4. Geographic Distribution

Where Do LGBTQ+ Candidates Run?

Show code 
source(here::here("code", "00_setup.R"))
library(sf)

df       <- readRDS(paths$analysis_full_rds)
muni_sf  <- readRDS(paths$geo_muni_rds)
state_sf <- readRDS(paths$geo_state_rds)

1 Overview

Brazil is a country of continental proportions, with 5,570 municipalities spanning everything from Amazonian riverine communities to the megalopolis of Sao Paulo. LGBTQ+ visibility and acceptance vary enormously across this geography — shaped by urbanization, religion, regional political culture, and local institutional capacity.

This chapter maps the spatial distribution of LGBTQ+ candidates in the 2024 municipal elections, looking for concentration patterns, regional disparities, and the urban/rural dimension.

Results by Position Type

Where analyses involve candidacy rates or election rates, results are presented separately for city councilors (proportional representation) and mayors/vice-mayors (plurality/majority) using tabbed panels. Geographic distribution maps pool across positions because every municipality simultaneously holds elections for both position types, so the spatial pattern is primarily determined by municipality-level factors rather than position type.

2 National Distribution

2.1 LGBTQ+ Candidates Per Municipality

We begin by counting how many LGBTQ+ candidates ran in each of Brazil’s municipalities. The municipality-level count is the raw number of candidates identified as LGBTQ+ (via TSE self-declaration and/or VOTE LGBT matching) within each of the 5,570 municipalities.

Show code 
# Aggregate candidate counts per municipality
muni_counts <- df %>%
  mutate(geobr_code = as.numeric(geobr_code)) %>%
  group_by(geobr_code) %>%
  summarise(
    n_candidates     = n(),
    n_lgbtq          = sum(lgbtq_candidate, na.rm = TRUE),
    n_trans           = sum(trans_candidate, na.rm = TRUE),
    lgbtq_share      = if_else(n_candidates > 0, n_lgbtq / n_candidates, 0),
    .groups = "drop"
  )

# Join to shapefile
muni_map <- muni_sf %>%
  left_join(muni_counts, by = c("code_muni" = "geobr_code")) %>%
  mutate(
    n_lgbtq       = replace_na(n_lgbtq, 0),
    n_trans        = replace_na(n_trans, 0),
    n_candidates  = replace_na(n_candidates, 0),
    lgbtq_share   = replace_na(lgbtq_share, 0),
    lgbtq_cat     = case_when(
      n_lgbtq == 0              ~ "0",
      n_lgbtq == 1              ~ "1",
      n_lgbtq >= 2 & n_lgbtq <= 5 ~ "2-5",
      n_lgbtq >= 6              ~ "6+"
    ),
    lgbtq_cat = factor(lgbtq_cat, levels = c("0", "1", "2-5", "6+"))
  )
Show code 
p_count <- ggplot(muni_map) +

geom_sf(aes(fill = lgbtq_cat), color = NA, linewidth = 0) +
  geom_sf(data = state_sf, fill = NA, color = "gray30", linewidth = 0.3) +
  scale_fill_manual(
    values = c("0" = "#F0F0F0", "1" = "#FDBE85", "2-5" = "#E6550D", "6+" = "#A63603"),
    name   = "N LGBTQ+ candidates",
    na.value = "#F0F0F0"
  ) +
  labs(
    title    = "Number of LGBTQ+ Candidates by Municipality",
    subtitle = "Brazil's 2024 municipal elections",
    caption  = "Source: TSE self-disclosure + VOTE LGBT matching. Gray lines = state borders."
  ) +
  theme_void() +
  theme(
    plot.title    = element_text(face = "bold", size = 16),
    plot.subtitle = element_text(size = 12, color = "gray40"),
    plot.caption  = element_text(size = 9, color = "gray50", hjust = 0),
    legend.position = "bottom"
  )

p_count

save_figure(p_count, "04_map_lgbtq_count_muni", width = 12, height = 8)
Figure 1: LGBTQ+ Candidates Per Municipality (2024 Municipal Elections)
Show code 
# Summary statistics
n_muni_any  <- sum(muni_counts$n_lgbtq > 0, na.rm = TRUE)
n_muni_zero <- sum(muni_counts$n_lgbtq == 0 | is.na(muni_counts$n_lgbtq), na.rm = TRUE)
Geographic Reach

Out of 5,570 municipalities, 1,449 had at least one LGBTQ+ candidate, while 4,116 had none. Most LGBTQ+ candidacies are concentrated in a relatively small number of urban centers.

Maps Pool Across Positions

The choropleth maps above aggregate all LGBTQ+ candidates regardless of position type. City councilors constitute 93% of all candidates, so position-specific maps would be visually indistinguishable from these pooled versions. Position-disaggregated rates and counts are available in the tabbed sections below.

2.2 LGBTQ+ Share Per Municipality

LGBTQ+ share is the proportion of a municipality’s candidates who are identified as LGBTQ+. The choropleth below maps this share across all municipalities, distinguishing between those with zero LGBTQ+ candidates and those with progressively higher shares.

Show code 
# Create share categories (only for municipalities with candidates)
muni_map <- muni_map %>%
  mutate(
    share_cat = case_when(
      n_candidates == 0        ~ "No candidates",
      lgbtq_share == 0         ~ "0%",
      lgbtq_share > 0 & lgbtq_share <= 0.01 ~ "0.1-1%",
      lgbtq_share > 0.01 & lgbtq_share <= 0.03 ~ "1-3%",
      lgbtq_share > 0.03      ~ ">3%"
    ),
    share_cat = factor(share_cat,
                       levels = c("No candidates", "0%", "0.1-1%", "1-3%", ">3%"))
  )

p_share <- ggplot(muni_map) +
  geom_sf(aes(fill = share_cat), color = NA, linewidth = 0) +
  geom_sf(data = state_sf, fill = NA, color = "gray30", linewidth = 0.3) +
  scale_fill_manual(
    values = c(
      "No candidates" = "#F0F0F0",
      "0%"            = "#DEEBF7",
      "0.1-1%"        = "#9ECAE1",
      "1-3%"          = "#4292C6",
      ">3%"           = "#084594"
    ),
    name     = "LGBTQ+ share",
    na.value = "#F0F0F0"
  ) +
  labs(
    title    = "LGBTQ+ Candidate Share by Municipality",
    subtitle = "Percentage of all candidates who are LGBTQ+",
    caption  = "Source: TSE self-disclosure + VOTE LGBT matching."
  ) +
  theme_void() +
  theme(
    plot.title    = element_text(face = "bold", size = 16),
    plot.subtitle = element_text(size = 12, color = "gray40"),
    plot.caption  = element_text(size = 9, color = "gray50", hjust = 0),
    legend.position = "bottom"
  )

p_share

save_figure(p_share, "04_map_lgbtq_share_muni", width = 12, height = 8)
Figure 2: LGBTQ+ Candidate Share (%) Per Municipality

