Voter Registration Data Infographic

Voter Registration Data Infographic

Why voter data?

Since childhood, I’ve been interested in politics– a fault I attribute to my dad. We were your typical liberal Prius-driving NPR listeners and CNN-blaring-in-the-background household. Upon moving to Santa Barbara in 2017 for my undergraduate degree, I’ve volunteered or worked in every election, traveling up and down the county to knock on doors for local candidates with pro-environment and pro-housing values.

Some campaigns over the years

Most visualizations of voter data happen at the national or state level, rather than within counties. Furthermore, election data from the 2024 general election has not been fully certified yet for most counties in California. I aim to provide a quick overview of the political composition of our county using voter registration and census data for people interested in local politics. A few questions I’d like to answer include:

• How are Democratic and Republican voters spatially distributed throughout the county?
• What is the political composition of the county?
• Is there a partisan difference between voter ages?
• How do voter registrations change over time? Do they increase in frequency closer to elections?

About the data

The Santa Barbara County voter registration data used for this project was requested from SB County Elections, and is dated for May 9, 2025. It contains voter registration information (voter name, address, contact information, voting precinct, and political party affiliation) for all voters in the county.

For census data, I use the tidycensus and tigris packages, with 2020 and 2023 geometries and population surveys, and tidygeocoder for addresses.

Visualizations

1. How are voters distributed across the county? I wanted to plot the difference in percentage of Democrats compared to Republicans across the county. I chose a chloropleth map to visualize this, because it is a good way to display spatial relationships. In Santa Barbara, urban centers in South County tend to lean heavily Democratic, while more rural areas are Republican. One challenge with displaying partisanship on a chloropleth map, however, is the visual overrepresentation of sparsely populated but large geographic areas. It appears as though a large portion of the county is red, when in reality very few people live in those areas, and the county as a whole leans solidly Democratic. Census blocks tend to be smaller than tracts, and areas with no population were excluded, therefore lessening the visual effect of area.

2. What is the partisan breakdown of the county? I’ve included a waffle plot displaying the proportion of voters by party to visualize the percentage of voters belonging to each political party.

3. How old are Democrats vs. Republicans? For this visualization, I chose stacked bar charts to visualize the age breakdowns within each party and across the county. On average, Democrats are slightly younger than Republicans.

4. When do people register to vote? Voters generally register to vote closer to elections, and we can see that there is a noticeable spike in the line graph of number of registrations each general election year (labeled on the x-axis). When we look at the weekly and daily registration trends leading up to the 2024 general election, we see a spike in registrations around the registration deadline and on election day.

Code for Visualizations

# -------------- Setup --------------
# Load packages
library(tidyverse)
library(janitor)
library(here)
library(tidycensus)
library(ggmap)
library(sf)
library(tigris)
library(showtext)
library(sysfonts)
library(tidygeocoder)
library(tmap)
library(tmaptools)
library(waffle)
library(patchwork)
library(leaflet)

# Set census API key
census_api_key(Sys.getenv("CENSUS_API_KEY"))

# -------------- Customize Theme --------------
# Customize font and theme
font_add_google("EB Garamond", "EBGaramond")
showtext_auto()
theme_set(theme_bw())

# Custom colors
party_colors <- c(
    "DEM" = "#126ca8",
    "REP" = "#c83236",
    "NPP" = "#c9b1d4",
    "Other" = "gray80"
)

# Create custom function
theme_voter_viz <- function(base_size = 20, base_family = "EBGaramond") {
    theme_classic() %+replace%
        theme(
            panel.grid.major.y = element_blank(),
            text = element_text(size = base_size, family = base_family),
            axis.text = element_text(family = base_family),
            plot.title = element_text(face = "bold", family = base_family),
            plot.subtitle = element_text(family = base_family),
            plot.caption = element_text(family = base_family),
            legend.text = element_text(family = base_family)
        )
}

# -------------- Load data --------------
full_file <- read_delim(
    here::here("posts/2025-03-17-voter-data-viz/data/Countywide_Voter Registration 030325.TXT"),
    delim = "\t"
) %>%
    clean_names()

Visualization 1: Spatial Analysis of Voter Distribution

# -------------- Geocoding & Retrieval of Census Data --------------
# Prepare addresses for geocoding
# address_full <- full_file %>%
#     mutate(street_address_test = paste(address_number, street_name, sep = " "))
#
# # Geocoding function
# geocode_addresses <- function() {
#     census_list <- list()
#     for (j in 1:25) {
#         start_index <- ((j - 1) * 10000) + 1
#         end_index <- j * 10000
#
#         census_list[[j]] <- address_full[start_index:end_index, ] %>%
#             tidygeocoder::geocode(
#                 street = street_address_test,
#                 city = city,
#                 state = state,
#                 method = "census"
#             )
#     }
#
#     final_census_df <- bind_rows(census_list) %>%
#         filter(party_code %in% c("DEM", "REP", "NPP")) %>%
#         drop_na(lat)
#
#     # Save for future use
#     saveRDS(final_census_df, "geocoded_addresses.rds")
#
#     return(final_census_df)
# }
#
# final_census_df <- geocode_addresses()

final_census_df <- readRDS(here::here("posts/2025-03-17-voter-data-viz/data/geocoded_addresses.rds"))

# File paths for saved data
sb_pop_path <- here::here("posts/2025-03-17-voter-data-viz/data/sb_pop_2023.rds")
sb_blocks_path <- here::here("posts/2025-03-17-voter-data-viz/data/sb_blocks_2020.rds")

# Check if files exist, otherwise create and save them
if (!file.exists(sb_pop_path)) {
    # Get Santa Barbara census tract population data
    sb_pop_2023 <- get_acs(
        geography = "tract",
        variables = "B01003_001", # Total population variable
        state = "CA",
        county = "Santa Barbara",
        year = 2023,
        geometry = TRUE
    ) %>%
        st_transform(crs = 4326) %>%
        filter(GEOID != 06083990000) # Channel Islands tract

    # Save to RDS file
    saveRDS(sb_pop_2023, sb_pop_path)
} else {
    # Load from saved file
    sb_pop_2023 <- readRDS(sb_pop_path)
}

if (!file.exists(sb_blocks_path)) {
    # Get census block geography for Santa Barbara County
    sb_blocks_2020 <- get_decennial(
        geography = "block",
        variables = "P1_001N", # Total population variable
        state = "CA",
        county = "Santa Barbara",
        year = 2020,
        geometry = TRUE
    ) %>%
        st_transform(crs = 4326) %>%
        # filter out ocean tracts
        filter(!grepl("^06083990000", GEOID))

    # Save to RDS file
    saveRDS(sb_blocks_2020, sb_blocks_path)
} else {
    # Load from saved file
    sb_blocks_2020 <- readRDS(sb_blocks_path)
}

# Tract-level map code here
# -------------- Spatial Visualization: Census Tracts --------------
# Convert geocoded addresses to sf object
final_census_sf <- st_as_sf(
    final_census_df,
    coords = c("long", "lat"),
    crs = 4326
)

# Join voters to tracts and calculate party percentages
party_by_tract <- st_join(final_census_sf, sb_pop_2023) %>%
    st_drop_geometry() %>%
    count(GEOID, party_code) %>%
    group_by(GEOID) %>%
    mutate(tract_total = sum(n), percentage = n / tract_total * 100) %>%
    ungroup()

# Two-party comparison calculation
two_party_comparison <- party_by_tract %>%
    filter(party_code %in% c("DEM", "REP")) %>%
    group_by(GEOID) %>%
    mutate(two_party_total = sum(n)) %>%
    mutate(two_party_pct = n / two_party_total * 100) %>%
    select(GEOID, party_code, two_party_pct) %>%
    pivot_wider(names_from = party_code, values_from = two_party_pct) %>%
    mutate(diff = DEM - REP)

# Join and create map
tract_diff_map <- sb_pop_2023 %>%
    select(GEOID, geometry) %>%
    left_join(two_party_comparison, by = "GEOID") %>%
    filter(!is.na(diff))

# Create custom breaks and labels for the scale
breaks <- seq(-100, 100, by = 20)
labels <- sapply(breaks, function(x) {
    if (x > 0) {
        paste0("+", x)
    } else {
        x
    }
})

