Click the Original, Code and Reconstruction tabs to read about the issues and how they were fixed.

Original


Source: Department of Foreign Affairs and Trade - Publication: Trade and Investment at a Glance 2021.


Objective

Objective: The objective of the publication is to provide a "statistical snapshot of Australia's trade and investment performance over 2019-20". The objective of the data visualisation is to compare Australia's economic activities with other economies through the early stages of the COVID-19 pandemic. The forward by The Hon Dan Tehan, MP - Former Minister for Trade, Tourism and Investment - makes the claim that "Australia's economic and health response to COVID-19 has been among the best in the world".

Audience: The visualisation is available on the internet. It is assumed that the audience is the general public.

Issues

The overriding problem is that through its design, there is the risk of the visualisation failing to meet it's objective (to compare Australia's economic activities with other economies through the early stages of the COVID-19 pandemic).

The visualisation has the following main issues that each contribute to the problem:

Issue 1 - cluttered appearance

The amount of data presented makes for a cluttered visualisation (an example of visual bombardment). It is difficult to distinguish the line that represents the Australian economic performance from the other economies.

Distinguishing the different lines isn't helped by the fact that each line follows a similar broad trajectory.

Issue 2 - small line labels

As a positive observation, the lines are labelled and the colour of the text matches the colour of the lines. However the size of the labels is too small - they are difficult to read.

Issue 3 - distance of the labels from the lines

While the use of data labels is encouraged (as opposed to legends) the labels in this plot are "attached" to their line by another line! This adds to the clutter and the audience's difficulty in interpreting the visualisation.

Issue 4 - lack of distinction between the lines

While each economy is represented by a different colour (hues are appropriate for nominal data), there is an overall lack of distinction between the lines. The colour saturation and value are the same.

Other visual variables such as shape, texture and line type could have been used.

The other potential problem is the number of different colours (seven) used for the lines. This colour scale is qualitative (for discrete variables) and any more than four different colours can start to be problematic.

Issue 5 - small and faint x-axis labels

The x-axis is important because it anchors the audience in points in time. The visualisation's objective is to compare economies over time. This is made difficult by the use of overly small and faint labels on the x-axis.

These issues have the potential to inadvertently "deceive the receiver" (Kirk's fourth design principle) or at least make interpretation difficult.

Reference

Code

The following code was used to fix the issues identified in the original.

library(ggplot2)
library(lubridate)
library(dplyr)


Tracker_Australia_Min_Max <- read.csv("/home/ddd_v1_w_383L_668172/asn611151_1/asn611152_1/work/Tracker_Australia_Min_Max.csv")


Tracker_Australia_Min_Max <- mutate(Tracker_Australia_Min_Max, Date = as.Date(Date, format= "%Y-%m-%d"))

ggp <- ggplot(Tracker_Australia_Min_Max, aes(x = Date)) +    # Create default ggplot2 line plot
  geom_line(aes(y = MinValue), colour="grey", size=0.2) +
  geom_line(aes(y = MaxValue), colour="grey", size=0.2) + 
  geom_line(aes(y = Australia), colour="black", size=0.6) + 
  labs(x = "Date", y = "Economic Activity", title = "Australian economic activity 2019-2023 compared to the strongest\n and weakest from the OECD and EuroZone")


ggp <- ggp +                                # Add color between lines
  geom_ribbon(aes(x = Date,
                  ymin = MinValue,
                  ymax = MaxValue),
              fill = "gray", alpha = 0.6)

Data notes

As seen in the image, the data source for the initial plot was from Bloomberg Economics.

I was unable to obtain access to this data set, but found a similar, publicly available data set published by the OECD - The Weekly Tracker of GDP Growth. This data also represents national economic activity.

To most closely approximate the data used in the initial plot, in addition to the Australian data, I chose three "emerging economies" - Columbia, Mexico and India, two "advanced economies" - Japan and USA and two groupings - EuroZone and OECD. The maximum and minimum economic activity was calculated from this data and used for the reconstructed plot.

The key difference between the two plots is the inclusion of China in the initial plot. Not being an OECD country the data for China was unavailable from the the OECD website.

Data Reference

Reconstruction

The following plot fixes the main issues in the original.

