Two Decades of New Mexico Electoral Politics

if (!requireNamespace("pacman", quietly = TRUE)) {
  install.packages("pacman")
}

suppressPackageStartupMessages(
  pacman::p_load(
    dplyr,
    ggplot2,
    tidyr,
    viridis,
    scales,
    patchwork,
    purrr,
    rlang,
    DT,
    knitr
  )
)

# Load final results data from GitHub
results <- read.csv("https://raw.githubusercontent.com/jaytimm/newmexico-political-data/main/data/elections/nm_election_results_2000-24.csv", 
                    stringsAsFactors = FALSE)

# Define office categories
statewide_offices <- c("Governor", "Attorney General", "Secretary of State", 
                      "State Treasurer", "State Auditor", "Commissioner of Public Lands",
                      "President of the United States", "United States Senator")

legislative_offices <- c("State Representative", "State Senate")

federal_offices <- c("United States Representative", "United States Senator")

# Load county-level statewide results from GitHub
county_results <- read.csv("https://raw.githubusercontent.com/jaytimm/newmexico-political-data/main/data/elections/nm_county_statewide_results_2000-24.csv", 
                           stringsAsFactors = FALSE)

# Load precinct-level data from GitHub (2004-2024)
precinct_data <- read.csv("https://raw.githubusercontent.com/jaytimm/newmexico-political-data/main/data/elections/nm_precinct_results_2004-24.csv", 
                          stringsAsFactors = FALSE)

# Show first 100 records
results |>
  head(100) |>
  datatable(
    options = list(
      pageLength = 10,
      scrollX = TRUE
    ),
    rownames = FALSE
  )

# Create heatmap of vote margins by office and year
# Filter to 2002-2022 and order by importance
statewide_offices_list <- c("Governor", "Secretary of State", "Attorney General", 
                            "State Treasurer", "State Auditor")

# Get winners and their names
winners <- results |>
  filter(office_name %in% statewide_offices_list) |>
  filter(election_date >= 2002 & election_date <= 2022) |>
  filter(election_date != 2016) |>
  filter(is_winner == TRUE) |>
  select(election_date, office_name, candidate_name, candidate_party_name) |>
  mutate(
    last_name = gsub(".* ", "", candidate_name),  # Get everything after last space
    last_name = gsub(" and .*", "", last_name)  # Handle ticket names
  )

# Calculate margins using existing pct column from results
heatmap_data <- results |>
  filter(office_name %in% statewide_offices_list) |>
  filter(election_date >= 2002 & election_date <= 2022) |>
  filter(election_date != 2016) |>
  filter(candidate_party_name %in% c("Democratic", "Republican")) |>
  group_by(election_date, office_name, candidate_party_name) |>
  summarise(pct = first(pct), .groups = "drop") |>
  pivot_wider(names_from = candidate_party_name, values_from = pct) |>
  mutate(
    Democratic = ifelse(is.na(Democratic), 0, Democratic),
    Republican = ifelse(is.na(Republican), 0, Republican),
    margin = Democratic - Republican
  ) |>
  left_join(winners, by = c("election_date", "office_name")) |>
  select(election_date, office_name, margin, last_name) |>
  mutate(
    office_name = factor(office_name, levels = rev(statewide_offices_list))
  )

# Create the heatmap
ggplot(heatmap_data, aes(x = as.factor(election_date), y = office_name, fill = margin)) +
  geom_tile(color = "white", linewidth = 1) +
  geom_text(aes(label = paste0(last_name, "\n", round(abs(margin), 1))), 
            color = "black", size = 3.5) +
  scale_fill_gradient2(
    low = "#d62728", 
    mid = "white", 
    high = "#1f77b4",
    midpoint = 0,
    limits = c(-40, 40),
    guide = "none"
  ) +
  labs(
    title = "Statewide Office Control in New Mexico (2002-2022)",
    x = "Election Year",
    y = "Office"
  ) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    plot.title = element_text(size = 16, face = "plain")
  )

# Filter to presidential elections in selected counties
selected_counties <- c("Bernalillo", "Torrance", "Sandoval", "Valencia", "Santa Fe", "Luna", "McKinley", "Doña Ana")

