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

Original


Source: FirstLinks - a Morningstar Company (2016).


Objective and Audience

The visualisation highlights the mix of industries and their overarching sectors, as recorded by the ABS in the March 2016 National Accounts release. This chart is representative of the total share of Australia’s employees, as broken down by industry. We could assume from this, the objective is to provide information around the future of Australia’s industry mix and how this could be used for future economic growth. Additionally, we should be aware that this chart should be used in conjunction with further economic data to provide a holistic picture of Australia’s economy.

The visualisation targets those who are interested in the Australian (and global) economy. Specifically, those wishing to seize industry-focused opportunities for business growth and development.

Issues

Overall, the visualisation is deceptive and includes perceptual issues. These issues are centered around the use of a pie chart to depict the data and other design-based features which interfere with the viewers ability to interpret the data.

Pie charts are considered ‘good’ when there are limited categories, appropriate use of colour, data is well-presented to tell the story, and labels are accurately used.

  • Issue 1 - Representation of industry sectors: Pie charts use angles to represent proportions and percentages. In the visualisation the audience will struggle to do this due to the large number of industry sectors (19) which make it difficult to visually process area size in comparison to a standard x and y-axis chart. Additionally, the chart has been made three-dimensional and presented on a tilted angle. When comparing ‘Admin & Support Services’ which makes up 3.7% with ‘Personal & Other Services’ which makes up 3.9%, to the eye ‘Admin & Support Services’ appears to be larger due to the angle of the segment. These factors make it difficult for the reader to understand the breakdown between industries. We could conclude that the deception is not intentional and instead a poor choice of visual design.

  • Issue 2 - Chart labels: The charts labels are not clear and do not aid the viewer when interpreting the percentage value of the different industries and sectors. Additionally, there is inconsistency in the placement of text and size, with some labels appearing on the segments and others not. This makes it difficult to differentiate between the industry and sector labels.

  • Issue 3 - Use of colour: The use of colour is inappropriate and can be difficult to view for long periods of time given the fluorescent tones. Additionally, this chart is not suitable for an audience that is colour blind. Most commonly, colour blindness is characterized by an inability to differentiate between different shades of colors, such as red, green, or blue. This chart has all three shades which may make it difficult for the viewer to interpret and differentiate between segments.

Reference

Code

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

# Load libraries
library(ggplot2) 
library(dplyr)  
library(knitr) 
library(readr)
library(here)
library(colorblindr)

# Load data
industry_data <- read_csv(here("industry_data.csv"))

# Create proportions and percentages
industry_data = mutate(industry_data, 
                Industry_proportion = Value / sum(Value))

industry_data = mutate(industry_data, 
                Industry_percent = Industry_proportion*100)

# Order industry data in descending order and define factor
industry_data$Industry <- industry_data$Industry %>%
  factor(levels = industry_data$Industry[order(-industry_data$Industry_percent)])

# Define Sector factors
industry_data$Sector <- factor(industry_data$Sector, levels=c('Primary','Secondary','Tertiary','Quinary','Quaternary'))

# Create plot
p <-ggplot(industry_data,aes(x = Industry_percent, y = Industry, fill = Sector))
p + geom_bar(stat="identity", width = 0.8) +
  geom_segment(aes(x = 0, y = Industry, xend = Industry_percent, yend = Industry), linetype = 0) +
  scale_fill_manual(values = c("#999999", "#E69F00", "#009E73", "#F0E442", "#CC79A7")) +
  theme_minimal() +
  labs(title = "Australia's Employment by Industry Sectors",
       x = "Shares of Total Employment as a Percent (%)",
       y = "Industry Sector") +
  geom_text(aes(label=round(Industry_percent,2)) ,hjust = -0.5, size = 2,  position = position_dodge(width = 0.5), inherit.aes = TRUE)+
  scale_x_continuous(limits = c(0,15))

Data Reference

Reference List * RMIT ‘MATH2404 Data Visualisation and Communication’ course notes and webinar content.

Reconstruction

The reconstructed plot will fix the main issues in the original by making the following amendments:

  • The first issue has been amended by representing the data in a bar chart as opposed to a pie chart. Bar charts are appropriate to visually present categorical data. By representing the data this way, it will allow the audience to easily interpret and compare industry sectors given each bar is stacked in a comparable format. Additionally, this chart will not be designed in a three-dimensional manner and presented on a tilted angle.

  • The second issue has been amended by the use of the clear labels on the y-axis to highlight the industry mix and shares of total employment as a percentage on the x-axis. The percentage allocation for each industry is included to the right of the corresponding bar in reduced font size. While colours have been used to represent the overarching sector in which the industry sits, with a legend to the right of the plot. The use of text in the updated visualisation is consistent and used sparingly to not detract from the chart.

  • The third issue has been amended by updating the colours used to represent each sector. The colours are shades which are easier to view given their non-fluorescent nature in addition to the fact that they are appropriate for an audience that may be colour blind.

The reconstructed plot: