Recall 2019 Federal Election Data
library(tidyverse)library(sf)aec_map <- read_sf(here::here("data/vic-july-2018-esri/E_AUGFN3_region.shp"))votes <- read_csv("https://results.aec.gov.au/24310/Website/Downloads/HouseDopByDivisionDownload-24310.csv", skip = 1)winners_fix <- votes %>% mutate(DivisionNm = ifelse(DivisionNm=="McEwen", "Mcewen", DivisionNm)) %>% filter(Elected=="Y" & CountNumber==0 & CalculationType=="Preference Count") %>% # get the winner right_join(aec_map, by = c("DivisionNm" = "Elect_div")) # join the dataauscolours <- c("ALP" = "#DE3533", "LNP" = "#ADD8E6", "KAP" = "#8B0000", "GVIC" = "#10C25B", "XEN" = "#ff6300", "LP" = "#0047AB", "NP" = "#0a9cca", "IND" = "#000000")ggplot(winners_fix) + geom_sf(aes(fill = PartyAb, geometry = geometry)) + scale_fill_manual(values = auscolours)
There are two sources of data:
- Electoral boundary
- The votes for candidates in each electorate
Recall 2016 ABS Census Data
- DataPacks https://datapacks.censusdata.abs.gov.au/datapacks/
- GeoPackages https://datapacks.censusdata.abs.gov.au/geopackages/
The Australian Statistical Geography Standard (ASGS)
State or Territory (STE)
vicmap_ste <- read_sf(geopath, layer = "census2016_eiuwa_vic_ste_short") ggplot(vicmap_ste) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_ste)## [1] 1Greater Capital City Statistical Areas (GCCSA)
- Each region with variable population
vicmap_gccsa <- read_sf(geopath, layer = "census2016_eiuwa_vic_gccsa_short")ggplot(vicmap_gccsa) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_gccsa)## [1] 4Section of State (SOS)
- Major urban, other urban, bounded locally & rural balance
vicmap_sos <- read_sf(geopath, layer = "census2016_eiuwa_vic_sos_short")ggplot(vicmap_sos) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sos)## [1] 6Remoteness Areas (RA)
vicmap_ra <- read_sf(geopath, layer = "census2016_eiuwa_vic_ra_short")ggplot(vicmap_ra) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_ra)## [1] 6Section of State Ranges (SOSR)
vicmap_sosr <- read_sf(geopath, layer = "census2016_eiuwa_vic_sosr_short")ggplot(vicmap_sosr) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sosr)## [1] 12Statistical Area Level 4 (SA4)
- Each region with population of 100,000 - 500,000
vicmap_sa4 <- read_sf(geopath, layer = "census2016_eiuwa_vic_sa4_short")ggplot(vicmap_sa4) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sa4)## [1] 19Significant Urban Areas (SUA)
vicmap_sua <- read_sf(geopath, layer = "census2016_eiuwa_vic_sua_short")ggplot(vicmap_sua) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sua)## [1] 22Commonwealth Electoral Division (CED)
vicmap_ced <- read_sf(geopath, layer = "census2016_eiuwa_vic_ced_short")ggplot(vicmap_ced) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_ced)## [1] 39Statistical Area Level 3 (SA3)
- Each region with population of 30,000 - 130,000
vicmap_sa3 <- read_sf(geopath, layer = "census2016_eiuwa_vic_sa3_short") ggplot(vicmap_sa3) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sa3)## [1] 68Local Government Area (LGA)
