Data Sources — dui-by-state
All data used in this project is from official U.S. government sources or hand-curated from those same sources. No data is sourced from Wikipedia or other crowd-edited references.
State Reference Data
State FIPS Codes & Abbreviations
- Source: U.S. Census Bureau
- File:
state.txt - URL: https://www2.census.gov/geo/docs/reference/state.txt
- Format: Pipe-delimited text
- Fields used: STATE (FIPS numeric), STUSAB (2-letter abbreviation), STATE_NAME
- License: Public domain (U.S. government work)
- Notes: Official FIPS 5-2 state codes. Includes 50 states + DC + territories; project filters to 50 states + DC only.
State Geographic Reference (lat/lng centroids, land area, Census region/division)
- Source: U.S. Census Bureau — 2024 Gazetteer Files (internal point coordinates and area measurements)
- URL: https://www.census.gov/geographies/reference-files/time-series/geo/gazetteer-files.html
- Format: Hand-curated CSV stored at
data/raw/state_reference.csv - Fields: state_fips, state_abbr, state_name, lat, lng, land_area_sq_mi, water_area_sq_mi, region, division
- License: Public domain (U.S. government work)
- Notes: The Gazetteer file is WAF-protected against automated download. Values were transcribed manually from the official Census Gazetteer and the Census Bureau’s regional classification. Lat/lng are Census internal points (not geographic centroids, not capital cities). Area figures are from the 2020 Census. Regions and divisions follow the standard Census Bureau classification unchanged since 1984.
- Last verified: 2026-08-01
State Population Estimates
- Source: U.S. Census Bureau — Population and Housing Unit Estimates
- File:
NST-EST2024-ALLDATA.csv - URL: https://www2.census.gov/programs-surveys/popest/datasets/2020-2024/state/totals/NST-EST2024-ALLDATA.csv
- Format: CSV
- Fields used: NAME, STATE (FIPS), POPESTIMATE2020–POPESTIMATE2024
- License: Public domain (U.S. government work)
- How the source collects the data: Model-based postcensal estimates: the 2020 decennial base rolled forward with births, deaths, and migration. Not a survey/fresh count.
- How the source defines the data: Resident population as of July 1 of each year. Used here only as a denominator for per-100k rates.
- Methodology changes / series breaks: Postcensal estimates are re-based (a new “vintage”) after each decennial census; using one vintage (2020–2024 series) end-to-end avoids mixing revisions. Error grows with distance from the 2020 base.
- Known controversies / debates: Census undercount debates; immaterial at state denominator scale.
- Notes: Summary level 40 = state; project filters out national total (SUMLEV=10) and territories (STATE FIPS > 56).
DUI / Traffic Safety Data
NHTSA FARS — Fatality Analysis Reporting System (2015–2024)
- Source: National Highway Traffic Safety Administration (NHTSA)
- URL pattern: https://static.nhtsa.gov/nhtsa/downloads/FARS/{year}/National/FARS{year}NationalCSV.zip
- Format: ZIP containing CSV files
- File used:
accident.CSVwithin each annual ZIP - Key fields: ST_CASE, STATE, STATENAME, YEAR, FATALS, DRUNK_DR
- License: Public domain (U.S. government work)
- How the source collects the data: A census of every fatal traffic crash on a U.S. public road, compiled by NHTSA (NCSA) from each state’s crash reports, death certificates, and related records via trained state analysts. Universe = crashes on public roads that killed at least one person within 30 days. Not a sample.
- How the source defines the data:
- Fatal crash: at least one death within 30 days of a public-road motor-vehicle crash.
DRUNK_DR: count of drivers in the crash with any alcohol involvement as coded in FARS.DRUNK_DR > 0is a raw-coded flag (at least one drinking driver), which is not the same as NHTSA’s imputed “alcohol-impaired (BAC ≥ 0.08)” estimate — see the imputed-estimates source below. Raw coding tends to undercount vs the imputed model.- Alcohol-impaired (NHTSA’s published definition): a driver/motorcycle operator with BAC ≥ 0.08 g/dL.