2.3 Top 20 Municipalities by LGBTQ+ Count

The table below lists the 20 municipalities with the highest absolute number of LGBTQ+ candidates, along with total candidates, trans candidates, and the LGBTQ+ share.

Show code 
top_munis <- muni_counts %>%
  filter(n_lgbtq > 0) %>%
  arrange(desc(n_lgbtq)) %>%
  head(20) %>%
  left_join(
    muni_sf %>% st_drop_geometry() %>% select(code_muni, name_muni, abbrev_state),
    by = c("geobr_code" = "code_muni")
  ) %>%
  mutate(
    lgbtq_share_fmt = format_pct(lgbtq_share),
    rank = row_number()
  ) %>%
  select(
    Rank          = rank,
    Municipality  = name_muni,
    State         = abbrev_state,
    `Total Cand.` = n_candidates,
    `LGBTQ+ Cand.` = n_lgbtq,
    `Trans Cand.` = n_trans,
    `LGBTQ+ %`    = lgbtq_share_fmt
  )

kable(top_munis, align = c("r", "l", "c", "r", "r", "r", "r"))
Table 1: Top 20 Municipalities by Number of LGBTQ+ Candidates
Rank Municipality State Total Cand. LGBTQ+ Cand. Trans Cand. LGBTQ+ %
1 Rio de Janeiro RJ 1046 28 3 2.7%
2 Salvador BA 866 27 3 3.1%
3 Belo Horizonte MG 898 27 5 3.0%
4 Porto Alegre RS 538 27 8 5.0%
5 São Paulo SP 1040 26 8 2.5%
6 Couto Magalhães TO 54 22 0 40.7%
7 Fortaleza CE 793 20 2 2.5%
8 Uberlândia MG 549 19 5 3.5%
9 Curitiba PR 776 19 1 2.4%
10 Florianópolis SC 327 18 3 5.5%
11 Ijuí RS 132 18 15 13.6%
12 Manaus AM 851 17 3 2.0%
13 Belém PA 606 14 3 2.3%
14 Natal RN 450 14 3 3.1%
15 Aracaju SE 508 14 7 2.8%
16 Paço do Lumiar MA 283 12 0 4.2%
17 São Luís MA 537 12 0 2.2%
18 Ribeirão Preto SP 426 12 3 2.8%
19 Goiânia GO 709 12 2 1.7%
20 Gravatá PE 228 11 1 4.8%
Urban Concentration

The composition of the top 20 municipalities reveals the degree to which LGBTQ+ candidacies concentrate in state capitals and large metropolitan centers, consistent with the sociological expectation that urban environments facilitate LGBTQ+ visibility and political mobilization.

Pooled Across Positions

The top municipalities table pools across position types. City councilors dominate the counts (93% of all candidates), so the ranking would be nearly identical if restricted to councilors only.

3 State-Level Patterns

The table below aggregates LGBTQ+ candidacy to the state level, reporting the total number of candidates, LGBTQ+ and trans counts, the LGBTQ+ rate per 1,000 candidates (a size-normalized measure), the LGBTQ+ share, and the LGBTQ+ election rate within each state.

Show code 
render_state <- function(data, tab_name) {
  state_summary <- data %>%
    group_by(state_abbrev) %>%
    summarise(
      n_total  = n(),
      n_lgbtq  = sum(lgbtq_candidate, na.rm = TRUE),
      n_trans  = sum(trans_candidate, na.rm = TRUE),
      lgbtq_rate_1k = round(n_lgbtq / n_total * 1000, 1),
      lgbtq_share   = n_lgbtq / n_total,
      election_rate_lgbtq = if_else(
        n_lgbtq > 0,
        mean(elected[lgbtq_candidate], na.rm = TRUE),
        NA_real_
      ),
      .groups = "drop"
    ) %>%
    mutate(region = state_to_region(state_abbrev)) %>%
    arrange(desc(n_lgbtq))

  state_summary %>%
    mutate(
      lgbtq_share = format_pct(lgbtq_share),
      election_rate_lgbtq = if_else(
        is.na(election_rate_lgbtq), "---",
        format_pct(election_rate_lgbtq)
      )
    ) %>%
    select(
      State       = state_abbrev,
      Region      = region,
      `Total Cand.` = n_total,
      `LGBTQ+`    = n_lgbtq,
      Trans       = n_trans,
      `LGBTQ+ per 1K` = lgbtq_rate_1k,
      `LGBTQ+ %`  = lgbtq_share,
      `LGBTQ+ Elect. Rate` = election_rate_lgbtq
    ) %>%
    cat_kable(align = c("l", "l", "r", "r", "r", "r", "r", "r"))

  p_state <- state_summary %>%
    ggplot(aes(x = reorder(state_abbrev, lgbtq_rate_1k),
               y = lgbtq_rate_1k, fill = region)) +
    geom_col(alpha = 0.9) +
    coord_flip() +
    scale_fill_manual(values = pal_region, name = "Region") +
    labs(
      x       = NULL,
      y       = "LGBTQ+ candidates per 1,000 total candidates",
      title   = paste0("LGBTQ+ Candidacy Rate by State (", tab_name, ")"),
      subtitle = "Normalized by total candidates in each state",
      caption  = "Higher rates may reflect both self-disclosure norms and VOTE LGBT coverage."
    )