# Create the divergent map
ggplot() +
    geom_sf(data = tract_diff_map, aes(fill = diff)) +
    scale_fill_gradient2(
        low = party_colors["REP"],
        mid = "#FFFFFF",
        high = party_colors["DEM"],
        midpoint = 0,
        name = "Dem-Rep Difference (%)",
        breaks = breaks,
        labels = labels,
        guide = guide_colorbar(
            direction = "horizontal",
            barwidth = 15,
            barheight = 0.5,
            title.position = "top",
            title.hjust = 0.5
        )
    ) +
    labs(
        title = "Democratic vs. Republican Voter Distribution by Census Tract\n",
        subtitle = "Percentage difference between parties (excluding other affiliations)"
    ) +
    theme_void() +
    theme_voter_viz() +
    theme(
        legend.position = "bottom",
        legend.title = element_text(size = 10),
        legend.text = element_text(size = 8),
        legend.margin = margin(t = 10, b = 10),
        axis.title = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank(),
        axis.line = element_blank()
    )

ggsave(
    "tract_diff_map.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 13
)

# Get bounding box
tract_map_bbox <- st_bbox(tract_diff_map)

# Block-level map code here
# -------------- Spatial Visualization: Block Groups --------------
# Join voters to blocks and calculate party percentages
party_by_block <- st_join(final_census_sf, sb_blocks_2020) %>%
    st_drop_geometry() %>%
    count(GEOID, party_code) %>%
    group_by(GEOID) %>%
    mutate(block_total = sum(n), percentage = n / block_total * 100) %>%
    ungroup()

# Two-party comparison calculation
two_party_comparison <- party_by_block %>%
    filter(party_code %in% c("DEM", "REP")) %>%
    group_by(GEOID) %>%
    mutate(
        two_party_total = sum(n),
        two_party_pct = n / two_party_total * 100
    ) %>%
    select(GEOID, party_code, two_party_pct) %>%
    pivot_wider(names_from = party_code, values_from = two_party_pct) %>%
    mutate(diff = DEM - REP)

# Join and create map
block_diff_map <- sb_blocks_2020 %>%
    select(GEOID, geometry) %>%
    left_join(two_party_comparison, by = "GEOID")

# Create custom breaks and labels for the scale
breaks <- seq(-100, 100, by = 20)
labels <- sapply(breaks, function(x) {
    if (x > 0) {
        paste0("+", x)
    } else {
        x
    }
})

# Create divergent plot
ggplot() +
    geom_sf(data = block_diff_map, aes(fill = diff), color = NA, size = 0.5) +
    geom_sf(data = filter(block_diff_map, is.na(diff)), fill = "gray80", color = "gray30") +
    scale_fill_gradient2(
        low = party_colors["REP"],
        mid = "#FFFFFF",
        high = party_colors["DEM"],
        midpoint = 0,
        name = "Dem-Rep Difference (%)",
        breaks = breaks,
        labels = labels,
        guide = guide_colorbar(
            direction = "horizontal",
            barwidth = 15,
            barheight = 0.5,
            title.position = "top",
            title.hjust = 0.5
        ),
        na.value = "gray80"
    ) +
    coord_sf(
        xlim = c(tract_map_bbox["xmin"], tract_map_bbox["xmax"]),
        ylim = c(tract_map_bbox["ymin"], tract_map_bbox["ymax"])
    ) +
    labs(
        title = "Democratic vs. Republican Voter Distribution by Census Block\n",
        subtitle = "Percentage difference between parties (excluding other affiliations)"
    ) +
    theme_void() +
    theme_voter_viz() +
    theme(
        legend.position = "bottom",
        legend.title = element_text(size = 10),
        legend.text = element_text(size = 8),
        legend.margin = margin(t = 10, b = 10),
        axis.text = element_blank(),
        axis.title = element_blank(),
        axis.ticks = element_blank(),
        axis.line = element_blank()
    )

ggsave(
    "block_diff_map.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 13
)

