Edinburgh Research Explorer
Design Patterns for Data Comics
Citation for published version:
Bach, B, Wang, Z, Farinella, M, Murray-Rust, D & Henry Riche, N 2018, Design Patterns for Data Comics.
in ACM Conference on Human Factors in Computing Systems (CHI)., 38, ACM, CHI 2018, Montreal,
Canada, 21/04/18. https://doi.org/10.1145/3173574.3173612
Digital Object Identifier (DOI):
10.1145/3173574.3173612
Link:
Link to publication record in Edinburgh Research Explorer
Document Version:
Peer reviewed version
Published In:
ACM Conference on Human Factors in Computing Systems (CHI)
General rights
Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s)
and / or other copyright owners and it is a condition of accessing these publications that users recognise and
abide by the legal requirements associated with these rights.
Take down policy
The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer
content complies with UK legislation. If you believe that the public display of this file breaches copyright please
contact openaccess@ed.ac.uk providing details, and we will remove access to the work immediately and
investigate your claim.
Download date: 09. Aug. 2022
Design Patterns for Data Comics
Benjamin Bach
University of Edinburgh
Edinburgh, UK
bbach@inf.ed.ac.uk
Zezhong Wang
University of Edinburgh
Edinburgh, UK
wangzezhong2016@gmail.com
Matteo Farinella
Columbia University
New York, NY
mf3094@columbia.edu
Dave Murray-Rust
University of Edinburgh
Edinburgh, UK
d.murray-rust@ed.ac.uk
Nathalie Henry Riche
Microsoft Research
Redmond, WA
nath@microsoft.com
ABSTRACT
Data comics for data-driven storytelling are inspired by the
visual language of comics and aim to communicate insights in
data through visualizations. While comics are widely known,
few examples of data comics exist and there has not been
any structured analysis nor guidance for their creation. We
introduce data-comic design-patterns, each describing a set of
panels with a specific narrative purpose, that allow for rapid
storyboarding of data comics while showcasing their expres-
sive potential. Our patterns are derived from i) analyzing
common patterns in infographics, datavideos, and existing
data comics, ii) our experiences creating data comics for dif-
ferent scenarios. Our patterns demonstrate how data comics
allow an author to combine the best of both worlds: spatial
layout and overview from infographics as well as linearity and
narration from videos and presentations.
ACM Classification Keywords
H.5.2 User Interfaces: Theory and methods
Author Keywords
Visualization, Comics, Story-telling
INTRODUCTION
Communicating with data or ‘data-driven storytelling’ is cap-
turing the attention of the visualization research commu-
nity [40, 38]. Crafting visualizations that effectively com-
municate insights on data while keeping the audience engaged
relies on many factors: the characteristics of data, the commu-
nication medium, the narrative and the audience, as well as the
purpose of the communication. There are different common
genres of data-driven storytelling such as magazine style arti-
cles (text+figures), infographics, videos, presentations, as well
as comics [48]. Each genre affords different opportunities and
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full cita-
tion on the first page. Copyrights for components of this work owned by others than
ACM must be honored. Abstracting with cr edit is permitted. To copy otherwise, or re-
publish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from permissions@acm.org.
CHI 2018, April 21–26, 2018, Montréal, QC, Canada.
Copyright is held by the owner/author(s). Publication rights licensed to ACM.
ACM ISBN 978-1-4503-5620-6/18/04 ...$15.00.
http://dx.doi.org/10.1145/3173574.3173612
c)
b)
a)
Figure 1. Data comic example employing design patterns: a) temporal
sequence, b) bar-chart panel, and c) flashback. (©Zezhong Wang)
limitations for communication (e.g. leveraging text as well as
pictures, enforcing linear structure, providing a high-graphical
resolution, supporting reader interactions).
This article focuses on data comics [57, 10], a format inspired
by the visual language of comics and the way people read
and understand graphic narratives. Comics can be broadly
described as sequential art [20] and juxtaposed images and
words [41]: they map time into space through the use of panels
and their ordering. Comics are not yet a common genre of data-
driven storytelling. However, we believe they are promising as
they combine assets from two common practices: they provide
the freedom of 2D spatial layout present in infographics and
annotated charts, while supporting the linearity of narration
naturally enforced by videos and live presentations. Arranging
panels in space enables the reader to gain an overview of the
narrative and its structure (page) while keeping the details
of each of its points (panel content) accessible at any time,
allowing the reader to consume the story at their own pace.
