
Wood, 2005). (3) When mapping with discrete class
breaks, a general rule is to have no less than three
and no more than seven classes best to distinguish
classied data values (Harrower & Brewer, 2003).
This rule was followed by ten (91.7%) of the eleven
states mapping only normalized data and discrete
class breaks, with most maps having ve classes. (4)
In terms of color hues, seven of the states mapping
only normalized data (43.8%) employed a cold color
hue (e.g. green, blue, and purple), ve (31.3%)
employed a warm color hue (e.g. red, orange, yellow),
and four (25.0%) had a mixed use of cold and warm
colors. (5) For enumeration units, twelve (75.0%)
used county, Delaware used ZIP codes, Connecticut
and Rhode Island used local enumeration units such
as town or municipality, and Missouri used Jurisdic-
tions,which upon inspection appeared to correspond
to the county boundaries.
The choice of enumeration unit is critical. Due to
the modiable aerial unit problem (MAUP), dierent
ways of subdividing an area can inuence the nal
aggregate values (Chen et al., 2022). Counties in
many states are administrative units where policy is
made. In some smaller states, like Connecticut and
Rhode Island, counties are often ignored in favor of
smaller administrative units, such as towns (Chen
et al., 2021). These smaller units allow for ner-scale
analysis. ZIP codes, like towns, are also generally smal-
ler than counties; however, their use in epidemiological
mapping is controversial (Chen et al., 2022). ZIP codes
are often discontinued or modied, do not cover the
entire United States, and their representation as poly-
gons is often not representative of what they cover
(Grubesic & Matisziw, 2006). Therefore, ZIP codes
for the analysis of health data should be avoided, if
possible in favor of census tracts or another more
meaningful enumeration unit, and used cautiously in
consideration of their limits only when unavoidable
(Chen et al., 2022;Grubesic & Matisziw, 2006).
Next, we reviewed the unnormalized maps of the
thirty-one states mapping totals in choropleths to see
if the state dashboards followed other cartographic
principles regarding map symbology. Below are our
ndings: (1) the most mapped variable was cumulative
cases per enumeration unit (n= 25, 80.6%), two states
(6.5%) mapped total cases in 7-day windows, and two
states mapped total cases in 14-day windows. New Jer-
sey mapped total daily cases, and Montana mapped
total active cases.(2) We found that the majority of
states with unnormalized choropleths on their dash-
boards (n= 17, 54.0%) employed discrete class breaks.
In comparison, four states (12.9%) used a continuous
color scheme. As previously stated, the general consen-
sus is that a discrete color scheme may be more easily
understood than a continuous one. We have no data
on class breaks for ten (32.3%) of these states due to
missing legends. (3) All seventeen states employing dis-
crete class breaks used between three and seven classes,
with most maps using ve class breaks, all aligning with
the categorizing pricinple of using 37classes.(4)In
terms of color hues, twenty states (64.5%) employed a
cold color hue (e.g. green, blue, and grey), ten states
(32.2%) employed a warm color hue (e.g. red, orange,
yellow), and one state (3.2%) had a mixed use of cold
and warm colors. (5) As for enumeration units, twenty
(80.6%) used county, three (9.6%) used ZIP codes, New
Hampshire used local enumeration units of town, and
Alaska used the local unit of Borough/Census Area.
North Carolina had both county and ZIP codes avail-
able as options for users to map the case data; however,
as the county was the default visualization, we counted
as using them. One state, Louisiana, used census
tracts in their published maps. As census tracts are
smaller than counties, used throughout the United
States, and created with consideration of human popu-
lations, this choice is highly in line with recommen-
dations in the literature. Surprisingly, among all 50
states, only Louisiana created maps using census tracts
at this point in the pandemic.
For the three states that did not use choropleth
maps on their dashboards (i.e. Nebraska, Texas, and
Wyoming), all mapped using counties. Texas and
Wyoming employed continuous class breaks for
their proportional symbols, which were blue circles.
Nebraskas map did not show COVID-19-related
data at the time of the survey.