# Get statewide margins
statewide_prez <- results |>
  filter(office_name == "President of the United States") |>
  filter(candidate_party_name %in% c("Democratic", "Republican")) |>
  group_by(election_date, candidate_party_name) |>
  summarise(pct = first(pct), .groups = "drop") |>
  pivot_wider(names_from = candidate_party_name, values_from = pct) |>
  mutate(
    Democratic = ifelse(is.na(Democratic), 0, Democratic),
    Republican = ifelse(is.na(Republican), 0, Republican),
    margin = Democratic - Republican,
    county_name = "Statewide"
  ) |>
  select(election_date, county_name, margin)

# Get winners by county
prez_winners <- county_results |>
  filter(office_name == "President of the United States") |>
  filter(county_name %in% selected_counties) |>
  filter(is_winner == TRUE) |>
  select(election_date, county_name, candidate_name, candidate_party_name) |>
  mutate(
    last_name = gsub(".* ", "", candidate_name),  # Get everything after last space
    last_name = gsub(" and .*", "", last_name)  # Handle ticket names
  )

# Get statewide winner
statewide_winners <- results |>
  filter(office_name == "President of the United States") |>
  filter(is_winner == TRUE) |>
  group_by(election_date) |>
  slice(1) |>
  select(election_date, candidate_name) |>
  mutate(
    last_name = gsub(".* ", "", candidate_name),  # Get everything after last space
    last_name = gsub(" and .*", "", last_name),  # Handle ticket names
    county_name = "Statewide"
  ) |>
  select(election_date, county_name, last_name)

prez_county <- county_results |>
  filter(office_name == "President of the United States") |>
  filter(county_name %in% selected_counties) |>
  filter(candidate_party_name %in% c("Democratic", "Republican")) |>
  group_by(election_date, county_name, candidate_party_name) |>
  summarise(pct = first(pct), .groups = "drop") |>
  pivot_wider(names_from = candidate_party_name, values_from = pct) |>
  mutate(
    Democratic = ifelse(is.na(Democratic), 0, Democratic),
    Republican = ifelse(is.na(Republican), 0, Republican),
    margin = Democratic - Republican
  ) |>
  left_join(prez_winners, by = c("election_date", "county_name")) |>
  select(election_date, county_name, margin, last_name) |>
  bind_rows(
    statewide_prez |>
      left_join(statewide_winners, by = c("election_date", "county_name"))
  )

# Order counties (Statewide at top)
prez_county$county_name <- factor(prez_county$county_name, 
                                   levels = c("Statewide", "Valencia", "Torrance", "Sandoval", "Bernalillo", "Santa Fe", "Luna", "McKinley", "Doña Ana"))

# Calculate range from data
margin_range <- range(prez_county$margin, na.rm = TRUE)
margin_max <- max(abs(margin_range))

ggplot(prez_county, aes(x = as.factor(election_date), y = county_name, fill = margin)) +
  geom_tile(color = "white", linewidth = 1) +
  geom_text(aes(label = paste0(last_name, "\n", round(abs(margin), 1))), 
            color = "black", size = 3.5) +
  scale_fill_gradient2(
    low = "#d62728", 
    mid = "white", 
    high = "#1f77b4",
    midpoint = 0,
    limits = c(-margin_max, margin_max),
    guide = "none"
  ) +
  labs(
    title = "Presidential Election Results: Selected Counties",
    x = "Election Year",
    y = "County"
  ) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    plot.title = element_text(size = 16, face = "plain")
  )

# Calculate State House breakdown by year
house_data <- results |>
  filter(office_name == "State Representative") |>
  filter(is_winner == TRUE)

house_composition <- house_data |>
  group_by(election_date, candidate_party_name) |>
  summarise(seats = n(), .groups = "drop") |>
  complete(election_date, candidate_party_name, fill = list(seats = 0)) |>
  filter(candidate_party_name != "Other") |>
  group_by(election_date) |>
  summarise(
    dem_seats = sum(seats[candidate_party_name == "Democratic"]),
    rep_seats = sum(seats[candidate_party_name == "Republican"]),
    label = paste0(dem_seats, "/", rep_seats),
    .groups = "drop"
  )

ggplot(house_composition, aes(x = as.factor(election_date))) +
  geom_col(aes(y = dem_seats), fill = "#1f77b4", alpha = 0.8) +
  geom_col(aes(y = -rep_seats), fill = "#d62728", alpha = 0.8) +
  geom_text(aes(y = pmax(dem_seats, abs(-rep_seats)) + 2, label = label), size = 4) +
  geom_hline(yintercept = 0, color = "black", linewidth = 0.5) +
  scale_y_continuous(
    breaks = seq(-40, 50, by = 10),
    labels = abs(seq(-40, 50, by = 10))
  ) +
  labs(
    title = "State House Composition (2000-2024)",
    x = "Election Year",
    y = "Number of Seats"
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(size = 16, face = "plain")
  )