Arranging panels in a sequence guides a reader’s attention and
helps to build a narrative, with a beginning, middle and end,
tension, argument, immersion, and emotion. By filling the
space between those two worlds [27], comics allow an author
to combine the best of both worlds in a single format.
This combination opens up a large potential for expressive
storytelling with data visualizations. While few examples [44,
12] of data comics exist, there are no guidelines, methodolo-
gies or tools that help crafting them. To facilitate the creation
of data comics and to inform the design of interactive and
eventually semi-automated design tools, this paper proposes a
set of design patterns for data comics. A pattern describes a a
set of panels with specific layout and content relation. Thus, a
pattern describes both usages of space and a relation between
the content inside the panels. For example, the data comic
in Figure 1 uses the flashback pattern (c) by introducing a
reference between two non-consecutive panels. Other patterns
include temporal-sequence to depict temporal change (a), of-
ten simplified into small multiples [7]. Such design patterns
provide an extensible set of templates that can help in the
creation phase of comics as well as being used to teach data
comics in classes and workshops.
We composed and refined our set of design patterns by follow-
ing three complementary methods. First, the authors of this
paper—an interdisciplinary team involving a comic artist, an
art graduate, as well as researchers in visualization— created
over ten data comics for different data and audience (Section
3). By crafting and iterating over these comics, we reflected
on an initial set of design patterns that we found useful in the
creation process and that reoccurred across the comics we gen-
erated. Second, we identified a design space to systematically
describe our patterns leading to the description of new patterns
(Section 4). The design space is informed by i) coding spatial
layouts common in the space-oriented genres of infograph-
ics and traditional comics, combined with ii) linear narrative
progression between panels (frames) in the time-oriented gen-
res, building from Hullman et al. [33]. Third, to assess the
value of our design patterns for teaching and creation of data
comics, we ran a workshop with 23 participants creating data
comics for multiple datasets (Section 5). We printed a set of
design patterns cards akin to IDEO [3] and provided them to
the workshop participants. Participants leveraged these cards
to create a set of comics from data we provided them. Subjec-
tive feedback from the workshop suggests that the cards were
useful in the ideation and execution process. All material in
this paper including examples of existing data comics, comics
that inspired our patterns, our own data comics, design cards,
as well as the entire material of the workshop, can be found
online: http://datacomics.net.
BACKGROUND
Genres in Data-Driven Storytelling
The term genre has been coined by Segel and Heer [48] to de-
scribe seven types of narrative visualizations: magazine style,
annotated chart, infographic, video, slide-show, flowchart, po-
sitioned poster, and eventually comics. For the purpose of
this work, we classify some of them as space-oriented (info-
graphics, posters, annotated charts) and others as time-oriented
(videos, slide-shows). Space-oriented genres present and com-
municate through space: layout, size, visual and graphical
elements organized on a canvas, equally and instantaneously
accessible to the reader, leaving it to the reader which elements
to watch, how long, and in which order. The time-oriented gen-
res present and communicate linearly over time; they have an
order with a fixed beginning, a middle, and an end; a develop-
ment over time, an argumentation and they can present change
over time through animations [6]. Observers are locked in
the order and pace in which they perceive the content. Re-
search has shown that narrative structure is intrinsically easier
to remember and facilitate readers engagement and persua-
sion [25]. However, this requires the audience to remember,
to continuously pay attention, and to adapt to the pace chosen
by the presenter. In short, while animated explanations can be
useful to present dynamic content [31], animations often risk
to show “too much, too fast” [54].
A third group of genres can be described as both spatial and
temporal. This includes magazine-style (text with figures)
as well as comics. Comic scholar Groensteen writes that
“[c]omics is not only the art of fragments, of scattering, of distri-
bution; it is also an art of conjunction, of linking together. [26,
p.22]. Comics combine linear narration with non-linear figures
and, as Eisner notes, the “partnership of words with imagery
become the local permutation. The resulting configuration is
called comics and it fills a gap between print and film.” [20,
p.xvii].
Comics for Data-Oriented Storytelling
Comics have been used in facilitating classroom education [32,
52, 50], conveying scientific phenomena [53, 22], and to sto-
ryboard interaction design [19]. These first examples demon-
strate the potential of comics to present scientific information
in a clear and accessible way. The comic page layout allows a
reader to consider information on multiple levels: a detailed
explanation in sequence, as well as browsing the concepts and
explanations simultaneously [51]. A single panel acquires
meaning from its individual content as well as from its po-
sition within the global visual/narrative structure—a process
defined as “braiding” (tressage) [27]. Comics can use tran-
sitions [41], simultaneously providing detail and overview
(co-presence and configuration) [28], thereby preserving vi-
sual permanence [56] and allowing readers to read the story at
their own pace. Besides their linear structure, comics have a
rich history of integrating unexpected logics into their struc-
ture, such as hierarchical linguistic grammars [17], repetition
[49], or musical elements that carry temporality, emotion and
a joyous disregard for staying in their expected place [15].