On April 8, 2023, nine days after the United States
Senate voted to end the COVID-19 emergency
declaration, we reviewed the list of URLs and dash-
boards we used in this study. First, we conrmed
that the CDC and Prevention COVID-19 data tracker
we based Figure 1 on still allowed users to view chor-
opleth maps displaying total cases; however, a normal-
ized option still exists, as previously observed (Centers
for Disease Control and Prevention, 2021). Next, we
found that thirteen (26%) of previously identied
state dashboard URLs no longer led to a publicly
facing dashboard (Table 2). Five enforced a sign-in
procedure for map viewing, while the others were una-
vailable. It is likely that as time progresses, more of
these links will no longer function, contiributing to a
phenomenon known as link rot (Klein et al., 2014).
3. Discussion and conclusion
Since the outbreak of the COVID-19 pandemic, Web
GIS technologies, particularly dashboards, have pro-
vided unprecedented opportunities for sharing health
information. Choropleth maps are overwhelmingly
favored to visualize COVID-19 case data in these dash-
boards (Mooney & Juhász, 2020). Unfortunately, our
ndings reveal that more than half of the states did not
rigorously follow fundamental cartographic principles,
such as data normalization, to create thematic maps.
Even among those states that mapped with appropriate
data, we identied other cartographic issues, such as
less-than-ideal numbers of classes, color schemes, and
inappropriate choice of enumeration units. These
ndings raise serious concerns regarding Web
mapping as they serve as an authoritative outlet for
delivering health information. The lack of adherence
Table 2. Availability of the URLs for state COVID-19 dashboards.
URL at the time of the survey
(February 11th, 2021)
URL availability on follow-up
(April 8, 2023)
Arkansas No
California Yes
Colorado Yes
Connecticut Yes
Delaware Yes
Florida No
Georgia Yes
Illinois No
Indiana Yes
Iowa No
Kansas Yes
Louisiana Yes
Maine Yes
Maryland Yes
Massachusetts Yes
Michigan Yes
Minnesota Yes
Mississippi,21882,420,873.html No
Missouri Yes
Nebraska No
Nevada Yes
Hampshire Yes
New Jersey Yes
New Mexico Yes
New York
North Carolina Yes
North Dakota Yes
Ohio Yes
Oklahoma No
Oregon No
Pennsylvania Yes
Rhode Island Yes
South Carolina No
South Dakota Yes
Tennessee Yes
Utah Yes
Vermont Yes
Virginia Yes
Washington Yes
West Virginia Yes
Wisconsin Yes
Indicates a dashboard URL now points to an ArcGIS Online sign-in page, and thus may still exist there but be unavailable.
to cartographic principles in map creation could unex-
pectedly mislead public perception of the pandemics
impact (Engel et al., 2022;Geyer & Lengerich, 2023). If
these dashboards were used to assist policymakers,
there is a possibility that a biased epidemiological pat-
tern arising from the maps could have a lasting policy
impact. This review helps to showcase the extent of
these problems as part of the infodemic surrounding
On the other hand, we see positive changes in the
creation of dashboards. For instance, while mapping
totals, the Florida dashboard included a note to the
user that comparison of counties is not possible
because case data are not adjusted by population
(Florida Department of Public Health, 2021). Simi-
larly, Connecticut initially used unnormalized data
in its dashboard but later switched to normalized
data in its latest version (Adams et al., 2020).
Beyond normalizing data in choropleth maps,
many other methods could be employed to improve
the interpretation of epidemiological data, such as
the cartogram, to show the severity of a health out-
come, where the areal unit is altered proportionally
to the population density aected (Roth et al., 2010;
Tobler, 2004;Zhang, 2020). A complementary
approach when using dynamic web maps is to incor-
porate additional information (such as total cases
and total population) in a pop-up window to present
a more comprehensive view of the health data when
a user clicks on an enumeration unit (Thomas et al.,
2022). Similarly, other visualization methods, such as
dot density maps, hot spot maps based on Getis Ord
Gi* statistic, and relative risk cluster maps created
using Poisson spacetime scan statistic, can also be
employed (Dent, 1990;Desjardins et al., 2020;Getis
& Ord, 1992). These visualization and statistical tech-
niques can open new avenues to displaying epidemio-
logical data from multifaceted perspectives without
using a choropleth to display absolute values.