# Create a bounding box for a zoomed-in look at south county
south_county_bbox <- tigris::places(state = "CA", cb = TRUE) %>%
    filter(
        NAME %in%
            c(
                "Santa Barbara",
                "Goleta",
                "Carpinteria",
                "Isla Vista",
                "Summerland",
                "Montecito",
                "Eastern Goleta Valley",
                "Mission Canyon",
                "University of California-Santa Barbara",
                "Toro Canyon"
            )
    ) %>%
    st_union() %>%
    st_buffer(dist = 0.02)

ggplot() +
    geom_sf(data = block_diff_map, aes(fill = diff), color = NA, size = 0.5) +
    geom_sf(data = filter(block_diff_map, is.na(diff)), fill = "gray80", color = "gray30") +
    scale_fill_gradient2(
        low = party_colors["REP"],
        mid = "#FFFFFF",
        high = party_colors["DEM"],
        midpoint = 0,
        name = "Dem-Rep Difference (%)",
        breaks = breaks,
        labels = labels,
        guide = guide_colorbar(
            direction = "horizontal",
            barwidth = 15,
            barheight = 0.5,
            title.position = "top",
            title.hjust = 0.5
        ),
        na.value = "gray80"
    ) +
    coord_sf(
        xlim = st_bbox(south_county_bbox)[c(1, 3)],
        ylim = st_bbox(south_county_bbox)[c(2, 4)]
    ) +
    theme_void() +
    theme_voter_viz() +
    theme(
        legend.position = "bottom",
        legend.title = element_text(size = 10),
        legend.text = element_text(size = 8),
        legend.margin = margin(t = 10, b = 10),
        axis.text = element_blank(),
        axis.title = element_blank(),
        axis.ticks = element_blank(),
        axis.line = element_blank()
    )

ggsave(
    "zoom_block_diff_map.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 10
)

# -------------- Spatial Visualization: Interactive Census Block Groups --------------
# Create a diverging color palette for leaflet map
pal <- colorNumeric(
    palette = colorRampPalette(c(
        party_colors["REP"],
        "#FFFFFF",
        party_colors["DEM"]
    ))(100),
    domain = c(-100, 100), # Set fixed domain from -100 to +100
    na.color = "transparent"
)

# Create leaflet map with OpenStreetMap tiles
leaflet(block_diff_map) %>%
    addTiles() %>%
    addPolygons(
        fillColor = ~ pal(diff),
        weight = 1,
        opacity = 1,
        color = "white",
        dashArray = "3",
        fillOpacity = 0.7,
        label = ~ paste0("Dem-Rep Difference: ", round(diff, 1), "%")
    ) %>%
    addLegend(
        pal = pal,
        values = ~diff,
        opacity = 0.7,
        title = "Dem-Rep Difference (%)",
        position = "bottomright"
    )

Note that UCSB is missing from the data. This could be because the census designates some universities as “group quarters” and counts them differently compared to other residences. Furthermore, students are likely underrepresented in the 2020 census due to the COVID-19 pandemic and remote learning, as well as general confusion about current address vs. permanent address.

Visualization 2: Party Proportions

# --------------- Visualization 2: Party Proportions Waffle Chart --------------
# Create waffle chart of party distribution
waffle_counts <- full_file %>%
  mutate(
    party_code = if_else(
      party_code %in% c("DEM", "REP", "NPP", "AI"),
      party_code,
      "Other"
    )
  ) %>%
  count(party_code) %>%
  mutate(
    percent = n / sum(n) * 100,
    party_name = case_when(
      party_code == "DEM" ~ "Democrat",
      party_code == "REP" ~ "Republican",
      party_code == "NPP" ~ "No Party Preference",
      party_code == "AI" ~ "American Independent",
      TRUE ~ "Other"
    ),
    party_label = paste0(party_name, " (", round(percent, 1), "%)"),
    party_code = factor(
      party_code,
      levels = c("DEM", "REP", "NPP", "AI", "Other")
    )
  ) %>%
  arrange(party_code)

# Create named vector for legend labels
party_labels <- setNames(
    waffle_counts$party_label,
    waffle_counts$party_code
)