  cat_plot(p_state, paste0("04-state-rate-", pos_suffix(tab_name)), height = 9)
}

render_position_tabset(render_state, df)
State Region Total Cand. LGBTQ+ Trans LGBTQ+ per 1K LGBTQ+ % LGBTQ+ Elect. Rate
SP Southeast 74044 514 109 6.9 0.7% 7.1%
MG Southeast 68745 400 72 5.8 0.6% 11.7%
BA Northeast 32879 257 37 7.8 0.8% 7.1%
RS South 26576 240 55 9.0 0.9% 10.3%
PR South 31719 161 40 5.1 0.5% 4.7%
CE Northeast 12223 156 34 12.8 1.3% 9.3%
PE Northeast 15002 139 34 9.3 0.9% 5.8%
RJ Southeast 16813 139 28 8.3 0.8% 5.9%
PA North 17136 127 24 7.4 0.7% 7.1%
SC South 17935 115 15 6.4 0.6% 10.4%
GO Center-West 18668 107 13 5.7 0.6% 5.6%
PB Northeast 9163 83 27 9.1 0.9% 5.1%
MA Northeast 15543 80 13 5.1 0.5% 12.5%
RN Northeast 6901 77 15 11.2 1.1% 9.2%
SE Northeast 5148 63 20 12.2 1.2% 0.0%
MT Center-West 10358 61 13 5.9 0.6% 3.5%
ES Southeast 9291 58 14 6.2 0.6% 3.5%
MS Center-West 6897 46 7 6.7 0.7% 5.6%
AL Northeast 5310 45 6 8.5 0.8% 6.7%
AM North 7691 42 14 5.5 0.5% 2.6%
TO North 6527 38 4 5.8 0.6% 16.2%
PI Northeast 7977 36 4 4.5 0.5% 3.0%
AC North 2183 17 1 7.8 0.8% 0.0%
RO North 4532 17 5 3.8 0.4% 0.0%
AP North 1486 15 4 10.1 1.0% 0.0%
RR North 1258 10 2 7.9 0.8% 0.0%

State Region Total Cand. LGBTQ+ Trans LGBTQ+ per 1K LGBTQ+ % LGBTQ+ Elect. Rate
MG Southeast 4709 15 1 3.2 0.3% 28.6%
SP Southeast 4317 15 0 3.5 0.3% 7.7%
RS South 2464 11 0 4.5 0.4% 9.1%
PR South 2302 9 0 3.9 0.4% 22.2%
BA Northeast 2326 8 1 3.4 0.3% 0.0%
RN Northeast 789 4 0 5.1 0.5% 25.0%
CE Northeast 998 3 1 3.0 0.3% 0.0%
MA Northeast 1296 3 0 2.3 0.2% 33.3%
PA North 959 3 0 3.1 0.3% 0.0%
TO North 669 3 0 4.5 0.4% 0.0%
GO Center-West 1311 2 0 1.5 0.2% 0.0%
PE Northeast 1067 2 0 1.9 0.2% 0.0%
PI Northeast 1001 2 0 2.0 0.2% 0.0%
RJ Southeast 777 2 0 2.6 0.3% 0.0%
AL Northeast 524 1 0 1.9 0.2% 0.0%
AP North 110 1 1 9.1 0.9% 0.0%
ES Southeast 570 1 0 1.8 0.2% 0.0%
MS Center-West 475 1 0 2.1 0.2% 0.0%
MT Center-West 746 1 0 1.3 0.1% 0.0%
PB Northeast 1079 1 0 0.9 0.1% 0.0%
RO North 346 1 0 2.9 0.3% 0.0%
RR North 87 1 0 11.5 1.1% 0.0%
SC South 1685 1 0 0.6 0.1% 0.0%
AC North 128 0 0 0.0 0.0% —
AM North 423 0 0 0.0 0.0% —
SE Northeast 438 0 0 0.0 0.0% —

Note

This tab pools city councilors (proportional representation) and mayors/vice-mayors (plurality). Position-specific results in the other tabs may be more informative.

State Region Total Cand. LGBTQ+ Trans LGBTQ+ per 1K LGBTQ+ % LGBTQ+ Elect. Rate
SP Southeast 78361 529 109 6.8 0.7% 7.1%
MG Southeast 73454 415 73 5.6 0.6% 12.3%
BA Northeast 35205 265 38 7.5 0.8% 6.9%
RS South 29040 251 55 8.6 0.9% 10.2%
PR South 34021 170 40 5.0 0.5% 5.7%
CE Northeast 13221 159 35 12.0 1.2% 9.1%
PE Northeast 16069 141 34 8.8 0.9% 5.7%
RJ Southeast 17590 141 28 8.0 0.8% 5.8%
PA North 18095 130 24 7.2 0.7% 7.0%
SC South 19620 116 15 5.9 0.6% 10.3%
GO Center-West 19979 109 13 5.5 0.5% 5.4%
PB Northeast 10242 84 27 8.2 0.8% 5.0%
MA Northeast 16839 83 13 4.9 0.5% 13.3%
RN Northeast 7690 81 15 10.5 1.1% 10.0%
SE Northeast 5586 63 20 11.3 1.1% 0.0%
MT Center-West 11104 62 13 5.6 0.6% 3.4%
ES Southeast 9861 59 14 6.0 0.6% 3.4%
MS Center-West 7372 47 7 6.4 0.6% 5.4%
AL Northeast 5834 46 6 7.9 0.8% 6.5%
AM North 8114 42 14 5.2 0.5% 2.6%
TO North 7196 41 4 5.7 0.6% 15.0%
PI Northeast 8978 38 4 4.2 0.4% 2.9%
RO North 4878 18 5 3.7 0.4% 0.0%
AC North 2311 17 1 7.4 0.7% 0.0%
AP North 1596 16 5 10.0 1.0% 0.0%
RR North 1345 11 2 8.2 0.8% 0.0%

4 Regional Patterns

The table below aggregates to Brazil’s five macro-regions, reporting total and LGBTQ+ candidate counts, the LGBTQ+ rate per 1,000 candidates, and election rates for all candidates and LGBTQ+ candidates separately.

