Georgia elections

What it covers

The Georgia scraper returns election results from the Georgia Secretary of State at results.sos.ga.gov. It covers all elections available on the site — primaries, runoffs, generals, and special elections — from 2000 to the present.

Optionally, it can also return a vote-method breakdown (Advance in Person, Election Day, Absentee by Mail, Provisional) for each contest.

Note: The Georgia SOS site is a JavaScript-rendered Angular SPA. All scraping uses a headless Chromium browser via Playwright. County scraping is parallelised across up to max_workers browsers (default 4). Georgia has 159 counties; a full state + county scrape for one year typically takes 20–60 minutes.

Arguments

Argument	Default	Description
`state`	—	`"GA"`, `"georgia"`, or `"Georgia"`
`year_from`	`NULL`	Start year, inclusive; `NULL` = no lower bound
`year_to`	`NULL`	End year, inclusive; `NULL` = no upper bound
`level`	`"all"`	`"all"` — statewide + county; `"state"` — statewide only (much faster); `"county"` — county-level only
`max_workers`	`4L`	Parallel Chromium browsers for county scraping
`include_vote_methods`	`FALSE`	If `TRUE`, also return vote-method breakdown tables

Examples

Statewide + county results for a single year

res <- scrape_elections(state = "GA", year_from = 2024, year_to = 2024)

# Statewide candidate totals
res$state %>%
  filter(office == "President of the United States") %>%
  select(election_name, candidate, party, votes, pct, is_incumbent)

# County-level breakdown
res$county %>%
  filter(office == "President of the United States") %>%
  select(county, candidate, party, votes, pct) %>%
  arrange(county, desc(votes))

Statewide totals only (faster — skips county scraping)

state_df <- scrape_elections(
  state     = "GA",
  year_from = 2024,
  year_to   = 2024,
  level     = "state"
)

state_df %>%
  select(election_name, office, candidate, party, votes, pct, is_incumbent)

With vote-method breakdown

res <- scrape_elections(
  state                = "GA",
  year_from            = 2024,
  year_to              = 2024,
  include_vote_methods = TRUE
)

# Statewide vote-method totals
res$vote_method_state %>%
  filter(office == "United States Senator") %>%
  select(candidate, party, votes_advance_in_person, votes_election_day,
         votes_absentee, votes_provisional, votes_total)

# County vote-method breakdown
res$vote_method_county %>%
  filter(office == "United States Senator", county == "Fulton") %>%
  select(candidate, votes_advance_in_person, votes_election_day,
         votes_absentee, votes_provisional, votes_total)

Faster county scraping with more parallel browsers

Georgia has 159 counties. By default, county pages are scraped with 4 parallel Chromium browsers. Increasing max_workers can cut runtime significantly on machines with enough CPU/RAM.

res <- scrape_elections(
  state       = "GA",
  year_from   = 2024,
  year_to     = 2024,
  level       = "all",
  max_workers = 8L
)

Multi-year scrape

res <- scrape_elections(
  state       = "GA",
  year_from   = 2020,
  year_to     = 2024,
  level       = "all",
  max_workers = 4L
)

Filter to a specific county

res <- scrape_elections(state = "GA", year_from = 2022, year_to = 2022)

res$county %>%
  filter(county == "Fulton") %>%
  select(election_name, office, candidate, party, votes, pct) %>%
  arrange(office, desc(votes))

Output columns

`$state` (when `level = "all"` or `"state"`)

Column	Description
`election_name`	Human-readable election name (e.g. `"November General Election"`)
`election_year`	Calendar year of the election
`election_slug`	URL slug (e.g. `"2024NovGen"`)
`election_date`	Date string from the page header
`result_status`	Reporting status (e.g. `"OFFICIAL RESULTS"`)
`office`	Office name
`localities_reporting`	`"X/Y"` localities reporting string
`candidate`	Candidate name (cleaned — `(I)` and party suffix stripped)
`party`	Party abbreviation
`is_incumbent`	`TRUE` if `(I)` appears in raw candidate name
`is_winner`	Always `NA` — winner marker absent from GA SOS HTML
`votes`	Vote total
`pct`	Percentage of votes
`url`	URL of the election page

`$county` (when `level = "all"` or `"county"`)

Same columns as $state, plus:

Column	Description
`county`	County name

`$vote_method_state` / `$vote_method_county` (when `include_vote_methods = TRUE`)

Same identifying columns as above, with votes split by method instead of a single votes + pct:

Column	Description
`votes_advance_in_person`	Advance in Person ballots
`votes_election_day`	Election Day ballots
`votes_absentee`	Absentee by Mail ballots
`votes_provisional`	Provisional ballots
`votes_total`	Sum of all methods

Performance notes

level = "state" is much faster. County scraping opens up to max_workers parallel Chromium browsers across Georgia’s 159 counties. A full state + county scrape for one year takes 20–60 minutes. Use level = "state" when county breakdowns are not needed.
include_vote_methods is significantly slower. Each contest panel requires an extra Playwright click-and-re-render step. Budget an extra 5–15 minutes per election page.
Virtual scroll. The GA SOS site uses Angular virtual scrolling, rendering only panels visible in the viewport. The scraper automatically scrolls to load all panels before parsing — without this, pages with 100+ contests would silently return incomplete data.
Multiple elections per year. The GA SOS site lists all elections for a year (primaries, runoffs, generals, specials). All are scraped and combined in the output.