# Calculate State Senate breakdown by year
senate_data <- results |>
  filter(office_name == "State Senate") |>
  filter(is_winner == TRUE)

senate_composition <- senate_data |>
  group_by(election_date, candidate_party_name) |>
  summarise(seats = n(), .groups = "drop") |>
  complete(election_date, candidate_party_name, fill = list(seats = 0)) |>
  filter(candidate_party_name != "Other") |>
  group_by(election_date) |>
  summarise(
    dem_seats = sum(seats[candidate_party_name == "Democratic"]),
    rep_seats = sum(seats[candidate_party_name == "Republican"]),
    label = paste0(dem_seats, "/", rep_seats),
    .groups = "drop"
  )

ggplot(senate_composition, aes(x = as.factor(election_date))) +
  geom_col(aes(y = dem_seats), fill = "#1f77b4", alpha = 0.8) +
  geom_col(aes(y = -rep_seats), fill = "#d62728", alpha = 0.8) +
  geom_text(aes(y = pmax(dem_seats, abs(-rep_seats)) + 2, label = label), size = 4) +
  geom_hline(yintercept = 0, color = "black", linewidth = 0.5) +
  scale_y_continuous(
    breaks = seq(-25, 30, by = 5),
    labels = abs(seq(-25, 30, by = 5))
  ) +
  labs(
    title = "State Senate Composition (2000-2024)",
    x = "Election Year",
    y = "Number of Seats"
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(size = 16, face = "plain")
  )

# Senate: 2 seats (Class 1: 2000,2006,2012,2018,2024 | Class 2: 2002,2008,2014,2020)
senate <- results |>
  filter(office_name == "United States Senator", is_winner == TRUE) |>
  mutate(
    yr = as.numeric(election_date),
    name = gsub(".*,? ([A-Z][a-z]+)$", "\\1", candidate_name),
    party = ifelse(candidate_party_name == "Democratic", "Democratic", "Republican")
  )

build_seat <- function(years, class_name) {
  map_dfr(years, ~{
    s <- senate |> filter(yr == .x) |> slice(1)
    if(nrow(s) > 0) {
      next_yr <- c(years[years > .x], 2025)[1] - 1
      data.frame(name = s$name, party = s$party, class = class_name,
                 start_year = .x, end_year = next_yr, stringsAsFactors = FALSE)
    }
  }) |> mutate(label = paste0(name, " (", class, ")"))
}

senate_timeline <- bind_rows(
  build_seat(c(2000, 2006, 2012, 2018, 2024), "Class 1"),
  build_seat(c(2002, 2008, 2014, 2020), "Class 2")
)

create_timeline(senate_timeline, "U.S. Senate",
                "Senator tenure by class from 2000-2024", group_col = "class")

# House: 3 districts with 2-year terms (each election is a separate term)
house_timeline <- results |>
  filter(office_name == "United States Representative", is_winner == TRUE) |>
  mutate(
    yr = as.numeric(election_date),
    name = gsub(".*,? ([A-Z][a-z]+)$", "\\1", candidate_name),
    party = ifelse(candidate_party_name == "Democratic", "Democratic", "Republican"),
    dist = gsub(".*District (\\d+).*", "\\1", district_name)
  ) |>
  arrange(dist, yr) |>
  group_by(dist) |>
  mutate(
    next_election = lead(yr, default = 2025),
    end_year = pmin(next_election - 1, 2024)
  ) |>
  ungroup() |>
  mutate(label = paste0(name, " (D", dist, ")")) |>
  select(label, party, dist, start_year = yr, end_year)

create_timeline(house_timeline, "U.S. House New Mexico Delegation",
                "Representative tenure by district from 2000-2024", group_col = "dist")