Show code 
render_region <- function(data, tab_name) {
  region_summary <- data %>%
    filter(!is.na(region)) %>%
    group_by(region) %>%
    summarise(
      n_total      = n(),
      n_lgbtq      = sum(lgbtq_candidate, na.rm = TRUE),
      n_trans       = sum(trans_candidate, na.rm = TRUE),
      lgbtq_share  = n_lgbtq / n_total,
      lgbtq_rate_1k = round(n_lgbtq / n_total * 1000, 1),
      elect_rate_all   = mean(elected, na.rm = TRUE),
      elect_rate_lgbtq = mean(elected[lgbtq_candidate], na.rm = TRUE),
      .groups = "drop"
    ) %>%
    arrange(desc(n_lgbtq))

  region_summary %>%
    mutate(
      lgbtq_share      = format_pct(lgbtq_share),
      elect_rate_all   = format_pct(elect_rate_all),
      elect_rate_lgbtq = format_pct(elect_rate_lgbtq)
    ) %>%
    select(
      Region       = region,
      `Total Cand.` = n_total,
      `LGBTQ+`     = n_lgbtq,
      Trans        = n_trans,
      `LGBTQ+ %`   = lgbtq_share,
      `LGBTQ+ per 1K` = lgbtq_rate_1k,
      `Elect. Rate (All)` = elect_rate_all,
      `Elect. Rate (LGBTQ+)` = elect_rate_lgbtq
    ) %>%
    cat_kable(align = c("l", "r", "r", "r", "r", "r", "r", "r"))

  p_region_rate <- region_summary %>%
    ggplot(aes(x = reorder(region, lgbtq_rate_1k), y = lgbtq_rate_1k, fill = region)) +
    geom_col(alpha = 0.9, show.legend = FALSE) +
    geom_text(aes(label = lgbtq_rate_1k), hjust = -0.2, size = 4) +
    coord_flip() +
    scale_fill_manual(values = pal_region) +
    scale_y_continuous(expand = expansion(mult = c(0, 0.2))) +
    labs(x = NULL, y = "LGBTQ+ per 1,000 candidates",
         title = paste0("LGBTQ+ Candidacy Rate by Region (", tab_name, ")"))

  p_region_elect <- region_summary %>%
    select(region, `All candidates` = elect_rate_all, `LGBTQ+ candidates` = elect_rate_lgbtq) %>%
    pivot_longer(-region, names_to = "group", values_to = "rate") %>%
    ggplot(aes(x = reorder(region, rate), y = rate, fill = group)) +
    geom_col(position = "dodge", alpha = 0.9, width = 0.7) +
    geom_text(aes(label = format_pct(rate)),
              position = position_dodge(width = 0.7), hjust = -0.2, size = 3.5) +
    coord_flip() +
    scale_fill_manual(values = c("All candidates" = "#3498DB", "LGBTQ+ candidates" = "#E74C3C"),
                      name = NULL) +
    scale_y_continuous(labels = percent, expand = expansion(mult = c(0, 0.25))) +
    labs(x = NULL, y = "Election Rate",
         title = paste0("Election Rate by Region: LGBTQ+ vs All (", tab_name, ")"))

  cat_plot(p_region_rate / p_region_elect,
           paste0("04-region-rates-", pos_suffix(tab_name)),
           width = 10, height = 10)
}

render_position_tabset(render_region, df)
Region Total Cand. LGBTQ+ Trans LGBTQ+ % LGBTQ+ per 1K Elect. Rate (All) Elect. Rate (LGBTQ+)
Southeast 168893 1111 223 0.7% 6.6 10.8% 8.4%
Northeast 110146 936 190 0.8% 8.5 18.4% 7.2%
South 76230 516 110 0.7% 6.8 15.9% 8.6%
North 40813 266 54 0.7% 6.5 12.8% 6.1%
Center-West 35923 214 33 0.6% 6.0 14.1% 4.9%

Region Total Cand. LGBTQ+ Trans LGBTQ+ % LGBTQ+ per 1K Elect. Rate (All) Elect. Rate (LGBTQ+)
Southeast 10373 33 1 0.3% 3.2 32.9% 16.7%
Northeast 9518 24 2 0.3% 2.5 39.2% 8.3%
South 6451 21 0 0.3% 3.3 37.8% 14.3%
North 2722 9 1 0.3% 3.3 34.4% 0.0%
Center-West 2532 4 0 0.2% 1.6 38.2% 0.0%

Note

This tab pools city councilors (proportional representation) and mayors/vice-mayors (plurality). Position-specific results in the other tabs may be more informative.

Region Total Cand. LGBTQ+ Trans LGBTQ+ % LGBTQ+ per 1K Elect. Rate (All) Elect. Rate (LGBTQ+)
Southeast 179266 1144 224 0.6% 6.4 12.0% 8.7%
Northeast 119664 960 192 0.8% 8.0 20.1% 7.2%
South 82681 537 110 0.6% 6.5 17.6% 8.9%
North 43535 275 55 0.6% 6.3 14.2% 6.0%
Center-West 38455 218 33 0.6% 5.7 15.8% 4.8%

Northeast vs Southeast

The Southeast — anchored by Sao Paulo and Rio de Janeiro — has the largest absolute number of LGBTQ+ candidates, reflecting its population weight. However, when normalized per 1,000 candidates, the pattern may shift. The Northeast, with the largest overall candidate pool, shows a distinct LGBTQ+ candidacy profile shaped by both the region’s political dynamics and the reach of VOTE LGBT identification efforts.

5 Urban vs Rural Proxy

We lack direct population or urbanization data at the municipality level in this dataset. However, the total number of candidates in a municipality serves as a rough proxy for its size: larger municipalities field more candidates.

We compare municipalities that had at least one LGBTQ+ candidate to those with zero, using the total candidate count as a size proxy. The table reports the number of municipalities, mean and median candidate counts, and total candidates for each group.