- Methodology changes / series breaks:
- FARS is broadly consistent year to year, but coding variables and their availability have changed over the decades (new/retired data elements). Within this project’s 2015–2024 window the core fields used are stable.
- Raw
DRUNK_DRcoding vs multiple-imputation estimates are two different measures — never chart a raw-coded count next to an imputed BAC≥0.08 estimate as if equivalent. - Early-release annual files are revised in later editions; figures here reflect the first release (2024 released April 2026).
- Known controversies / debates: Whether to report raw-coded or imputed alcohol involvement is the central measurement choice; NHTSA treats the imputed estimates as authoritative. Missing-BAC rates vary widely by state, which biases raw counts.
- Coverage: 2015–2024 (10 years). Data covers all 50 states + DC.
- Notes: This is the standard dataset for alcohol-involved fatal crashes in the U.S.
NHTSA Traffic Safety Facts — State Alcohol-Impaired-Driving Estimates
- Source: National Highway Traffic Safety Administration (NHTSA) — CrashStats
- URL: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813813
- Format: PDF
- License: Public domain (U.S. government work)
- Notes: Pre-aggregated state-level summary published annually. Includes alcohol-impaired fatalities per 100 million vehicle miles traveled (VMT). Used to cross-check FARS aggregate calculations.
DUI Laws & Penalties
NCSL — Per Se DUI/DWI Laws
- Source: National Conference of State Legislatures (NCSL)
- URL: https://www.ncsl.org/transportation/dui-dwi-per-se-laws
- Format: HTML table (scraped)
- License: Public domain legislative reference
- Notes: Most authoritative compiled source for state per se BAC laws. NCSL is a nonpartisan organization serving state legislatures.
roadlawguide.com — DUI Penalties by State
- Source: roadlawguide.com
- URL: https://roadlawguide.com/dui-dwi-laws/penalties-by-state/
- Format: HTML (scraped)
- Notes: Structured first-offense penalty comparison table. Used as a secondary cross-reference. Always verify against official state statutes before publishing.
ailawyer.pro — DUI Penalties by State
- Source: ailawyer.pro
- URL: https://ailawyer.pro/tools/dui-penalties-by-state
- Format: HTML (scraped)
- Notes: Structured penalty data per state for 2026. Used as tertiary cross-reference. Always verify against official state statutes.
DUI Arrests
FBI UCR — Arrests by Age, Sex, and Race, Summarized Yearly, 2023
- Source: Federal Bureau of Investigation, Uniform Crime Reporting Program
- Distributor: Inter-university Consortium for Political and Social Research (ICPSR)
- Study: ICPSR 39298
- URL: https://www.icpsr.umich.edu/web/NACJD/studies/39298
- Format: Tab-separated values (extracted from ZIP)
- Fields used: STATE (UCR code), OFFENSE (220=DUI), age/sex arrest counts
- License: ICPSR Terms of Use — no redistribution of raw data; derivative analysis and aggregated statistics permitted; citation required
- Coverage: 50 states + DC, 2023 (agency-level, aggregated to state)
- How the source collects the data: Voluntary administrative reporting by law enforcement agencies to the FBI’s UCR program. Not all agencies report, so it is neither a census nor a probability sample — it is a self-selected subset weighted toward participating agencies.
- How the source defines the data: Arrest = a UCR arrest event for offense code 220 (driving under the influence). An arrest is not a conviction, and counts events, not unique people. Coverage is measured against the population of reporting agencies, not the state’s full population.
- Methodology changes / series breaks:
- Major break: SRS → NIBRS-only on January 1, 2021. The FBI retired the Summary Reporting System and moved to NIBRS-only collection. Many agencies had not yet transitioned, so 2021 national coverage dropped sharply and figures rely on estimation; 2022 recovered to ~93.5% coverage partly by re-accepting SRS data. Arrest/offense counts across the 2020→2021→2022 boundary reflect reporting-coverage changes, not just real crime change — do not read year-over-year moves there as behavioral. (2023 data used here is post-transition; still coverage-limited.)
- Because coverage varies by state and year, absolute counts are not comparable across states or years; use per-reporting-population rates only.
- Known controversies / debates: The NIBRS transition and its coverage gap drew significant criticism (DOJ review); UCR undercounts are a long-standing, well-documented limitation.