This article focuses on observing choropleth maps
and whether they adhere to established cartographic
conventions on United States State government
ocial dashboards. Our ndings are consistent with
other studies that have identied widespread misuse
of choropleths throughout the pandemic (Adams
et al., 2020;Engel et al., 2022;Everts, 2020;Kronen-
felda & Yoo, 2020). With this article, we hope that
public health agencies may take the necessary steps
to monitor how data are collected (Tao et al., 2020),
comply with map-making principles, and integrate
other important demographic metrics, such as age
and sex, when making maps (Kontis et al., 2020).
Importantly, we suggest that health professionals, pol-
icymakers, and cartographers should be included in
the discussion when constructing these public-facing
web maps (Plewe, 2007;Rushton et al., 2000). These
combined eorts may help improve health
communication in future health crises as we are nor-
malizing life with this pandemic.
All gures were produced using ESRI ArcMap 10.7.1.
We thank Adam Gallaher, Ashley Benitez Ou, Dr. Debs
Ghosh, and Rich Mrozinski for supporting data collection
for this research.
Disclosure statement
No potential conict of interest was reported by the author(s).
This research was supported by pilot grant funding from the
Institute for Collaboration on Health, Intervention, and Pol-
icy (InCHIP) at the University of Connecticut.
Data availability
We collected dashboard data on February 11, 2021,
from the United States State government dashboards.
We revisited this dataset on April 8, 2023, to deter-
mine the number of links that were still active. Bound-
ary les were sourced from the United States Census
Bureau. The COVID-19 case data were obtained
from the CDC at
Adams, A., Li, W., Zhang, C., & Chen, X. (2020). The dis-
guised pandemic: The importance of data normalization
in COVID-19 web mapping. Public Health,183, 36.
Brewer, C. A. (2016). Designing better maps: A guide for GIS
users. Esri Press.
Brewer, C. A., & Pickle, L. (2002). Evaluation of methods for
classifying epidemiological data on choropleth maps in
series. Annals of the Association of American
Geographers,92(4), 662681.
Census Bureau. (2022). U.S. and world population clock.
Retrieved June 24, 2022 from:
Centers for Disease Control and Prevention. (2021).
COVID-19 cases, deaths, and trends in the US | CDC
COVID Data Tracker. Centers for Disease Control and
Prevention. Retrieved January 28, 2021 from: https://
Centers for Disease Control and Prevention. (2022). United
States COVID-19 cases, deaths, and laboratory testing
(NAATs) by state, territory, and jurisdiction | CDC
COVID data tracker. Centers for Disease Control and
Prevention. Retrieved May 28, 2022 from: https://covid.
Chen, X., Ye, X., Widener, M. J., Delmelle, E., Kwan, M.-P.,
Shannon, J., Racine, E. F., Adams, A., Liang, L., & Jia, P.
(2022). A systematic review of the modiable areal unit
problem (MAUP) in community food environmental
research. Urban Informatics,1(1), 22.
Chen, X., Zhang, A., Wang, H., Gallaher, A., & Zhu, X.
(2021). Compliance and containment in social distan-
cing: Mathematical modeling of COVID-19 across town-
ships. International Journal of Geographical Information
Science, 35(3), 120.
Clarkson, M. D. (2023). Web-based COVID-19 dashboards
and trackers in the United States: Survey study. JMIR
Human Factors,10, e43819.
Dent, B. D. (1990). Cartography: Thematic map design. W.C.
Desjardins, M. R., Hohl, A., & Delmelle, E. M. (2020). Rapid
surveillance of COVID-19 in the United States using a
prospective space-time scan statistic: Detecting and eval-
uating emerging clusters. Applied Geography,118,
Dobson, M. W. (1973). Choropleth maps without class
intervals? A comment. Geographical Analysis,5(4), 358
Dong, E., Du, H., & Gardner, L. (2020). An interactive web-
based dashboard to track COVID-19 in real time. The
Lancet Infectious Diseases,20(5), 533534. https://doi.
Engel, C., Rodden, J., & Tabellini, M. (2022). Policies to
inuence perceptions about COVID-19 risk: The case
of maps. Science Advances,8(11), eabm5106. https://doi.
Everts, J. (2020). The dashboard pandemic. Dialogues in
Human Geography,10(2), 260264.
Fareed, N., Swoboda, C. M., Chen, S., Potter, E., Wu,
D. T. Y., & Sieck, C. J. (2021). U.S. COVID-19 state gov-
ernment public dashboards: An expert review. Applied
Clinical Informatics,12(02), 208221.
Florida Department of Public Health. (2021). Floridas
COVID-19 data and surveillance dashboard. Retrieved
January 24, 2021 from:
Geraghty, E. M., & Artz, M. (2022). Learning from COVID-
19: GIS for pandemics. Esri Press.
Getis, A., & Ord, J. K. (1992). The analysis of spatial associ-
ation by use of distance statistics. Geographical Analysis,
24(3), 189206.
Geyer, N. R., & Lengerich, E. J. (2023). Lionvu: A data-dri-
ven geographical web-GIS tool for community health and
decision-making in a catchment area. Geographies,3(2),
Grin, A. L. (2020). Trustworthy maps. Journal of Spatial
Information Science,20(1), 519.
Grubesic, T. H., & Matisziw, T. C. (2006). On the use of ZIP
codes and ZIP code tabulation areas (ZCTAs) for the
spatial analysis of epidemiological data. International
Journal of Health Geographics,5(1), 58.
Harrower, M., & Brewer, C. A. (2003). An
online tool for selecting colour schemes for maps. The
Cartographic Journal,40(1), 2737.
Jenks, G. F. (1963). Generalization in statistical mapping.
Annals of the Association of American Geographers,53(1),
Jenks, G. F., & Caspall, F. C. (1971). Error on choroplethic
maps: Denition, measurement, reduction. Annals of
the Association of American Geographers,61(2), 217
Juergens, C. (2020). Trustworthy COVID-19 mapping: Geo-
spatial data literacy aspects of choropleth maps. KN
Journal of Cartography and Geographic Information,70
(4), 155161.
Kamel Boulos, M. N., & Geraghty, E. M. (2020).
Geographical tracking and mapping of coronavirus dis-
ease COVID-19/severe acute respiratory syndrome coro-
navirus 2 (SARS-CoV-2) epidemic and associated events
around the world: How 21st century GIS technologies are
supporting the global ght against outbreaks and epi-
demics. International Journal of Health Geographics,19
Klein, M., Van de Sompel, H., Sanderson, R., Shankar, H.,
Balakireva, L., Zhou, K., Tobin, R., & Bar-Ilan, J.
(2014). Scholarly context not found: One in ve articles
suers from reference rot. PLoS ONE,9(12), e115253.
Kontis, V., Bennett, J. E., Rashid, T., Parks, R. M., Pearson-
Stuttard, J., Guillot, M., Asaria, P., Zhou, B., Battaglini,
M., Corsetti, G., McKee, M., Di Cesare, M., Mathers, C.
D., & Ezzati, M. (2020). Magnitude, demographics and
dynamics of the eect of the rst wave of the COVID-
19 pandemic on all-cause mortality in 21 industrialized
countries. Nature Medicine,26(12), 19191928. https://
Kronenfelda, B. J., & Yoo, K. (2020). An evaluation of
COVID-19 dashboards from cartographic and epidemio-
logical perspectives. Autocarto 2020 Presentations:
Hazard identication, perception, and management.
Krygier, J., & Wood, D. (2005). Making maps: A visual guide
to map design for GIS. Guilford Press.
Lan, Y., Desjardins, M. R., Hohl, A., & Delmelle, E. (2021).