Show code 
urban_rural <- muni_counts %>%
  mutate(has_lgbtq = n_lgbtq > 0) %>%
  group_by(has_lgbtq) %>%
  summarise(
    n_municipalities = n(),
    mean_candidates  = round(mean(n_candidates, na.rm = TRUE), 1),
    median_candidates = median(n_candidates, na.rm = TRUE),
    total_candidates = sum(n_candidates, na.rm = TRUE),
    .groups = "drop"
  ) %>%
  mutate(
    has_lgbtq = if_else(has_lgbtq, "1+ LGBTQ+ candidates", "0 LGBTQ+ candidates")
  )

urban_rural %>%
  rename(
    Group             = has_lgbtq,
    `N Municipalities` = n_municipalities,
    `Mean Candidates`  = mean_candidates,
    `Median Candidates` = median_candidates,
    `Total Candidates` = total_candidates
  ) %>%
  kable(align = c("l", "r", "r", "r", "r"))
Table 2: Municipalities With vs Without LGBTQ+ Candidates
Group N Municipalities Mean Candidates Median Candidates Total Candidates
0 LGBTQ+ candidates 4116 61.9 51 254768
1+ LGBTQ+ candidates 1449 144.1 110 208833
Show code 
muni_counts %>%
  mutate(has_lgbtq = if_else(n_lgbtq > 0, "Has LGBTQ+ candidate(s)", "No LGBTQ+ candidates")) %>%
  filter(n_candidates > 0) %>%
  ggplot(aes(x = n_candidates, fill = has_lgbtq)) +
  geom_histogram(binwidth = 20, alpha = 0.7, position = "identity", color = "white") +
  scale_x_continuous(limits = c(0, 500)) +
  scale_fill_manual(values = c("Has LGBTQ+ candidate(s)" = "#E74C3C",
                                "No LGBTQ+ candidates" = "#3498DB"),
                    name = NULL) +
  labs(
    x        = "Total Candidates in Municipality (capped at 500 for visibility)",
    y        = "Number of Municipalities",
    title    = "Municipality Size: With vs Without LGBTQ+ Candidates",
    subtitle = "Distribution of total candidates per municipality, by presence of LGBTQ+ candidates"
  )
Figure 3: Municipality Size Distribution: With vs Without LGBTQ+ Candidates
Interpretation

The comparison of municipality sizes (proxied by total candidate counts) between places with and without LGBTQ+ candidates reveals the degree to which LGBTQ+ candidacies concentrate in larger, more urbanized settings — where social acceptance and community infrastructure may facilitate political participation.

Pooled Across Positions

The urban/rural comparison pools across position types because it examines municipality-level characteristics (candidate pool size as urbanization proxy) rather than position-specific outcomes.

6 Municipal Context

The previous sections mapped where LGBTQ+ candidates run. This section connects candidacy patterns to municipality-level characteristics: population size, economic development, and the local political environment measured by Bolsonaro’s 2022 first-round vote share.

6.1 LGBTQ+ Candidacy Rate by Municipality Size

Show code 
render_pop_bracket <- function(data, tab_name) {
  muni_pop_counts <- data %>%
    filter(!is.na(pop_bracket)) %>%
    group_by(pop_bracket) %>%
    summarise(n_munis = n_distinct(geobr_code), .groups = "drop")

  pop_summary <- data %>%
    filter(!is.na(pop_bracket)) %>%
    group_by(pop_bracket) %>%
    summarise(
      n_candidates = n(),
      n_lgbtq      = sum(lgbtq_candidate),
      rate_per_1k  = round(n_lgbtq / n_candidates * 1000, 1),
      elected_lgbtq = sum(lgbtq_candidate & elected, na.rm = TRUE),
      election_rate = round(elected_lgbtq / pmax(n_lgbtq, 1) * 100, 1),
      .groups = "drop"
    ) %>%
    left_join(muni_pop_counts, by = "pop_bracket")

  pop_summary %>%
    select(`Population Bracket` = pop_bracket,
           Municipalities = n_munis,
           `Total Candidates` = n_candidates,
           `LGBTQ+ Candidates` = n_lgbtq,
           `Rate per 1,000` = rate_per_1k,
           `LGBTQ+ Election Rate (%)` = election_rate) %>%
    cat_kable(align = c("l", rep("r", 5)), format.args = list(big.mark = ","))

  p_pop <- ggplot(pop_summary, aes(x = pop_bracket, y = rate_per_1k)) +
    geom_col(fill = pal_lgbtq["LGBTQ+"], alpha = 0.85) +
    geom_text(aes(label = rate_per_1k), vjust = -0.3, size = 4) +
    labs(
      x = "Municipality Population",
      y = "LGBTQ+ Candidates per 1,000",
      title = paste0("LGBTQ+ Candidacy Rate by Municipality Size (", tab_name, ")")
    )

  cat_plot(p_pop, paste0("04-pop-bracket-", pos_suffix(tab_name)))
}

render_position_tabset(render_pop_bracket, df)
Population Bracket Municipalities Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
< 10K 2,516 100,424 361 3.6 10.8
10K-50K 2,390 184,879 1,017 5.5 8.0
50K-200K 505 90,515 811 9.0 4.8
200K-500K 110 33,370 400 12.0 6.8
500K+ 43 22,574 450 19.9 6.9

Population Bracket Municipalities Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
< 10K 2,516 11,683 10 0.9 50.0
10K-50K 2,390 14,022 32 2.3 9.4
50K-200K 505 3,974 21 5.3 9.5
200K-500K 110 1,241 14 11.3 0.0
500K+ 43 656 14 21.3 0.0

Note

This tab pools city councilors (proportional representation) and mayors/vice-mayors (plurality). Position-specific results in the other tabs may be more informative.

