An Analysis of Voter Demographics in the District of Wentworth in the 2022 Federal Election

Utilization of Essential Libraries for Spatial Data Analysis and Visualization

The study utilizes the following libraries/packages:

• sf: for spatial data manipulation.

• tidyverse: for data manipulation, visualization, and analysis.

• kableExtra: for generating elegant tables.

• patchwork: for arranging multiple ggplot2 plots into a single layout.

• lwgeom: for geometric operations on spatial data.

Data Compilation: Sources Utilized in the Study

The following data sets have been used in the study:

• Distribution of Preferences by Candidate by Division 2022 from Australian Election Commission.

• Distribution of Preferences by Candidate by Division 2019 from Australian Election Commission.

• Election Boundaries Data for NSW from Australian Election Commission

• G07 State/Territory Data for New South Wales (GDA 1994) from Australian Bureau of Statistics.

• G17 State/Territory Data for New South Wales (GDA 1994) from Australian Bureau of Statistics.

• 2022 Referendum Results from Australian Election Commission

Analysis of Seats Secured by Independent Candidates in the 2022 Australian Federal Election

To identify the electoral seats won by independent candidates, we initially isolate records where candidates were elected as independents (IND) during the 2022 Australian federal election. Subsequently, we extract relevant columns such as the electoral division (DivisionNm), candidate’s full name (GivenNm and Surname), and the corresponding state abbreviation (StateAb). Additionally, a new column titled “Name” is generated by combining the given name and surname, both in title case. We eliminate duplicate entries based on the electoral division, state, and candidate name. Finally, the refined and formatted data is presented in a tabular structure (refer to Table 1), showcasing the electoral division, state/territory, and the name of the elected representative.

Table 1: Electoral Districts won by Independent Candidates in Federal Election 2022

Electoral District	State/Territory	Elected Representative
Fowler	NSW	Dai Le
Mackellar	NSW	Sophie Scamps
North Sydney	NSW	Kylea Jane Tink
Warringah	NSW	Zali Steggall
Wentworth	NSW	Allegra Spender
Clark	TAS	Andrew Wilkie
Goldstein	VIC	Zoe Daniel
Indi	VIC	Helen Haines
Kooyong	VIC	Monique Ryan
Curtin	WA	Kate Chaney

Analysis of Previously Held Liberal Seats Lost in the 2022 Australian Federal Election

To identify previously held Liberal seats from the dataset, we filter records where candidates were elected as members of the Liberal Party in the 2019 Australian federal election. We focus on specific electoral districts, namely Fowler, Mackellar, North Sydney, Warringah, Wentworth, Goldstein, Indi, Kooyong, and Curtin. Relevant columns such as the electoral district (DivisionNm) and state/territory abbreviation (StateAb) are then selected. Duplicate entries are eliminated based on the combination of electoral district and state/territory. Finally, the refined and formatted data is presented in tabular format (refer to Table 2), showcasing the electoral districts and their respective state/territories.

Table 2: Electoral Divisons of Liberal Party won by Independent Candidates in Federal Election 2022

Districts	State/Territory
Mackellar	NSW
North Sydney	NSW
Wentworth	NSW
Goldstein	VIC
Kooyong	VIC
Curtin	WA

Defining Voter Characteristics in Wentworth

Income Distribution Across Gender and Age Groups

We initiate our investigation by extracting electoral boundary data for New South Wales (NSW). Subsequently, we merge and format state/territory data for NSW to facilitate the delineation analysis of income, age, and gender. Following data tidying procedures, we filter the electorate names for Wentworth and intersect them with state electoral districts covering more than 10 per cent of the area.

During the data filtering process for our electorate, there’s no requirement to update our coordinate system from GDA 1994 to GDA 2020. This is because the electoral boundaries for the state of New South Wales were last updated in 2016 (see Redistribution that created this boundary for the state of New South Wales). Given that GDA 2020 was adopted in 2020 (see GDA2020 NSW Legislation Amendments), we can utilize GDA 1994 for both the electoral map and the census data set without the need for conversion.

We narrow down the data set to focus specifically on the Wentworth district, paying attention to areas such as Coogee, Heffron, Sydney, Vaucluse. Records related to ‘P’ (representing ‘Persons’) are excluded, as gender information is not available for these records..

Figure 1: The Figure Shows the Intersection of State Electoral Divisions (SED) with district Wentworth in New South Wales.

