The Crime That Feeds Itself

# 1. Load the data straight from the source
# Crime Statistics Agency (CSA) - Recorded offences by LGA, year ending June 2026.
data_url   <- "https://files.crimestatistics.vic.gov.au/2026-09/Data_Tables_LGA_Recorded_Offences_Year_Ending_June_2026.xlsx"
local_file <- "Data_Tables_LGA_Recorded_Offences_Year_Ending_June_2026.xlsx"

if (!file.exists(local_file)) {
  options(timeout = 600)
  download.file(data_url, local_file, mode = "wb")
}

# Find the sheets by name
sheets   <- excel_sheets(local_file)
sheet_01 <- sheets[str_detect(sheets, "01")][1]
sheet_02 <- sheets[str_detect(sheets, "02")][1]

# Helper: make column names simple (lower case, underscores)
clean_names_simple <- function(df) {
  names(df) <- names(df) %>% tolower() %>%
    str_replace_all("[^a-z0-9]+", "_") %>% str_remove_all("^_|_$")
  df
}

t1 <- read_excel(local_file, sheet = sheet_01) %>% clean_names_simple()
t2 <- read_excel(local_file, sheet = sheet_02) %>% clean_names_simple()

# 2. Give the columns short, standard names 
rename_by_pattern <- function(df, pattern, new_name, exclude = NULL) {
  hit <- str_detect(names(df), pattern)
  if (!is.null(exclude)) hit <- hit & !str_detect(names(df), exclude)
  names(df)[which(hit)[1]] <- new_name
  df
}

t1 <- t1 %>%
  rename_by_pattern("^year$", "year") %>%
  rename_by_pattern("government_area", "lga") %>%
  rename_by_pattern("region", "region") %>%
  rename_by_pattern("offence_count", "offences") %>%
  rename_by_pattern("rate", "rate") %>%
  mutate(year = as.integer(year),
         offences = suppressWarnings(as.numeric(offences)),
         rate = suppressWarnings(as.numeric(rate)))

t2 <- t2 %>%
  rename_by_pattern("^year$", "year") %>%
  rename_by_pattern("government_area", "lga") %>%
  rename_by_pattern("division", "division", exclude = "sub") %>%
  rename_by_pattern("subdivision", "subdivision") %>%
  rename_by_pattern("subgroup", "subgroup") %>%
  rename_by_pattern("offence_count", "offences") %>%
  mutate(year = as.integer(year),
         offences = suppressWarnings(as.numeric(offences)),
         justice = str_detect(division, "^E"))   # E = Justice procedures offences

# Helper: remove the CSA code at the start of a name, e.g. "E23 Breach bail conditions"
strip_code <- function(x) str_squish(str_remove(x, "^([A-Z][0-9]+|[0-9]+)\\s+"))

# 3. Population for each LGA (worked out from the CSA rate)
# rate = offences / population * 100,000  ->  population = offences / rate * 100,000
lga_pop <- t1 %>%
  filter(!is.na(rate), rate > 0, !is.na(region)) %>%
  mutate(pop = offences / rate * 1e5) %>%
  select(year, lga, region, pop)

first_yr <- min(t2$year)
last_yr  <- max(t2$year)

# 4. Statewide totals
state <- t2 %>%
  group_by(year) %>%
  summarise(all     = sum(offences, na.rm = TRUE),
            justice = sum(offences[justice], na.rm = TRUE),
            .groups = "drop") %>%
  mutate(other = all - justice,
         justice_share = justice / all * 100)

get <- function(col, yr) state[[col]][state$year == yr]

# Key numbers used in the text
total_first <- get("all", first_yr);      total_last <- get("all", last_yr)
just_first  <- get("justice", first_yr);  just_last  <- get("justice", last_yr)
other_first <- get("other", first_yr);    other_last <- get("other", last_yr)
total_chg <- (total_last / total_first - 1) * 100
just_chg  <- (just_last / just_first - 1) * 100
other_chg <- (other_last / other_first - 1) * 100
share_first <- get("justice_share", first_yr)
share_last  <- get("justice_share", last_yr)

# Text helpers
n_fmt  <- function(x) comma(round(x))
p_fmt  <- function(x) paste0(ifelse(x > 0, "+", ""), round(x, 1), "%")
up_dn  <- function(x) ifelse(x >= 0, "went up", "went down")

# Common chart settings
base_font <- list(family = "Arial", size = 14, color = "#222222")
cfg <- function(p) config(p, displaylogo = FALSE,
                          modeBarButtonsToRemove = c("lasso2d", "select2d"))

Why I looked at this

Every few months, we hear news about crime increasing in Victoria. According to Victoria Police (2026), breaches of orders were the biggest contributor to recorded crime in the year ending June 2026. This made me curious about which types of crime were actually increasing. After looking at the official data, I found that the biggest contributor was not burglary, assault, or car theft. Instead, it was offences related to breaking the rules of the justice system, such as breaching intervention orders, breaking bail conditions, and failing to appear in court. This report looks more closely at these offences and what the data tells us about crime in Victoria.