Population Bracket Municipalities Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
< 10K 2,516 112,107 371 3.3 11.9
10K-50K 2,390 198,901 1,049 5.3 8.0
50K-200K 505 94,489 832 8.8 4.9
200K-500K 110 34,611 414 12.0 6.5
500K+ 43 23,230 464 20.0 6.7

6.2 Electorate Conservatism and LGBTQ+ Candidacy

Bolsonaro’s 2022 first-round presidential vote share serves as a proxy for local electorate conservatism. Municipalities are grouped into quartiles from least (Q1) to most (Q4) conservative.

Show code 
render_bolsonaro <- function(data, tab_name) {
  bolso_summary <- data %>%
    filter(!is.na(bolsonaro_quartile)) %>%
    group_by(bolsonaro_quartile) %>%
    summarise(
      n_candidates = n(),
      n_lgbtq      = sum(lgbtq_candidate),
      rate_per_1k  = round(n_lgbtq / n_candidates * 1000, 1),
      mean_bolso   = round(mean(bolsonaro_share, na.rm = TRUE), 1),
      elected_lgbtq = sum(lgbtq_candidate & elected, na.rm = TRUE),
      election_rate = round(elected_lgbtq / pmax(n_lgbtq, 1) * 100, 1),
      .groups = "drop"
    )

  bolso_summary %>%
    select(`Quartile` = bolsonaro_quartile,
           `Mean Bolsonaro %` = mean_bolso,
           `Total Candidates` = n_candidates,
           `LGBTQ+ Candidates` = n_lgbtq,
           `Rate per 1,000` = rate_per_1k,
           `LGBTQ+ Election Rate (%)` = election_rate) %>%
    cat_kable(align = c("l", rep("r", 5)), format.args = list(big.mark = ","))

  p_bolso <- ggplot(bolso_summary, aes(x = bolsonaro_quartile, y = rate_per_1k)) +
    geom_col(fill = pal_ideology["Right"], alpha = 0.85) +
    geom_text(aes(label = rate_per_1k), vjust = -0.3, size = 4) +
    labs(
      x = "Bolsonaro 2022 Vote Share Quartile",
      y = "LGBTQ+ Candidates per 1,000",
      title = paste0("LGBTQ+ Candidacy Rate by Local Conservatism (", tab_name, ")")
    )

  cat_plot(p_bolso, paste0("04-bolsonaro-", pos_suffix(tab_name)))
}

render_position_tabset(render_bolsonaro, df)
Quartile Mean Bolsonaro % Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
Q1 (least conservative) 21.6 94,175 679 7.2 8.0
Q2 39.7 96,187 777 8.1 8.8
Q3 50.7 96,687 705 7.3 6.8
Q4 (most conservative) 61.5 96,230 578 6.0 5.0

Quartile Mean Bolsonaro % Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
Q1 (least conservative) 20.5 8,762 18 2.1 16.7
Q2 39.6 6,704 22 3.3 0.0
Q3 50.6 6,319 24 3.8 16.7
Q4 (most conservative) 62.0 6,526 18 2.8 11.1

Note

This tab pools city councilors (proportional representation) and mayors/vice-mayors (plurality). Position-specific results in the other tabs may be more informative.

Quartile Mean Bolsonaro % Total Candidates LGBTQ+ Candidates Rate per 1,000 LGBTQ+ Election Rate (%)
Q1 (least conservative) 21.5 102,937 697 6.8 8.2
Q2 39.7 102,891 799 7.8 8.5
Q3 50.7 103,006 729 7.1 7.1
Q4 (most conservative) 61.5 102,756 596 5.8 5.2

Conservatism Gradient

The relationship between electorate conservatism and LGBTQ+ candidacy rates is consistent with theoretical expectations: localities with more conservative electorates have fewer LGBTQ+ candidates per capita, possibly reflecting higher perceived costs of disclosure and reduced party incentives to recruit LGBTQ+ candidates.

6.3 Municipality Size vs LGBTQ+ Share

Show code 
scatter_data <- df %>%
  filter(!is.na(populacao_2022)) %>%
  group_by(geobr_code, region, populacao_2022) %>%
  summarise(
    n_cand  = n(),
    n_lgbtq = sum(lgbtq_candidate),
    .groups = "drop"
  ) %>%
  filter(n_lgbtq > 0) %>%
  mutate(lgbtq_share = n_lgbtq / n_cand * 100)

ggplot(scatter_data, aes(x = populacao_2022, y = lgbtq_share, color = region)) +
  geom_point(alpha = 0.4, size = 1.5) +
  geom_smooth(aes(group = 1), method = "loess", color = "black",
              linewidth = 1, se = TRUE, alpha = 0.2) +
  scale_x_log10(labels = scales::label_number(scale_cut = scales::cut_short_scale())) +
  scale_color_manual(values = pal_region) +
  labs(
    x = "Municipality Population (log scale)",
    y = "LGBTQ+ Candidate Share (%)",
    color = "Region",
    title = "LGBTQ+ Share Increases with Municipality Size",
    subtitle = "Each dot = one municipality with at least one LGBTQ+ candidate"
  )
Figure 4: Municipality Population vs LGBTQ+ Candidate Share
Pooled Across Positions

This scatter plot pools across position types because it examines municipality-level characteristics (population vs. LGBTQ+ candidate share). The relationship between municipality size and LGBTQ+ candidacy is driven by structural factors that apply across position types.

7 Geographic Concentration: Lorenz Curve

The Lorenz curve quantifies how concentrated LGBTQ+ candidates are across municipalities. Municipalities are sorted from fewest to most LGBTQ+ candidates on the x-axis, and the y-axis shows the cumulative share of all LGBTQ+ candidates. If candidates were evenly distributed, the curve would follow the 45-degree line of equality. Greater curvature indicates more concentration. The Gini coefficient (ranging from 0 to 1) summarizes the area between the Lorenz curve and the equality line, where higher values indicate greater concentration.