The figure labeled Figure 1 illustrates the intersection of State Electoral Divisions (SED) with the Wentworth district. Following this, we can proceed to filter the data for the identified SEDs to acquire information regarding the income, age, and gender distribution within our selected electoral district of Wentworth.

Figure 2: Comparing Income Distribution by Sex & Age in Wentworth with New South Wales

We then, generate two distinct plots visualizing the income distribution across various age groups and genders in New South Wales (NSW) (see Figure 2).

The first plot, labeled as f1, utilizes data specific to the Wentworth electoral district. It employs a dodged column chart to represent income groups against the count of individuals, with separate bars for each gender. The plot is facetted based on age groups (see Figure 2, Part A), with a black-and-white theme, Viridis color palette, and rotated x-axis text for better readability.

In contrast, the second plot (see Figure 2, Part B), labeled as f2, draws from a broader data set encompassing NSW. This plot excludes records labeled as ‘Persons’ and follows a similar structure to f1, facilitating a comparative analysis of income distribution across age groups and genders for the entire NSW region.

We can note that the income distribution across age groups and gender in Wentworth closely mirrors that of the state.

Median Weekly Rent and Income

We begin by performing several spatial operations to integrate geometric coordinates with statistical area (SA1) data in New South Wales (NSW) and determine their overlap with electoral regions. First, we import and select geometric coordinates associated with SA1 regions from a spatial file. Then, we read SA1 data from a CSV file and join it with the geometric coordinates based on SA1 codes.

Next, we calculate the centroids of SA1 regions. Subsequently, we determine which SA1 centroids intersect with electoral boundaries by utilizing the function. The resulting intersections are used to identify which SA1 regions overlap with specific electoral divisions. This information is then merged with the SA1 data to create a new data set. Finally, we group the data set by electoral divisions and calculate summary statistics such as the average weekly rent and average personal income for each division.

Next, we filter the grouped data to isolate statistics for the Wentworth electoral division. Then, we calculate the deviations of average personal income and weekly rent for other electoral divisions compared to Wentworth. This is achieved by subtracting the corresponding values of Wentworth from each division’s average statistics.

In the first plot (n1), we visualize the deviation of average personal income around Wentworth (see Figure 3, Part A) for other electorates in New South Wales. Each bar represents the difference in average income between each electorate and Wentworth. The x-axis denotes the electorates, while the y-axis indicates the average income deviation.

Similarly, in the second plot (n2), we depict the deviation of average weekly rent around Wentworth for other electorates in New South Wales (see Figure 3, Part B). Each bar represents the difference in average rent between each electorate and Wentworth. The x-axis denotes the electorates, while the y-axis indicates the average rent deviation. Both plots utilize a black-and-white theme with a base size of 10, and the x-axis text is rotated for improved readability.

Figure 3: Comparing Income & Rent in Wentworth with other electorates from New South Wales

The analysis reveals that Wentworth exhibits notably higher average income compared to the remaining 46 electorates, resulting in negative deviations when contrasted with other regions. Additionally, barring Warringah and Bradfield, all other electorates demonstrate lower average weekly rents compared to Wentworth. These findings underscore the distinctive economic landscape of Wentworth, necessitating policymakers to tailor strategies that address its unique socioeconomic dynamics.

This insight provides a holistic view of the constituency’s economic well-being, where higher incomes suggest prosperity and enhanced economic opportunities, while lower incomes may signal areas of economic disparity requiring targeted interventions. Moreover, housing affordability significantly impacts constituents’ quality of life, influencing their capacity to save, pursue education, and maintain decent standards of living. Policymakers and politicians who grasp these challenges can advocate for initiatives supporting affordable housing and alleviate housing-related financial burdens on constituents.

Figure 4: Median Weekly Rent & Personal Income in Wentworth

Now, we narrow down our focus on Wentworth to compare Median Weekly Rent & Income across SA1 regions in the district (Figure 4).

In the first plot (p1), we filter the data set to focus on the Wentworth electoral division, ensuring that the median weekly rent data is nonzero. We then create a ggplot object and to plot the spatial geometry of SA1 regions, coloring them based on the numeric values of median weekly rent. The plot is styled with a black-and-white theme, with a legend positioned at the bottom and a Viridis color scale representing median weekly rent values.