# Update party colors to include AI
party_colors_updated <- party_colors
party_colors_updated["AI"] <- "gray40"

# Create waffle chart
ggplot(waffle_counts, aes(fill = party_code, values = n)) +
  geom_waffle(
    color = "white", 
    size = 2.5, 
    n_rows = 10,
    make_proportional = TRUE
  ) +
  scale_fill_manual(
    values = party_colors_updated,
    labels = party_labels
  ) +
  coord_fixed() +
  theme_voter_viz() +
  theme(
    axis.text = element_blank(),
    axis.title = element_blank(),
    axis.ticks = element_blank(),
    axis.line = element_blank(),
    legend.spacing.x = unit(1.0, "cm"),
    legend.spacing.y = unit(1.0, "cm"),
  ) +
  labs(
    title = "Party Distribution in Santa Barbara County",
    fill = "Party"
  )

ggsave(
  "party_waffle_chart.svg",
  path = here::here("posts/2025-03-17-voter-data-viz/media"),
  width = 16,
  height = 13
)

Visualization 3: Age Composition by Party

# -------------- Visualization 3: Age Composition by Party --------------
# Calculate age distribution by party
voter_ages <- full_file %>%
    mutate(
        age_group = cut(
            interval(dob, today()) %/% years(1),
            breaks = c(18, 25, 35, 50, 65, Inf),
            labels = c("18-25", "26-35", "36-50", "51-65", "65+"),
            right = FALSE
        ),
        party_code = factor(party_code, levels = c("DEM", "REP"))
    ) %>%
    filter(!is.na(party_code)) %>%
    count(party_code, age_group) %>%
    group_by(party_code) %>%
    mutate(pct = n / sum(n)) %>%
    ungroup()

# Fetch and process Santa Barbara County census age data
sb_age_data <- get_acs(
    geography = "county",
    table = "B01001",
    year = 2023,
    cache_table = TRUE
) %>%
    filter(NAME == "Santa Barbara County, California") %>%
    mutate(
        age_group = case_when(
            variable %in%
                c("B01001_003", "B01001_004", "B01001_027", "B01001_028") ~
                "18-25",
            variable %in%
                c("B01001_005", "B01001_006", "B01001_029", "B01001_030") ~
                "26-35",
            variable %in%
                c(
                    "B01001_007",
                    "B01001_008",
                    "B01001_009",
                    "B01001_031",
                    "B01001_032",
                    "B01001_033"
                ) ~
                "36-50",
            variable %in%
                c(
                    "B01001_010",
                    "B01001_011",
                    "B01001_012",
                    "B01001_013",
                    "B01001_034",
                    "B01001_035",
                    "B01001_036",
                    "B01001_037"
                ) ~
                "51-65",
            variable %in%
                c(
                    "B01001_014",
                    "B01001_015",
                    "B01001_016",
                    "B01001_017",
                    "B01001_018",
                    "B01001_019",
                    "B01001_038",
                    "B01001_039",
                    "B01001_040",
                    "B01001_041",
                    "B01001_042",
                    "B01001_043"
                ) ~
                "65+",
            TRUE ~ NA_character_
        )
    ) %>%
    filter(!is.na(age_group)) %>%
    group_by(age_group) %>%
    summarize(n = sum(estimate)) %>%
    mutate(
        pct = n / sum(n),
        party_code = "SB County"
    ) %>%
    mutate(
        age_group = factor(
            age_group,
            levels = c("18-25", "26-35", "36-50", "51-65", "65+")
        )
    )

# Combine data and create plot
bind_rows(voter_ages, sb_age_data) %>%
    mutate(
        party_code = factor(party_code, levels = c("DEM", "REP", "SB County")),
        label = paste0(age_group, " (", scales::percent(pct, accuracy = 1), ")")
    ) %>%
    ggplot(aes(x = party_code, y = pct, fill = age_group)) +
    geom_col(width = 0.6) +
    geom_text(
        aes(label = label),
        position = position_stack(vjust = 0.5),
        color = "black",
        fontface = "bold",
        size = 5.5,
        family = "EBGaramond"
    ) +
    scale_fill_brewer(palette = "Purples", direction = 1) +
    scale_y_continuous(
        labels = scales::percent,
        breaks = seq(0, 1, 0.2)
    ) +
    labs(
        title = "Age Distribution of Voters in Santa Barbara County by Party",
        subtitle = "Comparing party affiliations with overall county population\n",
        x = NULL,
        y = "Proportion",
        caption = "\nSources: Santa Barbara Voter File (2025), US Census ACS 2023"
    ) +
    theme_voter_viz() +
    theme(
        axis.text.y = element_blank(),
        axis.ticks = element_blank(),
        legend.position = "none",
        text = element_text(family = "EBGaramond")
    )

