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

Original


Source: Guardian graphic | Source: SQM Weekly rents index (Sep-2022).


Objective

Guardian News online (Australian Edition) presented the following visualisation as part an article discussing housing affordability and the recent record-high rents across Australia and its direct impact to tenants on 27th September 2022. In the news article written by Caitlin Cassidy, this particular visualization title ‘Price change on combined units and houses’ is was attempting to reinforce the significance of the year on year increase in average weekly rents for units and houses across Australia’s capital cities. I believe this article was targeted at the general population who may be interested in housing affordability, increasing costs of living or broader thematic social and economic and policy issues.

For the purpose of this assignment, my focus will only be restricted to the critique of this visualization from an academic and a ‘visual best-practice’ point of view.

This particular visualization had the following three primary issues:

  • Arrangement: The arrangement of this visualizations axis is inappropriate as it plots both a dollar value (weekly rents) and a percentage change (year-on-year) on the same x axis. As a result the proportions are misrepresented and fails to convey the significance of the relative increase in weekly rents across Australian state capitals. Furthermore the visualization’s attempt to sort the data based on ‘12 month increase (%)’ is obfuscated given the larger visual proportion of the dollar values. There are in all 16 data points here (8 capital city rents and their respective yearly percentage change), non of which are clearly labeled or ordered appropriately

  • Text: The visualizations fails to convey the context or the significance of the year increase of the rents across Australian capital cities. For example, it is unclear as to which point in time this data is referring to. We can only deduce from the article and date of publishing that this data refers to the period of September 2022 vs September 2021. The graph title reads ‘price change’, however most people struggle to grasp the relative nature of percent changes and would therefore find it easier to grasp dollar changes (or increases in this case).

  • Data is ineffectively labelled: The ‘weekly rents’ amounts presumably refers to average rents, which is less obvious. to the general reader. This implies that these values are across houses and units. Theses high-level values will have its limitations as there are various configurations of dwellings (bedrooms,baths,car spaces, suburbs, metro/regional etc) that will significantly impact rental values. The visualization also fails to specify that the data refers only to rents in state capitals (metro areas only). This matters as regional areas too are facing significant rent cost pressures across the country at present.

Reference

Code

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

library("dplyr")
library("tidyr")
library("readxl")
library("ggplot2")

# Setup working drive to load in datasets
setwd("~/Personal/RMIT/Courses/Data Visualisation and Communication/Assessment 2")

#Read data from worksheet called Rents (58 variables originally, 74,678 obs)
rents <- read_excel("Housing_rental_data.xlsx")


#Reorder and select columns
rents <- rents [, c(1,3,2)]

#View dataframe
head(rents, n= 8)
## # A tibble: 8 x 3
##   State     PriorYear CurrentYear
##   <chr>         <dbl>       <dbl>
## 1 Sydney         508.        666.
## 2 Canberra       592.        657.
## 3 Darwin         513.        565.
## 4 Brisbane       434.        558.
## 5 Perth          454.        538 
## 6 Hobart         453.        501.
## 7 Melbourne      404.        501.
## 8 Adelaide       397.        487.
#Pivot longer
rents <- data.frame(rents %>% pivot_longer(!State, names_to = "Period", values_to = "WeeklyRents"))

#View dataframe again
head(rents,n= 8)
##      State      Period WeeklyRents
## 1   Sydney   PriorYear    508.4708
## 2   Sydney CurrentYear    666.4100
## 3 Canberra   PriorYear    591.8309
## 4 Canberra CurrentYear    656.8600
## 5   Darwin   PriorYear    513.2198
## 6   Darwin CurrentYear    565.2200
## 7 Brisbane   PriorYear    433.9451
## 8 Brisbane CurrentYear    557.7700
#Factorize Period
rents$Period <- factor(rents$Period, levels = c('PriorYear', 'CurrentYear'))
#Change decimal places on rents
rents$WeeklyRents <- round(rents$WeeklyRents,digits =0)

#Plotting the graph
ggp <- ggplot(rents,
              aes(x= reorder(State, - WeeklyRents), y = WeeklyRents, fill = Period)) +
  geom_bar(stat ="identity", position ="dodge") +
  scale_y_continuous (expand =c(0,0), limits = c(0,700), breaks =seq(0,700, by=100)) +
  
#Setup labels
  geom_text(aes(State,label = WeeklyRents),
            position = position_dodge(width = 1), vjust =2) +
  
#Setup titles and axis labels
  labs(title = "Increase in combined weekly rents across Australian capital cities",
       subtitle = "Combined houses and units : Sep-21 vs Sep-22",
       y = "Avg. weekly rents ($)", 
       x = "State capitals",
       caption ="Source: SQM research weekly rents index\ combined houses and units")

#view final graph
ggp

Data Reference

Reconstruction

The following plot fixes the main issues in the original.