Geovisualization of COVID-19: State of the art and
opportunities. Cartographica: The International Journal
for Geographic Information and Geovisualization,56(1),
Monmonier, M. S. (2018). How to lie with maps. The
University of Chicago Press.
Mooney, P., & Juhász, L. (2020). Mapping COVID-19: How
web-based maps contribute to the infodemic. Dialogues
in Human Geography,10(2), 265270.
Plewe, B. (2007). Web cartography in the United States.
Cartography and Geographic Information Science,34(2),
Rezk, A. A., & Hendawy, M. (2023). Informative carto-
graphic communication: A framework to evaluate the
eects of map types on usersinterpretation of COVID-
19 geovisualizations. Cartography and Geographic
Information Science,118.
Richards, T. B. (1999). Information technology. Geographic
information systems and public health: Mapping the
future. Public Health Reports,114(4), 359373. https://
Roth, R. E., Woodru, A. W., & Johnson, Z. F. (2010).
Value-by-alpha maps: An alternative technique to the
cartogram. The Cartographic Journal,47(2), 130140.
Rushton, G., Elms, G., & McMaster, R. (2000).
Considerations for improving geographic information
system research in public health. URISA Journal,12(2),
Tao, R., Downs, J., Beckie, T. M., Chen, Y., & McNelley, W.
(2020). Examining spatial accessibility to COVID-19 test-
ing sites in Florida. Annals of GIS,26(4), 319327. https://
Thomas, C., Valdizon, S., & Artz, M. (Eds.). (2022).
Mapping community health: GIS for health & human
services. Esri Press.
Tobler, W. R. (1973). Choropleth maps without class inter-
vals? Geographical Analysis,5(3), 262265. https://doi.
Tobler,W.R.(2004). Thirty ve years of computer cartograms.
Annals of the Association of American Geographers,94(1),
Zhang, C., & Li, W. (2005). The roles of web feature and web
map services in real-time geospatial data sharing for
time-critical applications. Cartography and Geographic
Information Science,32(4), 269283.
Zhang, C., Zhao, T., & Li, W. (2015). Geospatial semantic
web. Springer International Publishing.
Zhang, T. (2020). Integrating geographic information sys-
tem technique with Google Trends data to analyse
COVID-19 severity and public interest. Public Health,
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
In 2018, the Penn State Cancer Institute developed LionVu, a web mapping tool to educate and inform community health professionals about the cancer burden in Pennsylvania and its catchment area of 28 counties in central Pennsylvania. LionVu, redesigned in 2023, uses several open-source JavaScript libraries (i.e., Leaflet, jQuery, Chroma, Geostats, DataTables, and ApexChart) to allow public health researchers the ability to map, download, and chart 21 publicly available datasets for clinical, educational, and epidemiological audiences. County and census tract data used in choropleth maps were all downloaded from the sources website and linked to Pennsylvania and catchment area county and census tract geographies, using a QGIS plugin and Leaflet JavaScript. Two LionVu demonstrations are presented, and 10 other public health related web-GIS applications are reviewed. LionVu fills a role in the public health community by allowing clinical, educational, and epidemiological audiences the ability to visualize and utilize health data at various levels of aggregation and geographical scales (i.e., county, or census tracts). Also, LionVu is a novel application that can translate and can be used, for mapping and graphing purposes. A dialog to demonstrate the potential value of web-based GIS to a wider audience, in the public health research community, is needed.