Data comics support data-driven storytelling by making use
of peoples’ familiarity in reading and understanding comics
along with the particular qualities of the medium [57, 10].
There is not a precise definition of what is or is not a data
comic, rather data comics combine elements from different
genres, presented in a unique visual and narrative framework.
Bach et al. [10] describe four essential components that help
discussing the uniqueness of data comics: the presence of data
visualization; flow, whether linear or non-linear; narration
to include and illustrate contextual information and create
a compelling narrative as opposed to a factual presentation
of charts; and words-and-pictures combined visually using
multiple strategies. The patterns presented in this work are
most close to the dimensions of flow and narration.
While we can find many examples of data stories that include
some elements of comics (juxtaposition, combination of text
and image, etc), very few examples exist that combine these
components into a fully fledged data comic. Nature provides a
high quality example on the challenges of climate change [44]
in 2015 combining data visualization, narration, linear-and-
non linear flow, into a compelling, visually striking and even
emotional piece. Cisneros created an interactive comic using
Tableau Graphics [16]. Additional examples include a scien-
tific report on the change of water temperatures [12] and graph
comics [8]. Part of this may be due to a lack of tool support;
besides Zhao’s and Elmqvist’s editor prototype [57] that al-
lows designers to embellish data visualizations with comic
elements, data comics authors have to craft their comics man-
ually, which generally entails laboriously switching between
data visualization tools and illustration tools [13].
Design Patterns in Education and Design
Packs of cards containing design patterns are a common re-
source for creative practice, supporting creation and analysis
in both classroom and professional situations [3, 2, 1, 21]. For
visualization, a wealth of online collections of visualization
design patterns exist (e.g., [46, 47]), but without a system-
atic scientific foundation. VizItCards is a card set figuring
visualization types and tasks in visualization [29], specifically
designed for education.
For storytelling, Segel and Heer have described general design
strategies [48], including details-on-demand, multi-messaging,
progress bar, and time-slider. These strategies are similar to
our design patterns but do not describe the specific design
space of (static) data comics. For data-driven storytelling,
Bach et al. [11] describe general higher level narrative patterns
for data-driven storytelling (e.g., addressing-the-audience, ex-
ploration, or gradual reveal), some of which were described as
patterns that define the “flow” of a story: exploration, gradual
reveal, repetition, speed-up/slow-down. While none of these
mentioned patterns are specific about the storytelling genre,
the actual implementation of pattern depends on the genre
and its media. Some of these patterns can be translated into
narrative comic design patterns, as shown later.
DATA COMICS USE CASES
To obtain more examples of data comics and to explore design
patterns, we created 11 data comics aiming for a diverse set
of datasets and audiences. Creating one comic, including
gathering data, storyboarding, iterating, and finalizing took
between one and three days. The complete list of examples
can be found online.
1
In this section, we briefly describe three
examples (Figure 2) to introduce a few design patterns we
found in these comics (highlighted in black in the following).
Wannacry
The comic in Figure 2-left illustrates the outbreak of the Wan-
nacry virus in 2017. It is held in dark colors, contains lots
of illustrations, and few actual data visualization. The first
four panels create an
exposé
, introducing the virus and how
it works. The panels in row three show a
time-sequence
, one
panel per day showing the number of infected computers. The
last panel shows the
larger picture
of all the organizations
infected on the 5th day. Several individual panels
annotate
some selected companies by showing their logo. This comic
was intended for general information, e.g. on social networks.
O’Keefe’s Experiment
Figure 2-center shows a comic explaining one of the neuro-
science experiments which lead to the discovery of place cells
and the associated maps, for which John O’Keefe was awarded
the 2014 Nobel prize in Medicine and Physiology [45]. While
several explanatory videos were created after the Nobel prize
announcement
2
, the linear structure of time-based media limits
the communication of the complex process. We feel that data
comics afford more freedom here: a
larger panel
was used to
place the aerial view of the rat maze (adapted from Fig.2 of
the original paper) in relationship to the real-life experimen-
tal setup. Repeating this aerial view creates a
time-sequence
showing the animal moving through the maze, in parallel to
the brain recordings coming from the experimental rig. This
juxtaposition simulates the point of view of the experimenter,
making explicit the relationship between the animal position
within the maze and the brain activity. Finally, the sequence
builds up
to a more conventional representation of a ‘place
cell’ map (adapted from Fig. 2 of the original paper).
Through combination and nesting of basic data-comic design-
patterns we were able to visualize all the steps required to go
from an experimental setup to high level experimental results.
However, contrary to the frames of a video, the comic panels
allowed us to preserve the logic connections and visualize
them simultaneously on the page, so that at any point and in a
single gaze the reader can be reminded of the larger picture,
and move within the explanation at will.
USAid Data Reports
A third comic illustrates the development strategy of USAid in
Uganda. Local teams gather data which is compiled into charts
and maps and then explained in data reports [5] that support
decisions about future strategies. However, these reports are
too dense and text heavy to communicate the goals of USAid
to the general public.
Our data comic is designed in collaboration with USAid for
use on their website. Together, we defined audience, presen-
tation medium and settled on explaining the situation of a
1
http://comics.datacomics.net
2
E.g. https://www.youtube.com/watch?v=GmFspbHHZ_w
Figure 2. Three examples of our data comics. Wannacry outbreak (left) (©Zezhong Wang); O’Keefe’s experiment (center) (©Matteo Farinella); US
development aid in Uganda (right) (©Zezhong Wang).
14-year-old Ugandan girl, the focus of the current USAid de-
velopment strategies. The comic (Figure 2-right) starts with
a
larger panel
, locating Uganda. It then links a set of three
exposé
panels, introducing the central figure of the comic.
The exposé ends with a
panel-barchart
showing the share of
natural disasters the girl is confronted with. Then, the comic
contrasts two charts and shows an
annotated transition
(ar-
rows that grow the small yellow rectangle into a larger colored
treemap). Eventually, the linear sequence gets broken, leading
to three
facets
(the last three panels). The full comic with the
remainder can be seen online.
To conclude, these comics as well as the other 8 ones led to
a number of data-comic design-patterns (highlighted in bold
face) that we identified. The next section details a design space
to systematically describe these patterns and to generate more
patterns.
PATTERN DESIGN SPACE
The goal of our research is to facilitate the creation of comics
from data. We set out to identify patterns that capture the way
data comics can present information and create narration. We
coded patterns from 9 existing data comics
3
plus our own 11
comics mentioned in Section Data Comic Use Cases. From
these examples, we could describe an initial set of 29 design
patterns which informed our design space. The design space
then provided 1) a structured terminology, and 2) a frame of
thought to systematically brainstorming new patterns. This
allowed us to describe 13 more patterns, not captured by any
comic so far. Dimensions of the design spase were not defined
a priori; we followed an iterative process with 3 individual
authors creating and refining their own patterns and categories.
The final categories were decided by consensus (>12 hour long
discussions) and final review.
3
http://comics.datacomics.net
As data comics combine spatial presentation with narration,
we define a data-comic design-pattern as
a set of panels
with specific layout and content relation. A design pattern
hence presents a specific solution to a narrative purpose, by
proposing a graphical layout and the content of the individual
panels. Consequently, our design space combines a dimension
for spatial panel-layouts with a dimension for content relation
between panels. A design pattern can be described by a com-
bination of both dimensions. However, some combinations
can describe multiple patterns and their variations.
Some patterns are commonly used in infographics as both gen-
res use the visual space for graphical expression. For example,
many infographics use a layout grid and group elements hier-
archically; some show small multiples to convey changes over
time or show different facts of the data; some infographics are
designed to guide the reader’s view in a specific way by play-
ing with size, color, and layout. While the visual distinction
between infographics and comics is sometimes blurred, we
articulate three main characteristics of the data comics genre
as compared to other genres: the notion of panels, a narrative
relation between panels, and a guided layout.
Panels
The central narrative device in a comic is the panel, defined
as a frame “depicting a frozen moment” [4] and the camera
through which the reader perceives the “world” [42]. For
the sake of our research,
we define a panel as
a narrative
element that captures a moment in the narration and focuses
the reader’s attention.
While panels are inherent to comics it helps to extend their no-
tion to the other narrative genres such as infographics, posters,
slideshows, and videos (Figure 3). In infographics and posters
(the space-oriented genres), a panel corresponds be any graph-
ical component that can be understood as a narrative element