- Notes: Use for relative comparisons and per-reporting-population rates, never as absolute arrest counts. Always note reporting coverage when publishing.
- Citation: United States Department of Justice. Federal Bureau of Investigation. Uniform Crime Reporting Program Data: Arrests by Age, Sex, and Race, Summarized Yearly, United States, 2023. Inter-university Consortium for Political and Social Research [distributor], 2026.
- Retrieved: 2026-08-02
Alcohol Consumption
NIAAA — Apparent Per Capita Alcohol Consumption by State (2022)
- Source: National Institute on Alcohol Abuse and Alcoholism (NIAAA)
- Publication: Surveillance Report #121
- URL: https://www.niaaa.nih.gov/sites/default/files/surveillance-report121.pdf
- Format: PDF — Table 2 extracted via
pdfplumber - Fields used: state_name, ethanol_per_capita_gallons_2022, consumption_decile_2022
- License: Public domain (U.S. government work)
- How the source collects the data: Apparent consumption derived from alcohol sales/tax/shipment records (gallons of beverage converted to ethanol), divided by the population aged 14+. It measures alcohol sold in a state, not alcohol drunk by residents — a model/administrative proxy, not a survey of drinkers.
- How the source defines the data: Per-capita ethanol in gallons (beer + wine + spirits) for population ages 14+. Decile 1 = highest consumption, 10 = lowest.
- Methodology changes / series breaks: Cross-border sales distort state figures (tourism, low-tax states, military/tribal sales), so a high value can reflect purchasing location rather than resident behavior. Population-base and beverage conversion factors have been revised over the surveillance-report history; use one report year (2022 here) rather than splicing editions.
- Known controversies / debates: “Apparent consumption” systematically differs from self-reported survey consumption (e.g. BRFSS); the sales-based method is standard but known to misattribute cross-border purchases.
- Coverage: 50 states + DC, 2022. National average 2022: 2.50 gallons.
NHTSA Imputed Alcohol Fatality Estimates
NHTSA Traffic Safety Facts — State Alcohol-Impaired-Driving Estimates 2024
- Source: NHTSA — CrashStats
- URL: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813813
- Format: PDF — Table 2 extracted via
pdfplumber - Fields used: total_fatalities_2024, alcohol_impaired_fatalities_2024 (BAC≥.08), pct_alcohol_impaired_2024, high_bac_fatalities_2024 (BAC≥.15), pct_high_bac_2024
- License: Public domain (U.S. government work)
- How the source collects the data: Derived from FARS (same census of fatal crashes) but with a statistical model layered on top: for drivers with missing BAC, NHTSA applies multiple imputation to estimate BAC across the full 0–0.94 g/dL range, then aggregates. So these are modeled estimates, not raw counts.
- How the source defines the data: Alcohol-impaired fatality = a death in a crash involving a driver/motorcycle operator with (estimated) BAC ≥ 0.08 g/dL. High-BAC = BAC ≥ 0.15. The imputation fills in the ~large share of drivers who were never tested.
- Methodology changes / series breaks:
- This is the imputed series; FARS raw
DRUNK_DRis a different series. In this project the imputed estimate (~30% of fatals) is treated as authoritative and the raw FARS coding (~15%) is the undercount — do not mix them in one comparison. - NHTSA adopted the current multiple-imputation approach in the early 2000s (replacing an older 3-category discriminant method); pre-2000s imputed figures are not directly comparable. Not an issue for 2024-only use here.
- This is the imputed series; FARS raw
- Known controversies / debates: Imputation assumptions (which drivers get assigned a high BAC) are periodically debated, but the method is the accepted federal standard.
- Coverage: 50 states + DC, 2024. National 2024: 11,907 alcohol-impaired fatalities (30% of 39,254 total).