vicmap_lga <- read_sf(geopath, layer = "census2016_eiuwa_vic_lga_short")ggplot(vicmap_lga) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_lga)## [1] 82State Electoral Division (SED)
vicmap_sed <- read_sf(geopath, layer = "census2016_eiuwa_vic_sed_short") ggplot(vicmap_sed) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sed)## [1] 90Urban Centres and Localities (UCL)
vicmap_ucl <- read_sf(geopath, layer = "census2016_eiuwa_vic_ucl_short")ggplot(vicmap_ucl) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_ucl)## [1] 353Statistical Area Level 2 (SA2)
- Each region with populations in the range of 3,000-25,000
vicmap_sa2 <- read_sf(geopath, layer = "census2016_eiuwa_vic_sa2_short")ggplot(vicmap_sa2) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sa2)## [1] 464Postal Areas (POA)
vicmap_poa <- read_sf(geopath, layer = "census2016_eiuwa_vic_poa_short")ggplot(vicmap_poa) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_poa)## [1] 698State Suburbs (SSC)
vicmap_ssc <- read_sf(geopath, layer = "census2016_eiuwa_vic_ssc_short") ggplot(vicmap_ssc) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_ssc)## [1] 2931Statistical Area Level 1 (SA1)
- Each region with a population of range 200-800
vicmap_sa1 <- read_sf(geopath, layer = "census2016_eiuwa_vic_sa1_short")ggplot(vicmap_sa1) + geom_sf(aes(geometry = geom, fill = Median_age_persons))
nrow(vicmap_sa1)## [1] 14073Comparing SED 2016 and electorates divisions 2019
See here for winners_fix data was.
ggplot(winners_fix) + geom_sf(data = vicmap_sed, aes(geometry = geom, fill = Median_age_persons), alpha = 1, color = "white", size = 2) + geom_sf(aes(geometry = geometry), fill = "transparent", color = "red") + coord_sf(xlim = c(144.95, 145.24), ylim = c(-38.05, -37.85))
Closer look 🔬 Hotham electorate 1
electorate <- winners_fix %>% filter(DivisionNm=="Hotham")sed_intersect <- vicmap_sed %>% filter(st_intersects(geom, electorate$geometry, sparse = FALSE)[,1])ggplot(electorate, aes(geometry = geometry)) + geom_sf() + geom_sf_text(aes(label = DivisionNm)) + geom_sf(data = sed_intersect, aes(geometry = geom), color = "red", fill = "transparent") + geom_sf_text(data = sed_intersect, aes(label = sed_code_2016, geometry = geom), color = "red")- There are 10 SED regions that intersect with Hotham electorate.
Closer look 🔬 Hotham electorate 2
sed_intersect2 <- sed_intersect %>% mutate(geometry = st_intersection(geom, electorate$geometry), perc_area = 100 * st_area(geometry) / st_area(geom), perc_area = as.numeric(perc_area)) %>% filter(perc_area > 5)ggplot(sed_intersect2, aes(geometry = geometry)) + geom_sf(data = electorate) + geom_sf_text(data = electorate, aes(label = DivisionNm)) + geom_sf(color = "red", aes(fill = Median_age_persons)) + geom_sf_text(aes( label = glue::glue("{sed_code_2016} ({scales::comma(perc_area, 1)}%, {Median_age_persons})")), color = "red") + theme(legend.position = "bottom")- There are 5 SED areas with at least 5% intersection with the electoral area.
- How would you characterise the median age for Hotham?