How to use this report: All charts are interactive. Hover over or tap the chart to see the exact numbers. Click on the legend to show or hide data, and use the buttons above each chart to change the view.

1. The headline: “crime is up”

In the year ending June 2017, Victoria Police recorded 529,764 offences. By the year ending June 2026, this number was 620,155, a change of +17.1%. That is the number most headlines use.

But look what happens when we split it in two. Offences against the justice system (breaches of orders, bail breaches, failing to appear and similar) went up by +45%. Everything else changed by just +12.9%. Press “% change since 2017” to see the gap clearly.

# ---- Chart 1: total crime vs justice procedures vs everything else ------------
c1 <- state %>%
  select(year, all, other, justice) %>%
  pivot_longer(-year, names_to = "series", values_to = "offences") %>%
  group_by(series) %>%
  mutate(change = (offences / offences[year == first_yr] - 1) * 100) %>%
  ungroup() %>%
  mutate(series = recode(series,
    all     = "All recorded offences",
    other   = "Everything except justice offences",
    justice = "Justice procedures offences"))

c1_cols   <- c("All recorded offences" = "#555555",
               "Everything except justice offences" = "#0072B2",
               "Justice procedures offences" = "#D55E00")
c1_dash   <- c("dot", "solid", "solid")   # line style, so colour is not the only cue
c1_symbol <- c("circle", "square", "diamond")

p1 <- plot_ly()
for (i in seq_along(c1_cols)) {          # traces 1-3: number of offences
  s <- names(c1_cols)[i]
  p1 <- p1 %>% add_trace(
    data = filter(c1, series == s), x = ~year, y = ~offences,
    type = "scatter", mode = "lines+markers", name = s, visible = TRUE,
    line = list(color = c1_cols[[s]], width = 3, dash = c1_dash[i]),
    marker = list(color = c1_cols[[s]], size = 8, symbol = c1_symbol[i]),
    hovertemplate = paste0("<b>", s, "</b><br>Year ending June %{x}<br>",
                           "%{y:,.0f} offences<extra></extra>"))
}
for (i in seq_along(c1_cols)) {          # traces 4-6: % change since first year
  s <- names(c1_cols)[i]
  p1 <- p1 %>% add_trace(
    data = filter(c1, series == s), x = ~year, y = ~change,
    type = "scatter", mode = "lines+markers", name = s, visible = FALSE,
    line = list(color = c1_cols[[s]], width = 3, dash = c1_dash[i]),
    marker = list(color = c1_cols[[s]], size = 8, symbol = c1_symbol[i]),
    hovertemplate = paste0("<b>", s, "</b><br>Year ending June %{x}<br>",
                           "%{y:+.1f}% compared with ", first_yr, "<extra></extra>"))
}

p1 %>%
  layout(
    title = list(text = paste0("<b>", ifelse(just_chg > other_chg,
                                 "Justice offences grew much faster than other crime",
                                 "Justice offences vs other crime"), "</b><br>",
                               "<sup>Recorded offences in Victoria, years ending June ",
                               first_yr, "-", last_yr, "</sup>"),
                 x = 0, xanchor = "left"),
    font = base_font,
    xaxis = list(title = "Year ending June", dtick = 1),
    yaxis = list(title = "Recorded offences", tickformat = ",", rangemode = "tozero"),
    legend = list(orientation = "h", y = -0.2, yanchor = "top"),
    margin = list(t = 110, b = 90),
    hovermode = "closest",
    updatemenus = list(list(
      type = "buttons", direction = "right", x = 1, xanchor = "right", y = 1.17,
      buttons = list(
        list(label = "Number of offences", method = "update",
             args = list(list(visible = c(TRUE, TRUE, TRUE, FALSE, FALSE, FALSE)),
                         list(yaxis = list(title = "Recorded offences", tickformat = ",",
                                           rangemode = "tozero")))),
        list(label = paste0("% change since ", first_yr), method = "update",
             args = list(list(visible = c(FALSE, FALSE, FALSE, TRUE, TRUE, TRUE)),
                         list(yaxis = list(title = paste0("% change since ", first_yr),
                                           ticksuffix = "%", zeroline = TRUE,
                                           rangemode = "normal")))))))) %>% cfg()