Show code 
# Sort municipalities by LGBTQ+ count and compute cumulative shares
lorenz_data <- muni_counts %>%
  filter(!is.na(n_lgbtq)) %>%
  arrange(n_lgbtq) %>%
  mutate(
    cum_munis   = row_number() / n(),
    cum_lgbtq   = cumsum(n_lgbtq) / sum(n_lgbtq)
  )

# Compute Gini coefficient (area between Lorenz curve and equality line)
gini <- 1 - 2 * sum(
  diff(lorenz_data$cum_munis) *
  (head(lorenz_data$cum_lgbtq, -1) + tail(lorenz_data$cum_lgbtq, -1)) / 2
)

# Find key concentration stats
pct_munis_80 <- lorenz_data %>%
  filter(cum_lgbtq >= 0.80) %>%
  slice_min(cum_lgbtq) %>%
  pull(cum_munis)

ggplot(lorenz_data, aes(x = cum_munis, y = cum_lgbtq)) +
  geom_line(color = "#E74C3C", linewidth = 1.2) +
  geom_abline(slope = 1, intercept = 0, linetype = "dashed", color = "gray50") +
  geom_ribbon(aes(ymin = cum_munis, ymax = cum_lgbtq), alpha = 0.15, fill = "#E74C3C") +
  annotate("text", x = 0.35, y = 0.85,
           label = paste0("Gini = ", round(gini, 3)),
           size = 5, fontface = "bold", color = "#E74C3C") +
  annotate("text", x = 0.35, y = 0.78,
           label = paste0(round((1 - pct_munis_80) * 100, 0),
                          "% of municipalities account\nfor 80% of LGBTQ+ candidates"),
           size = 3.8, color = "gray30") +
  scale_x_continuous(labels = percent, expand = c(0, 0.01)) +
  scale_y_continuous(labels = percent, expand = c(0, 0.01)) +
  labs(
    x       = "Cumulative % of municipalities (sorted by LGBTQ+ count)",
    y       = "Cumulative % of LGBTQ+ candidates",
    title   = "Geographic Concentration of LGBTQ+ Candidates",
    subtitle = "Lorenz curve across 5,570 municipalities",
    caption  = "Dashed line = perfect equality. Shaded area = Gini coefficient."
  )

save_figure(last_plot(), "04_lorenz_curve", width = 7, height = 6)
Figure 5: Lorenz Curve: Geographic Concentration of LGBTQ+ Candidates Across Municipalities
Concentration

A Gini coefficient of 0.846 indicates very high geographic concentration. LGBTQ+ candidates are overwhelmingly clustered in a small number of municipalities, leaving the vast majority of Brazilian cities without any identified LGBTQ+ political representation.

Pooled Across Positions

Geographic concentration is measured across all LGBTQ+ candidates regardless of position type. Since city councilors constitute 93% of all candidates, position-specific Lorenz curves would be nearly identical to the pooled version shown here.

8 Trans Candidate Geography

Trans candidates represent a small but politically significant subgroup. Their geographic distribution warrants specific attention. The point map below shows the location and count of trans candidates at the municipality level, while subsequent tables disaggregate by state.

Show code 
# Get centroids of municipalities with trans candidates for a point map
trans_munis <- muni_map %>%
  filter(n_trans > 0)

trans_centroids <- trans_munis %>%
  st_centroid()

p_trans <- ggplot() +
  geom_sf(data = state_sf, fill = "#F5F5F5", color = "gray60", linewidth = 0.3) +
  geom_sf(data = trans_centroids,
          aes(size = n_trans),
          color = "#F39C12", alpha = 0.7) +
  scale_size_continuous(
    range = c(1.5, 8),
    name  = "N trans candidates",
    breaks = c(1, 2, 5, 10, 20)
  ) +
  labs(
    title    = "Where Trans Candidates Run",
    subtitle = "Each point is a municipality with at least one trans candidate",
    caption  = "Source: TSE self-disclosure + VOTE LGBT matching. Point size = number of trans candidates."
  ) +
  theme_void() +
  theme(
    plot.title    = element_text(face = "bold", size = 16),
    plot.subtitle = element_text(size = 12, color = "gray40"),
    plot.caption  = element_text(size = 9, color = "gray50", hjust = 0),
    legend.position = "bottom"
  )

p_trans

save_figure(p_trans, "04_map_trans_candidates", width = 12, height = 8)
Figure 6: Trans Candidate Locations Across Brazil
Show code 
df %>%
  filter(trans_candidate) %>%
  count(state_abbrev, sort = TRUE) %>%
  left_join(
    df %>% count(state_abbrev, name = "total"),
    by = "state_abbrev"
  ) %>%
  mutate(
    region = state_to_region(state_abbrev),
    rate_per_1k = round(n / total * 1000, 2),
    pct = format_pct(n / total)
  ) %>%
  select(
    State      = state_abbrev,
    Region     = region,
    `Trans Cand.` = n,
    `Total Cand.` = total,
    `Per 1K`   = rate_per_1k,
    `% of Total` = pct
  ) %>%
  kable(align = c("l", "l", "r", "r", "r", "r"))
Table 3: Trans Candidates by State (States with Trans Candidates)
State Region Trans Cand. Total Cand. Per 1K % of Total
SP Southeast 109 78361 1.39 0.1%
MG Southeast 73 73454 0.99 0.1%
RS South 55 29040 1.89 0.2%
PR South 40 34021 1.18 0.1%
BA Northeast 38 35205 1.08 0.1%
CE Northeast 35 13221 2.65 0.3%
PE Northeast 34 16069 2.12 0.2%
RJ Southeast 28 17590 1.59 0.2%
PB Northeast 27 10242 2.64 0.3%
PA North 24 18095 1.33 0.1%
SE Northeast 20 5586 3.58 0.4%
RN Northeast 15 7690 1.95 0.2%
SC South 15 19620 0.76 0.1%
AM North 14 8114 1.73 0.2%
ES Southeast 14 9861 1.42 0.1%
GO Center-West 13 19979 0.65 0.1%
MA Northeast 13 16839 0.77 0.1%
MT Center-West 13 11104 1.17 0.1%
MS Center-West 7 7372 0.95 0.1%
AL Northeast 6 5834 1.03 0.1%
AP North 5 1596 3.13 0.3%
RO North 5 4878 1.03 0.1%
PI Northeast 4 8978 0.45 0.0%
TO North 4 7196 0.56 0.1%
RR North 2 1345 1.49 0.1%
AC North 1 2311 0.43 0.0%
Show code 
# Aggregate trans counts to state level
trans_state <- df %>%
  group_by(state_abbrev) %>%
  summarise(
    n_total = n(),
    n_trans = sum(trans_candidate, na.rm = TRUE),
    trans_rate_1k = round(n_trans / n_total * 1000, 2),
    .groups = "drop"
  )