Closer look 🔬 Hotham electorate 3
Strategy 1
sort(sed_intersect2$Median_age_persons)## [1] 32 35 39 40 40Closer look 🔬 Hotham electorate 3
Strategy 1
sort(sed_intersect2$Median_age_persons)## [1] 32 35 39 40 40Strategy 2
mean(sed_intersect2$Median_age_persons)## [1] 37.2Closer look 🔬 Hotham electorate 3
Strategy 1
sort(sed_intersect2$Median_age_persons)## [1] 32 35 39 40 40Strategy 2
mean(sed_intersect2$Median_age_persons)## [1] 37.2Strategy 3
weighted.mean(sed_intersect2$Median_age_persons, sed_intersect2$perc_area)## [1] 36.35205Closer look 🔬 Hotham electorate 4
sa1_intersect <- vicmap_sa1 %>% filter(st_intersects(geom, electorate$geometry, sparse = FALSE)[,1]) sa1_intersect2 <- sa1_intersect %>% mutate(geometry = st_intersection(geom, electorate$geometry), perc_area = 100 * st_area(geometry) / st_area(geom), perc_area = as.numeric(perc_area)) %>% filter(perc_area > 5) ggplot(sa1_intersect) + geom_sf(color = "red", aes(fill = Median_age_persons, geometry = geom)) + geom_sf(data = electorate, color = "white", size = 2, fill = "transparent", aes(geometry = geometry)) + theme(legend.position = "bottom")
Closer look 🔬 Hotham electorate 5
Strategy 1
fivenum(sa1_intersect2$Median_age_persons)## [1] 0 34 38 42 82Closer look 🔬 Hotham electorate 5
Strategy 1
fivenum(sa1_intersect2$Median_age_persons)## [1] 0 34 38 42 82Strategy 2
mean(sa1_intersect2$Median_age_persons)## [1] 37.38235Closer look 🔬 Hotham electorate 5
Strategy 1
fivenum(sa1_intersect2$Median_age_persons)## [1] 0 34 38 42 82Strategy 2
mean(sa1_intersect2$Median_age_persons)## [1] 37.38235Strategy 3
weighted.mean(sa1_intersect2$Median_age_persons, sa1_intersect2$perc_area)## [1] 37.35034Closer look 🔬 Hotham electorate 5
Strategy 1
fivenum(sa1_intersect2$Median_age_persons)## [1] 0 34 38 42 82Strategy 2
mean(sa1_intersect2$Median_age_persons)## [1] 37.38235Strategy 3
weighted.mean(sa1_intersect2$Median_age_persons, sa1_intersect2$perc_area)## [1] 37.35034Strategy 4
ggplot(sa1_intersect2, aes(x = Median_age_persons)) + geom_histogram(binwidth = 1)
Closer look 🕵️ Zero median age
sa1_intersect2 %>% filter(Median_age_persons==0) %>% ggplot() + geom_sf() + geom_sf(data = electorate, color = "red", fill = "transparent", aes(geometry = geometry))
Dorling Cartogram
sa1_intersect4 <- sa1_intersect %>% mutate(centroid = st_centroid(geom))ggplot(sa1_intersect4) + geom_sf(data = electorate, aes(geometry = geometry), size = 2, fill = "grey60") + geom_sf(aes(geometry = centroid, color = Median_age_persons), size = 0.5, shape = 3) + scale_color_viridis_c(name = "Median age", option = "magma")
Closer look 🔬 Hotham electorate 7
sa1_intersect5 <- sa1_intersect4 %>% filter(st_intersects(centroid, electorate$geometry, sparse = FALSE)[,1], Median_age_persons!=0)Strategy 1
fivenum(sa1_intersect5$Median_age_persons)## [1] 20 34 38 42 82Strategy 2
mean(sa1_intersect5$Median_age_persons)## [1] 38.58015Strategy 4
ggplot(sa1_intersect5, aes(x = Median_age_persons)) + geom_histogram(binwidth = 1)
Closer look 🔬 Hotham electorate 7
sa1_intersect5 <- sa1_intersect4 %>% filter(st_intersects(centroid, electorate$geometry, sparse = FALSE)[,1], Median_age_persons!=0)Strategy 1
fivenum(sa1_intersect5$Median_age_persons)## [1] 20 34 38 42 82Strategy 2
mean(sa1_intersect5$Median_age_persons)## [1] 38.58015Strategy 4
ggplot(sa1_intersect5, aes(x = Median_age_persons)) + geom_histogram(binwidth = 1)
- There are many ways to characterise an electorate.
- Estimates of median age of an electorate is more consistent using SA1 map data than SED map data.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Lecturer: Emi Tanaka
Department of Econometrics and Business Statistics
ETC5512.Clayton-x@monash.edu
Week 6