Split Voting

This article explores the split_total and split_electorate datasets.

Split Voting

The split_total and split_electorate datasets contains the number of voters who cast their party vote for a different party to their candidate vote at the national level, by party. This dataset only has election years 2005 through to 2023 due to there not being split-ticket voting data available for previous years.

Upload data

View the shape of data frame.

df <- scgUtils::get_data("split_total")
# or
# df <- scgUtils::get_data("split_electorate")

Prepare Sankey

Prepare sankey data.

# Filter 2023 election data only
df <- df[df$Election == 2023, ]

# If electoral-level dataset is used, also filter by Electorate
# df <- df[df$Election == 2023 & df$Electorate == "Auckland Central", ]

# Rename unsuccessful minor parties into "Other" category
df <- update_names(df, nameCol = "List_Party")
df <- update_names(df, nameCol = "Electorate_Party")

df <- df %>%
  # Get sum of votes by flow
  group_by(List_Party, Electorate_Party) %>%
  summarise(Vote = round(sum(Votes)), .groups = 'drop') %>%
  ungroup() %>%
  # Get percent of party vote going to candidate vote
  group_by(List_Party) %>%
  mutate(Percent = round(Vote/sum(Vote)*100,2) %>%
  ungroup()

head(df)

List_Party	Electorate_Party	Vote	Percent
ACT Party	ACT Party	68692	28.46
ACT Party	Green Party	2095	0.87
ACT Party	Informal	1356	0.56
ACT Party	Labour Party	7715	3.20
ACT Party	Maori Party	715	0.30
ACT Party	NZ First	3106	1.29

Plot

Using the scgUtils package, create sankey to visualise the split vote from the candidate vote (left) to the party vote (right). Use the colour_prep function from the same package to match the party colours to the correct political party.

scgUtils::plot_sankey(
  data = df,
  source = "List_Party", # left side of sankey
  target = "Electorate_Party", # right side of sankey
  value = "Vote",
  colours = scgUtils::colour_prep(df, c("List_Party","Electorate_Party"), pal_name = "polNZ"),
  width = 900, # reduce width from default of 1200
  height = 700, # reduce height from default of 800
  shiftLabel = 8 # manually shift labels to outside of plot if odd number of parties on each side
) # %>%
  # save from viewer to html
  # htmlwidgets::saveWidget(file="sankey_2023.html", selfcontained = TRUE)

NB. Percentage and votes within tooltip on nodes do not exactly equal the party_vote or candidate_vote of each party due to rounding.