Full-text available
Background: The SARS-CoV-2 pandemic provided an opportunity to employ public-facing online data visualization tools to help citizens understand the evolving status of the outbreak. Given the heterogeneity of data sources, developers, tools, and designs employed in this effort, it raises questions about how visualizations were constructed during a time that daily batches of data were available but issues of data quality and standardization were unresolved. Objective: This report surveys online COVID-19 dashboards and trackers likely to be used by residents of the United States to monitor the spread of infection at local, national, and global scales. The work is intended to provide insights that will help application developers to increase the usefulness, transparency, and trustworthiness of dashboards and trackers for public health data in the future. Methods: Websites of coronavirus dashboards and trackers were identified in August 2020 using the Google search engine. They were examined to determine the data sources used, types of data presented, types of data visualizations, characteristics of the visualizations, and issues with "messy" data. The websites were surveyed three more times for changes in design and data sources, with the final survey in June 2022. Themes were developed to highlight issues concerning challenges in presenting COVID-19 data and techniques of effective visualizations. Results: A total of 111 websites were identified and examined (84 state-focused, 11 nation-wide, and 16 with global data), plus an additional 17 later added for state vaccination data. This work documents how data aggregators have played a central role in making data accessible to visualization developers. The designs of dashboard and tracker visualizations vary in type and quality, with some well-designed displays supporting interpretation of the data and others obscuring the meaning of the data and potentially misleading viewers. Five themes were identified to describe challenges in presenting COVID-19 data and techniques of effective visualizations. Conclusions: This analysis reveals the extent to which dashboards and trackers informing the American public about the COVID-19 pandemic relied upon an ad-hoc pipeline of data sources and data aggregators. The dashboards and trackers identified in this survey offer an opportunity to compare different approaches to the display of similar data.
Full-text available
Geospatial models can facilitate the delineation of food access patterns, which is particularly relevant for urban planning and health policymaking. Because community food environmental studies use different analysis units or study scales, the rigor and consistency of their evaluations cannot be ensured. This issue is known as the modifiable areal unit problem (MAUP). The paper provides a systematic review of past literature on place-based community food environmental research using different analysis units or geospatial models as they pertain to the MAUP. We identify these key findings: (1) the ZIP code zone is not recommended as an appropriate analysis unit for modeling community food access, as it did not have significant correlations with health indicators; (2) using a circular buffer of less than 0.5 km around household locations is most likely to reveal health correlations, compared with network buffers or container-based measures; (3) to reveal health effects of the community food environment, it is recommended to focus in selected regions or partitions of a study area with similar socioeconomic statuses, such as the central city or low socioeconomic status areas; (4) for studies utilizing a single statistical unit or distance measure, it is suggested to discuss the existence of the MAUP, such as evaluating the sensitivity of the model to the change of the unit or the distance measure. By highlighting the MAUP, this paper has policy implications—given that geospatial modeling of food accessibility provides support for health policy intervention, using different metrics may lead to different interpretations of health disparities and could thus misinform policy decisions. Therefore, any assessment of community food environments that may potentially lead to a policy change should consider the effects of the MAUP.
Full-text available
Choropleth disease maps are often used to inform the public about the risks posed by coronavirus disease 2019 (COVID-19). In a survey conducted in the U.S. state of Georgia in June 2020, we randomly assigned respondents to view either of two maps. The first map reported county-level COVID case counts; the second displayed case rates per 100,000 people. Respondents who saw case rate maps were less likely to perceive COVID as mostly an urban problem and reported higher levels of concerns about the virus. Case rate maps also increased support for policies aimed at mitigating the spread of the virus, although, for this outcome, the effect was quantitatively small and the maps did not change individuals' self-reported behavior. For several outcomes, the impact of the case rate map was strongest for rural residents and self-identified Republicans, both of whom were less worried about the virus and more skeptical about public health measures to mitigate its spread.
Full-text available
This consists of a few extracts from the Making Maps book. The "GPS" in the title is a typo and should be "GIS"
Full-text available
In the early development of COVID-19, large-scale preventive measures, such as border control and air travel restrictions, were implemented to slow international and domestic transmissions. When these measures were in full effect, new cases of infection would be primarily induced by community spread, such as the human interaction within and between neighboring cities and towns, which is generally known as the meso-scale. Existing studies of COVID-19 using mathematical models are unable to accommodate the need for meso-scale modeling, because of the unavailability of COVID-19 data at this scale and the different timings of local intervention policies. In this respect, we propose a meso-scale mathematical model of COVID-19, named the meso-scale Susceptible, Exposed, Infectious, Recovered (MSEIR) model, using town-level infection data in the state of Connecticut. We consider the spatial interaction in terms of the inter-town travel in the model. Based on the developed model, we evaluated how different strengths of social distancing policy enforcement may impact epi curves based on two evaluative metrics: compliance and containment. The developed model and the simulation results help to establish the foundation for community-level assessment and better preparedness for COVID-19.