# Calculate delegation breakdown by year
federal_data <- results |>
  filter(office_name %in% federal_offices) |>
  filter(is_winner == TRUE)

delegation <- federal_data |>
  filter(office_name == "United States Representative") |>
  group_by(election_date, candidate_party_name) |>
  summarise(seats = n(), .groups = "drop") |>
  complete(election_date, candidate_party_name, fill = list(seats = 0)) |>
  filter(candidate_party_name != "Other") |>
  mutate(
    delegation_label = paste0(seats, ifelse(candidate_party_name == "Democratic", "D", "R"))
  ) |>
  group_by(election_date) |>
  summarise(
    dem_seats = sum(seats[candidate_party_name == "Democratic"]),
    rep_seats = sum(seats[candidate_party_name == "Republican"]),
    label = paste0(dem_seats, "/", rep_seats),
    .groups = "drop"
  )

ggplot(delegation, aes(x = as.factor(election_date))) +
  geom_col(aes(y = dem_seats), fill = "#1f77b4", alpha = 0.8) +
  geom_col(aes(y = -rep_seats), fill = "#d62728", alpha = 0.8) +
  geom_text(aes(y = ifelse(dem_seats >= rep_seats, dem_seats + 0.15, -rep_seats - 0.15), label = label), size = 4) +
  geom_hline(yintercept = 0, color = "black", linewidth = 0.5) +
  scale_y_continuous(
    breaks = seq(-3, 3, by = 1),
    labels = abs(seq(-3, 3, by = 1))
  ) +
  labs(
    title = "U.S. House Delegation Composition (2000-2024)",
    subtitle = "Positive = Democratic seats, Negative = Republican seats",
    x = "Election Year",
    y = "Number of Seats"
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(size = 16, face = "plain")
  )

# Filter for 2024 and standardize column names
precinct_2024 <- precinct_data |>
  filter(election_date == 2024)

# Standardize column names
if("COUNTY_NAM" %in% names(precinct_2024)) {
  precinct_2024$county_name <- precinct_2024$COUNTY_NAM
}
if("VTD_NUM" %in% names(precinct_2024)) {
  precinct_2024$precinct_name <- as.character(precinct_2024$VTD_NUM)
}

precinct_2024 <- precinct_2024 |>
  mutate(cd = gsub(".*District (\\d+).*", "\\1", district_name))

# Get Harris (President) results
harris <- precinct_2024 |>
  filter(office_name == "President of the United States") |>
  mutate(pct_harris = (dem / votes_cast) * 100) |>
  select(county_name, precinct_name, pct_harris)

# Get House results by CD
house_2024 <- precinct_2024 |>
  filter(office_name == "United States Representative") |>
  group_by(county_name, precinct_name, cd) |>
  summarise(
    votes_cast = sum(votes_cast, na.rm = TRUE),
    pct_house_dem = (sum(dem, na.rm = TRUE) / votes_cast) * 100,
    .groups = "drop"
  )

# Merge and plot
comparison <- harris |>
  inner_join(house_2024, by = c("county_name", "precinct_name"))

ggplot(comparison, aes(x = pct_harris, y = pct_house_dem, color = cd)) +
  geom_point(alpha = 0.3, size = 0.8) +
  geom_abline(slope = 1, intercept = 0, linetype = "dashed", color = "black", linewidth = 1.2) +
  geom_smooth(method = "lm", se = FALSE, linewidth = 1.5) +
  scale_color_manual(values = c("#1f77b4", "#ff7f0e", "#2ca02c")) +
  xlim(25, 75) +
  ylim(25, 75) +
  labs(
    x = "Harris %", 
    y = "House Democratic Candidate %",
    title = "Democratic Performance: House vs Harris (2024)",
    subtitle = "Points above diagonal = House outperformed Harris; below = Harris outperformed House (x-axis zoomed 25-75% for visualization purposes)",
    color = "CD"
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(size = 16, face = "plain")
  )

# Summary table by CD
comparison |>
  group_by(cd) |>
  summarise(
    mean_harris = mean(pct_harris, na.rm = TRUE),
    mean_house = mean(pct_house_dem, na.rm = TRUE),
    mean_underperform = mean(pct_harris - pct_house_dem, na.rm = TRUE),
    .groups = "drop"
  ) |>
  arrange(mean_underperform) |>
  mutate(
    mean_harris = round(mean_harris, 1),
    mean_house = round(mean_house, 1),
    mean_underperform = round(mean_underperform, 1)
  ) |>
  knitr::kable(
    col.names = c("CD", "Mean Harris %", "Mean House Democratic Candidate %", "Mean Underperformance"),
    caption = "Harris Underperformance by Congressional District (2024)",
    align = "cccc",
    digits = 1,
    format = "html"
  )