ggsave(
    "age_composition_plot.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 13
)

Visualization 4: Registration Trends

# -------------- Visualization 4: Yearly Registration Trends --------------
# Analyze registrations over time
yearly_registration <- full_file %>%
    mutate(year = format(registration_date, "%Y")) %>%
    filter(year >= "2000") %>%
    filter(year <= "2024") %>%
    count(year)

# Create time series plot
ggplot(yearly_registration, aes(x = year, y = n)) +
    geom_line(group = 1, color = "purple4") +
    geom_point(color = "purple4") +
    labs(
        title = "Voter Registrations in Santa Barbara County",
        subtitle = "Number of new registrations from 2000 - 2024, showing spikes in new registrations during general election years",
        x = "Year",
        y = "Number of New Registrations"
    ) +
    scale_x_discrete(
        breaks = seq(
            from = min(yearly_registration$year),
            to = max(yearly_registration$year),
            by = 4
        )
    ) +
    theme(
        panel.grid.major.y = element_blank(),
        text = element_text(size = 24, family = "EBGaramond"),
        axis.text = element_text(family = "EBGaramond"),
        plot.title = element_text(face = "bold", family = "EBGaramond"),
        plot.subtitle = element_text(family = "EBGaramond"),
        plot.caption = element_text(family = "EBGaramond"),
        legend.text = element_text(family = "EBGaramond")
    )

ggsave(
    "yearly_registration_plot.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 13
)

It seems obvious that more people register to vote in election years. But how long do they wait to register? Let’s look at the 2024 March 5th primary election and November 5th presidential election to take a closer look at when people register to vote.

# Weekly trends code here
# -------------- Visualization 4: Weekly Registration Trends --------------
# Analyze 2024 registrations by week and party
weekly_registrations <- full_file %>%
    mutate(
        reg_year = lubridate::year(registration_date),
        reg_month = lubridate::month(registration_date),
        reg_week = lubridate::week(registration_date),
        reg_day = lubridate::day(registration_date),
        week_start_date = lubridate::floor_date(registration_date, "week")
    ) %>%
    filter(reg_year >= 2024, reg_year <= 2025) %>%
    count(reg_week, week_start_date, party_code) %>%
    group_by(reg_week, party_code) %>%
    slice(1) %>%
    ungroup() %>%
    filter(party_code %in% c("DEM", "REP", "NPP"))

# Create election day reference dates
election_dates <- tibble(
    date = as.Date(c("2024-03-05", "2024-11-05")),
    label = c("Primary", "General")
)

# Create stacked area chart
ggplot(
    weekly_registrations,
    aes(x = week_start_date, y = n, fill = party_code)
) +
    geom_area(alpha = 0.8) +
    geom_vline(
        data = election_dates,
        aes(xintercept = date),
        linetype = "dashed",
        color = "red"
    ) +
    labs(
        title = "Voter Registrations in Santa Barbara County (2024)",
        subtitle = "Composition of new registrations by week and party",
        y = "Number of New Registrations",
        fill = "Party"
    ) +
    scale_x_date(
        date_breaks = "1 month",
        date_labels = "%b" # Format: "Jan 01"
    ) +
    scale_fill_manual(values = party_colors) +
    theme_voter_viz() +
    theme(
        axis.title.x = element_blank(),
        legend.position = "bottom"
    )

ggsave(
    "weekly_registration_plot.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media"),
    width = 16,
    height = 13
)