# Join to state shapefile
state_trans_map <- state_sf %>%
  left_join(trans_state, by = c("abbrev_state" = "state_abbrev"))

p_trans_state <- ggplot(state_trans_map) +
  geom_sf(aes(fill = trans_rate_1k), color = "white", linewidth = 0.4) +
  scale_fill_gradient(
    low  = "#FFF5EB",
    high = "#D94801",
    name = "Trans candidates\nper 1,000",
    na.value = "#F0F0F0"
  ) +
  labs(
    title    = "Trans Candidacy Rate by State",
    subtitle = "Trans candidates per 1,000 total candidates",
    caption  = "Small absolute numbers; interpret with caution."
  ) +
  theme_void() +
  theme(
    plot.title    = element_text(face = "bold", size = 16),
    plot.subtitle = element_text(size = 12, color = "gray40"),
    plot.caption  = element_text(size = 9, color = "gray50", hjust = 0),
    legend.position = "bottom"
  )

p_trans_state

save_figure(p_trans_state, "04_map_trans_state_rate", width = 10, height = 7)
Figure 7: State-Level Trans Candidacy Rate
Small-N Caution

Trans candidates constitute a small fraction of all LGBTQ+ candidates, who themselves are a small fraction of all candidates. Geographic patterns at the municipality level should be interpreted cautiously. State-level aggregation is somewhat more stable, but rates based on very small denominators (e.g., states with 1-2 trans candidates) can be misleading.

9 Electoral Success Geography

The sections above map where LGBTQ+ candidates run. This section focuses on where they win.

Show code 
render_elected_geo <- function(data, tab_name) {
  elected_state <- data %>%
    filter(lgbtq_candidate, elected) %>%
    count(state_abbrev, name = "elected_lgbtq") %>%
    left_join(
      data %>% filter(lgbtq_candidate) %>% count(state_abbrev, name = "total_lgbtq"),
      by = "state_abbrev"
    ) %>%
    mutate(election_rate = round(elected_lgbtq / total_lgbtq * 100, 1)) %>%
    arrange(desc(elected_lgbtq))

  elected_state %>%
    select(State = state_abbrev,
           `LGBTQ+ Candidates` = total_lgbtq,
           `Elected` = elected_lgbtq,
           `Election Rate (%)` = election_rate) %>%
    cat_kable(align = c("l", "r", "r", "r"))

  elected_state_geo <- state_sf %>%
    left_join(elected_state, by = c("abbrev_state" = "state_abbrev")) %>%
    mutate(elected_lgbtq = replace_na(elected_lgbtq, 0))

  p_elected <- ggplot(elected_state_geo) +
    geom_sf(aes(fill = elected_lgbtq), color = "white", linewidth = 0.3) +
    scale_fill_gradient(low = "#f7fbff", high = "#08519c",
                        name = "Elected LGBTQ+\nCandidates") +
    labs(
      title = paste0("Where LGBTQ+ Candidates Win (", tab_name, ")"),
      subtitle = "Number of elected LGBTQ+ candidates by state"
    ) +
    theme_void() +
    theme(legend.position = "right")

  cat_plot(p_elected, paste0("04-map-elected-", pos_suffix(tab_name)), height = 8)
}

render_position_tabset(render_elected_geo, df)
State LGBTQ+ Candidates Elected Election Rate (%)
MG 400 45 11.2
SP 514 34 6.6
RS 240 24 10.0
BA 257 17 6.6
CE 156 14 9.0
SC 115 12 10.4
MA 80 9 11.2
PA 127 8 6.3
RJ 139 8 5.8
PR 161 7 4.3
RN 77 7 9.1
PE 139 6 4.3
TO 38 6 15.8
GO 107 5 4.7
PB 83 4 4.8
AL 45 3 6.7
ES 58 2 3.4
MS 46 2 4.3
MT 61 2 3.3
AM 42 1 2.4
PI 36 1 2.8

State LGBTQ+ Candidates Elected Election Rate (%)
MG 15 4 26.7
PR 9 2 22.2
MA 3 1 33.3
RN 4 1 25.0
RS 11 1 9.1
SP 15 1 6.7

Note

This tab pools city councilors (proportional representation) and mayors/vice-mayors (plurality). Position-specific results in the other tabs may be more informative.

State LGBTQ+ Candidates Elected Election Rate (%)
MG 415 49 11.8
SP 529 35 6.6
RS 251 25 10.0
BA 265 17 6.4
CE 159 14 8.8
SC 116 12 10.3
MA 83 10 12.0
PR 170 9 5.3
PA 130 8 6.2
RJ 141 8 5.7
RN 81 8 9.9
PE 141 6 4.3
TO 41 6 14.6
GO 109 5 4.6
PB 84 4 4.8
AL 46 3 6.5
ES 59 2 3.4
MS 47 2 4.3
MT 62 2 3.2
AM 42 1 2.4
PI 38 1 2.6

10 Summary

This chapter documents the geography of LGBTQ+ candidacies in Brazil’s 2024 municipal elections. Out of 5,570 municipalities, only 1,449 had at least one LGBTQ+ candidate, and the Lorenz curve and Gini coefficient quantify this concentration.

Municipality-level context reveals two key gradients: LGBTQ+ candidacy rates increase with population size and decrease with electorate conservatism (proxied by Bolsonaro’s 2022 vote share). The electoral success map shows where LGBTQ+ candidates not only run but win. Trans-specific maps show the spatial distribution of this smaller subgroup. These patterns have important implications for both representation and research design — any analysis of LGBTQ+ candidate outcomes must account for the non-random geographic selection into candidacy.