Similarly, in the second plot (p2), we filter the data set to concentrate on the Wentworth electoral division, ensuring that the median total personal income data is nonzero. We create another ggplot object and to visualize the spatial distribution of SA1 regions, coloring them based on the numeric values of median total personal income. This plot also follows a black-and-white theme, with a legend at the bottom and a “Viridis” color scale representing median total personal income values.

Yes Campaign, No Campaign Voter Referendum Analysis

To determine the population proportion of Indigenous and Torres Strait Islanders, we start by reading the raw Indigenous population data from the provided file and filter it to keep only the relevant State Electoral Division (SED) codes intersecting with the Wentworth district. The G07 Geo-package contains information on Indigenous status grouped by Age and Sex and hence, can be used for this analysis.

Then, we pivot the data to long format to facilitate analysis, separating various categories into distinct columns. Following this, we perform data cleaning procedures, which involve renaming and restructuring category labels to ensure consistency and clarity. Additionally, we remove unnecessary substrings from category names.

The resulting dataset specifically related to the Wentworth district’s Indigenous population is extracted, with details about age ranges, Indigenous status, and gender segregated into separate columns for ease of analysis.

Finally, we calculate the percentage of the Indigenous population within the Wentworth district and filter the results to focus solely on individuals identified as Indigenous.

The Indigenous & Torres Islander population in the district of Wentworth is 0.86 per cent with 6457 individuals. Hence, it is observed that Indigenous individuals constitute less than one percent of the total population in Wentworth. This scarcity of Indigenous and Torres Strait Islander people renders it impractical to compute Statistical Area 1 (SA1) statistics for this demographic group, consequently rendering such data unavailable. Instead, the smallest Australian Bureau of Statistics structure representing Aboriginal and Torres Strait Islander Communities is Indigenous Locations (ILOC). Each ILOC contains a minimum of about 90 resident populations and is aggregated from SA1s.

Table 3: Referendum Voting Statistics in the District of Wentworth

Yes Votes	No Votes	Total Votes	Yes Votes (in %)	No Votes (in %)
44976	26740	71716	62.71404	37.28596

Analysis (see Table 3) reveals that the percentage of “yes” votes for the referendum in the Wentworth district stands at 62.70%. Consequently, out of the total 71,716 individuals who participated in the referendum within the district, 44,976 individuals voted in favor of the referendum.

Table 4: Referendum Voting Statistics in the State of New South Wales

Yes Votes	No Votes	Total Votes	Yes Votes (in %)	No Votes (in %)
1694477	2447408	4141885	40.91077	59.08923

Furthermore, it is noted that in the state of New South Wales, only 40.90% of the total population (Table 4), which amounts to 4,141,885 individuals, voted in favor of the referendum. This indicates that the referendum failed to secure a majority vote in the state. Additionally, it is evident that the district of Wentworth does not accurately represent the broader state population in terms of referendum votes.

References

• R Core Team (2024). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.
• Pebesma, E., (2018). Simple Features for R: Standardized Support for Spatial Vector Data. The R Journal 10 (1), 439-446, https://doi.org/10.32614/RJ-2018-009
• Pebesma, E., & Bivand, R. (2023). Spatial Data Science: With Applications in R. Chapman and Hall/CRC. https://doi.org/10.1201/9780429459016
• Pebesma E (2024). lwgeom: Bindings to Selected ‘liblwgeom’ Functions for Simple Features. R package version 0.2-14, https://CRAN.R-project.org/package=lwgeom.
• Pedersen T (2024). patchwork: The Composer of Plots. R package version 1.2.0, https://CRAN.R-project.org/package=patchwork.
• Wickham H, Averick M, Bryan J, Chang W, McGowan LD, François R, Grolemund G, Hayes A, Henry L, Hester J, Kuhn M, Pedersen TL, Miller E, Bache SM, Müller K, Ooms J, Robinson D, Seidel DP, Spinu V, Takahashi K, Vaughan D, Wilke C, Woo K, Yutani H (2019). “Welcome to the tidyverse.” Journal of Open Source Software, 4(43), 1686. doi:10.21105/joss.01686 https://doi.org/10.21105/joss.01686.
• Zhu H (2024). kableExtra: Construct Complex Table with ’kable’and Pipe Syntax. R package version 1.4.0, https://CRAN.R-project.org/package=kableExtra.