Data Visualisations and Communications
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
A2: Critical analysis of the
efficacy of 'Falling Angels?'
COMM2501: Data Visualisations and Communications
Total words (excl. tables and figure captions): 1095
Data analysed:
‘Falling Angels?’ by Slobin, S, Cage, F & Palma, E published in Reuters on April 3,
2020.
Introduction
‘Falling Angels?’ is a timeseries analysis of corporate bond ratings and debt levels between
1983 - 2020 published by Reuters. Multiple data sets are used throughout the visualisation
with sources including S&P Global, Moody’s Investor Service, OECD reports, the US
Federal Reserve, and the US Treasury Office; unfortunately, none are referenced specifically.
I sought a story focused on a non-health aspect of COVID-19. Given my finance major, I
gravitated towards the financial aspect and settled on ‘Falling Angels?’
Analysis
Analytical methods
The authors have used a wide array of non-interactive analytical methods to convey their
story. As seen in figure 1, methods from simple bar charts, stacked area charts, and line
graphs to more complex radar charts and pictograms are utilised. These methods are used to
present statistics regarding the volume of bonds outstanding, credit relative to GDP,
distribution of bond ratings, financial stress index, bond re-ratings, and default rates.
Data story
The authors have followed the data story framework outlined by Dyke (Heeg, 2015) and have
presented one central idea. They leave the audience with the conclusion that due to the
financial stress caused by COVID-19, corporations are unable to repay bonds, thus causing
expected default rates to increase. This conclusion has been conveyed through a linear
structure as suggested by Dyke (Heeg, 2015), which is distinguished by subheadings. Doing
so allows the story to flow naturally and conveys the conclusion clearly.
Figure 1: A selection of the data visualisations used in ‘Fallen Angels?’
4
As aforementioned, ‘Falling Angels?’ employs a variety of visualisation modes which act
together to increase the story aesthetics. By increasing the aesthetics, the perception that the
story is easy to use increases too (Lidwell & Butler, 2010), thus better engaging the audience.
The story investigates a complex topic yet makes no attempt to explain the finance jargon or
principles used to the audience. Whilst some may identify this as a critique, it is likely a
deliberate choice based on the principal element of user experience (UX) design whereby
audience personas are identified and catered to (Unger & Chandler, 2012). Personas of this
story are likely those with existing financial literacy such as finance students or industry
professionals.
The short form and structure of being one continuous page is further indicative of the
identified persona’s context-of-use. In taking only 2-3 minutes to engage with the story, busy
finance students and industry professionals can derive the conclusion quickly within their
existing routines, such as on the commute to work. Such a consideration for the context-of-
use is a hallmark of good visual design as identified in Zarour and Alharbi’s (2017)
framework.
The visualisations presented are all non-interactive, which may inhibit UX as the audience
cannot explore the data themselves. Whilst potentially a deliberate choice given the busy
persona targeted, the exclusion of interactivity inhibits users who wish to explore the data in
more depth.
Shortcomings
Colour
The colour palette used throughout the story, shown in figure 2, presents an area for
improvement. When used together, the purples and blues produce difficult to distinguish
visual elements as shown in figure 3. The authors have also not used this palette consistently,
with some graphs using different colours entirely, thereby violating the consistency heuristic
(Williams et al., 2018).
Figure 2: Colour
palette used in
'Falling Angels?'
Figure 3: Line graph used in ‘Falling Angels?’ showing difficult to distinguish colour palette.
5
Story layout
The story layout results in shortcomings related to information hierarchy and unity principles
of visual design (Yu Siang, 2019). As seen in figure 4, the typography used to title
visualisations is larger, bold, and a different font to what is used for body text. This incorrect
use of hierarchy creates confusion about what is important, and directs the audience’s eyes
over the page in a non-uniform way, thereby violating the spatial organisation heurisitc
(Williams et al., 2018). Regarding unity, the width of the narative text and visualisations are
different, and there is inconsistency in the width of data visualisations, thereby also violating
the consistency heuristic (Williams et al., 2018).
Pictogram
Figure 5: Pictogram used in 'Falling Angels?'
This pictogram in figure 5 conveys the increase in the number of bond re-ratings over time.
However, it has several legibility shortcomings. Firstly, the dots are different sizes with no
scale provided, thereby violating the visual principle of scale (Yu Siang, 2019). The subtext
“rating actions may be for more than one company” may explain why there are different
sizes, but it is not made clear whether that is the case, thereby leaving the audience confused.
Figure 4: Screen capture of 'Falling Angels?' showcasing shortcomings
of hierarchy and unity.
6
Additionally, the partial occlusion of some dots by one another is a shortcoming and violates
the gestalt principles. Namely, the principle that objects which share a common region and
are graphically connected are assumed by audiences to be related (Evans, 2017) is violated
creating a type 1 error. Further, this design choice violates the spatial organisation heuristic as
the occlusion makes it difficult to identify individual dots and count the true number of
ratings on a given day (Williams et al., 2018).
Radar chart
The radar chart shown in figure 6 has several
legibility shortcomings. The most significant
regards the aforementioned gestalt principle.
By using an overlayed radar chart, an
audience may assume that there is an inverse
causal relationship between the two variables.
This results in a type 1 error, as no such
relationship exists.
A second legibility shortcoming regards the
nodes representing the scale of the axis.
Nodes are only used at year 2000, requiring
the audience to extrapolate them around the
circle to read the chart. This is a difficult task
and likely causes misestimation of bond
ratings in years other than 2000.