# Daily trends code here
# -------------- Visualization 4: Daily Registration Trends Before Election --------------
# Analyze daily registrations for Oct 5 - Nov 5 period
daily_registrations <- full_file %>%
    filter(
        registration_date >= as.Date("2024-10-05"),
        registration_date <= as.Date("2024-11-05"),
        party_code %in% c("DEM", "REP", "NPP")
    ) %>%
    count(registration_date, party_code)

# Count total number of registrations on key dates
daily_registrations_lookup <- full_file %>%
  filter(
    registration_date %in% c("2024-03-05", "2024-10-21", "2024-11-05")
  ) %>%
  count(registration_date)

# Create reference dates
reference_dates <- tibble(
    date = c(as.Date("2024-11-05"), as.Date("2024-10-21")),
    label = c("Election Day", "Registration Deadline")
)

# Create a stacked area chart with daily data
ggplot(
    daily_registrations,
    aes(x = registration_date, y = n, fill = party_code)
) +
    geom_area(alpha = 0.8) +
    # Add vertical lines for both reference dates
    geom_vline(
        data = reference_dates,
        aes(xintercept = date),
        linetype = "dashed",
        color = "red",
    ) +
    labs(
        title = "Voter Registrations in Santa Barbara County (Oct 5 - Nov 5, 2024)",
        subtitle = "Daily registrations in the month leading up to the election",
        y = "Number of New Registrations",
        fill = "Party"
    ) +
    scale_x_date(
        breaks = seq.Date(
            from = as.Date("2024-10-05"),
            to = as.Date("2024-11-05"),
            by = "1 day"
        ),
        labels = NULL,
        date_minor_breaks = "1 week",
        limits = c(as.Date("2024-10-05"), as.Date("2024-11-05"))
    ) +
    scale_y_continuous(
        breaks = seq(0, 6200, by = 1000)
    ) +
    scale_fill_manual(values = party_colors) +
    theme_voter_viz() +
    theme(
        axis.title.x = element_blank(),
        legend.position = "bottom"
    )

ggsave(
    "daily_registration_plot.svg",
    path = here::here("posts/2025-03-17-voter-data-viz/media/"),
    width = 16,
    height = 13
)

The most amount of registrations occur on Election Day. California offers same-day voter registration for voters who need to vote provisionally (did not register in advance, moved addresses, voting at different polling location, etc.) This is not true for all states, however, especially those actively restricting voting rights.

Discussion - design elements & aesthetic choices

Select tab to see discussion on the topic.

Graphic Form + General Design

Creating my spatial visualization took the most amount of time, and I chose to include an interactive block group level map, because partisan differences can be obscured by geography and population. I wanted my quick-glance infographic to present the data in a formal, news-oriented format, as if it were something you would see on the New York Times or 538.

Text, Themes, Colors, & Typography

My aesthetic choices and theming were heavily influenced by New York Times and other legacy news media data visualizations. I chose red and blue since they’re the traditional colors that represent political parties, in addition to purple and grey as neutral colors. The text for my individual plots provides an overview for each plot as a standalone, explaining relationships and key takeaways.

Primary Message + Contribution

I aim to provide sub-county level visualizations of partisanship for SB County. While the south portion of the county is heavily Democratic, rural areas and north county leans more Republican. Registered Republicans are slightly older than Democrats, but also wait until the last minute to register to vote.

These visualizations provide a quick way to glean context from local outcomes of elections. They can be used as tool to compare vote shares for candiates or issues against, or to look at partisan counts for redistricting efforts.

Accessibility

While my spatial visualization is not colorblind friendly, I have included alt-text to improve accessibility for those using screen readers. In the future, I would like to create a colorblind-friendly version of this infographic.

DEI

Because my voter data only includes age variables, I chose to focus on partisan differences between Democrats and Republicans in this analysis rather than incorporating other demographic variables. However, adding census-tract level analysis using additional variables like income, race/ethnicity, or housing burden from the American Community Survey could be interesting.

Furthermore, a partisan analysis lends itself to the concepts of diversity, equity, and inclusion. The modern Republican party under Trump has attempted to kill DEI initiatives and mentions of it from government. It’s worth visualizing voter sentiments locally, because while Trump won the 2024 election, SB county remains deeply blue, especially south county.