Full-text available
Mapping the prevalence and spread of infectious diseases has never been more critical than during the COVID-19 pandemic. A plethora of web-based GIS dashboards have been created that incorporate basic GIS functionality; these dashboards have served as platforms for rapid data sharing and real-time information, ultimately facilitating decision making. However, many of them have merely focused on presenting and monitoring cumulative or daily incidence of COVID-19 data, disregarding the temporal dimension. In this paper, we review the usefulness of GIS-based dashboards for mapping the prevalence COVID-19, but also missed opportunities to emphasize the temporal component of the disease (cyclicity, seasonality). We suggest that advanced geovisualization techniques can be used to integrate the temporal component in interactive animated maps illustrating (a) the daily relative risk and the number of days a geographic region has been in a disease cluster, (b) the ratio between the observed and expected number of cases over time and (c) mortality count dynamics in a space-time cube. We illustrate these approaches by using COVID-19 cases and death counts across the U.S. at the county level from January 25th, 2020 to October 1st, 2020. We discuss how each of these visualization approaches can promote the understanding of important public health concepts applied to the pandemic such as risk, spread and mortality. Finally, we suggest future avenues to promote research at the intersection of space-time visualization and infectious diseases.
The dynamic and rapid spread of COVID-19 presented a challenge in explaining complex issues to uninformed members of society. Although the conveying of this information was dependent on web maps, few studies have focused on the effectiveness of these maps for informing the general public. This study employs the map usability literature to define three map design dimensions: (1) clarity, (2) understanding, and (3) accuracy. Using these dimensions, we developed a framework for categorizing geovisualizations into scientific, rhetorical, and informative maps based on their purpose. We analyzed four examples of COVID-19 geovisualizations under this framework and conducted an online survey to evaluate their interpretation by users. Only 10% to 60% of users understood each map correctly and poor design practices significantly affected the user interpretation of the maps. The degrees of understanding conformed to our framework. Scientific maps were ambiguous owing to their complexity and the over-emphasis on map cartographic accuracy, whereas informative maps that balance accuracy, clarity, and understanding were more interpretable. Meanwhile, no rhetorical COVID-19 dashboards were included in our sample. We demonstrated that one map cannot provide a complete and comprehensive overview. Therefore, our framework calls for the conscious design of different types of maps to convey the different shades of cartographic reality.
Background In the United States, all 50 state governments deployed publicly viewable dashboards regarding the novel coronavirus disease 2019 (COVID-19) to track and respond to the pandemic. States dashboards, however, reflect idiosyncratic design practices based on their content, function, and visual design and platform. There has been little guidance for what state dashboards should look like or contain, leading to significant variation. Objectives The primary objective of our study was to catalog how information, system function, and user interface were deployed across the COVID-19 state dashboards. Our secondary objective was to group and characterize the dashboards based on the information we collected using clustering analysis. Methods For preliminary data collection, we developed a framework to first analyze two dashboards as a group and reach agreement on coding. We subsequently doubled coded the remaining 48 dashboards using the framework and reviewed the coding to reach total consensus. Results All state dashboards included maps and graphs, most frequently line charts, bar charts, and histograms. The most represented metrics were total deaths, total cases, new cases, laboratory tests, and hospitalization. Decisions on how metrics were aggregated and stratified greatly varied across dashboards. Overall, the dashboards were very interactive with 96% having at least some functionality including tooltips, zooming, or exporting capabilities. For visual design and platform, we noted that the software was dominated by a few major organizations. Our cluster analysis yielded a six-cluster solution, and each cluster provided additional insights about how groups of states engaged in specific practices in dashboard design. Conclusion Our study indicates that states engaged in dashboard practices that generally aligned with many of the goals set forth by the Centers for Disease Control and Prevention, Essential Public Health Services. We highlight areas where states fall short of these expectations and provide specific design recommendations to address these gaps.