Design Features

The design decisions behind the reconstructed plot were based on addressing the issues identified in the initial plot, while simultaneously being mindful of Kirk's four design principles. It was important to try not to introduce new issues while removing the existing issues.

Some fundamental design decisions were made.

Firstly the decision was made to retain the time series plot format. Time series plots are good examples of the visual law of connectedness, allowing the audience to efficiently identify the key pattern in the data - in this instance how economic activity changes over time.

The convention of representing time on the x-axis was retained, as was the convention of increasing time from left to right.

A greater time period was available for the reconstructed plot and so the scaling on the x-axis is smaller than the initial plot. This is not misleading because the trends in the two plots are not being compared with each other. The additional data is in fact useful for the analysis below.

Scaling on the y-axis is appropriate in that it allows for the spikes and dips to be visible and comparable, without unnecessary exaggeration.

Muted grid lines and plot background were chosen to minimise distraction from the data.

Additionally, five more specific design features were established.

Design Feature 1: a single line represents the Australian economic performance;

Design Feature 2: a "geom_ribbon" layer to represent the range of other economies in a band of a different colour (hue) and lighter (less saturation) to ensure the Australian line has prominence.

The combination of the first two design features provide a good balance of both form and function to the visualisation - Kirk's first design principle. It also makes the plot much more intuitive - Kirk's third design principle;

Design Feature 3: consideration of visual variables - the line representing the Australian economic performance has a greater weight, is of different colour (hue) and has greater saturation than the lines representing the maximum and minimum range of other economies, and the shading in between.

The colours were chosen to achieve a good level of contrast. Colour scales with red and green were avoided to help ensure accessibility for those with colour blindness;

Design Feature 4: larger labels on the x-axis; and

Design Feature 5: no labels (and especially no connecting lines) on the plot itself, because they are no longer needed due to Design Features 1 and 2.

These Design Features were specifically chosen (Kirk's second design principle - justifying selections) to address the identified issues in the initial plot as follows:

Issue 1 - cluttered appearance Directly addressed by Design Features 1 and 2.

Issue 2 - small line labels Not required due to Design Features 1 and 2. Removed as per Design Features 5.

Issue 3 - distance of the labels from the lines Not required due to Design Features 1 and 2.

Issue 4 - lack of additional visual distinctions Addressed by Design Feature 3.

Issue 5 - small and faint x-axis labels Addressed by Design Feature 4.

Final Analysis

Focusing on Kirk's fourth design principle - never deceive the receiver, a brief analysis was completed to assess the initial claim made in the the DFAT Publication which also relied on the initial plot as evidence.

Initial claim: "Australia's economic and health response to COVID-19 has been among the best in the world"

From the reconstructed plot, several points become obvious:

  1. Australia's economic activity dropped sharply as COVID-19 hit (first half of 2020) as did the activity of the other economies.

  2. Australia's economic activity did not drop as far as that of the other economies, before returning to more normal levels (post June 2021).

  3. Like the other economies, from the depression in early 2020, the Australian economic activity rose quickly for a period, then more slowly for a period, then sharply in Q1 2021, before resuming more normal levels. Up until the sharp spike, the Australian economy was one of the strongest, but after the spike it fell back to the middle of the field or lower for about a year (mid 2021 to mid 2022).

  4. Some economies experienced a series of subsequent smaller spikes in activity from Q2 2021 to Q2 2022 but Australia's economy wasn't included in any of these. It wasn't until a spike later in 2022 that Australia's economic activity also spiked.

The data presented in the DFAT publication ranged from January 2020 to February 2021. Across this range, from the reconstructed plot, it is reasonable to say that Australia's economic response was in fact one of the best in the world (the blue line is consistently close to the top of the range). The publication is somewhat misleading however, in that it's title "Trade and Investment at a Glance 2021" implies that it is providing commentary for the whole, or at least a substantial part of 2021 - not the first two months. As discussed, post the start of 2021, Australian economic activity was not world leading. It was average at best for significant periods of time.

References

Kirk, A. 2012. Data visualization: a successful design process. Birmingham, UK: Packt Publishing Ltd. Kommenda, N., C. Barr, and J. Holder. 2018. “Gender pay gap: what we learned and how to fix it.” https://www.theguardian.com/news/ng-interactive/2018/apr/05/women-are-paid-less-than-men-heres-how-to-fix-it.