DUI Enforcement & Procedures
Sobriety Checkpoint Legality
- Source: Multiple legal references (Governors Highway Safety Association, state constitutions, case law)
- Method: Compiled from verified legal references — 12 states ban checkpoints
- License: Public legislative reference
- States banning checkpoints: AK, ID, IA, MI, MN, MT, OR, RI, TX, WA, WI, WY
NCSL — State Ignition Interlock Laws (All-Offender)
- Source: National Conference of State Legislatures
- URL: https://www.ncsl.org/research/transportation/state-ignition-interlock-laws.aspx
- Format: HTML table (scraped)
- License: Public domain legislative reference
- Notes:
all_offender= “Yes” means IID mandatory for all DUI offenders including first offense. 31 states as of 2024.
Implied Consent & Refusal Penalties
- Source: Foundation for Advancing Alcohol Responsibility (responsibility.org)
- URL: https://www.responsibility.org/wp-content/uploads/2026/01/Implied-Consent-Refusal-Penalties_Aug-2025.pdf
- Format: PDF (96 pages of statutory text per state)
- License: Public reference
- Notes: All 50 states + DC have implied consent. 17 states impose criminal penalties (jail/fine) for test refusal; all others impose administrative suspension only. Wyoming is the only state without refusal penalties.
Open Container Laws
- Source: NHTSA / federal compliance records
- Notes: 40 states meet federal open container standards. 11 states allow open alcohol containers in vehicles under some circumstances.
BAC Limits
- Notes: All states: 0.08 g/dL. Utah: 0.05 g/dL (lowest in U.S., effective 2018).
Notes on Data Quality & Verification
- All government source files are saved verbatim to
data/raw/and never modified. - Any discrepancies between scraped law/penalty sources should be resolved by consulting the official state statute directly.
- Population figures are estimates; the decennial Census (2020) is the only complete enumeration.
- FARS fatality figures may be revised in subsequent NHTSA releases; figures here reflect the first annual release.
Methodology & definitions — the law-snapshot sources
Most of the law/penalty/enforcement tables below (NCSL per se laws, NCSL IID laws, roadlawguide, ailawyer, IIHS speed limits, NASID enforcement, checkpoint legality, implied-consent, open-container, body-cam, vehicle impound) share the same profile, so their required fields are captured once here rather than repeated:
- Collection method: hand-curated or scraped point-in-time snapshots of statute status, transcribed from an authority (NCSL, IIHS, NASID, NHTSA, or the state code). Not surveys, not counts — a coded reading of law as of a retrieval date.
- Definitions: each “Yes/No/category” is a coding judgment about statutory text (e.g. what counts as “mandatory-all-offender” IID, or a “felony threshold”). The coding rules are noted per table; edge cases are resolved against the official statute.
- Methodology changes / series breaks: these are current-snapshot tables with no time series — a value reflects the law on the retrieval/verification date only. Laws change frequently, so do not compare one snapshot against an older edition without re-coding both from the same rules. When a source revises its compilation, treat it as a new snapshot, not a continuation.
- Known controversies / debates: secondary compilations (roadlawguide, ailawyer) can disagree with each other and with statute; always defer to the official state statute before publishing a penalty claim.
Series breaks that matter most for comparisons (quick reference)
- FARS raw
DRUNK_DR≠ NHTSA imputed BAC≥0.08 estimate — different measures; the imputed one is authoritative. Never chart them side by side as equals. - FBI UCR SRS→NIBRS-only break at Jan 1, 2021 — arrest/offense counts around 2020–2022 reflect reporting-coverage changes, not just real change. Per-reporting-population rates only.
- NIAAA apparent consumption = alcohol sold, not resident drinking — cross-border sales distort state values.
- Census population = postcensal estimates, re-based each decennial vintage.
Structural Controls (added for analysis depth)
FARS 2024 — Prior DWI Convictions & Crash Speed Limits
- Source: NHTSA FARS 2024 —
vehicle.csv - URL: https://static.nhtsa.gov/nhtsa/downloads/FARS/2024/National/FARS2024NationalCSV.zip
- Fields used:
PREV_DWI(prior DWI convictions),VSPD_LIM(posted speed limit at crash location); joined withdrimpair.csvcode 9 to identify impaired drivers - License: Public domain (U.S. government work)
- Notes: Aggregated to state level.
PREV_DWIis a count of prior DWI convictions found on the driver’s record. Codes 98/99/998 = unknown and are excluded. Key finding: the vast majority (85–95%) of impaired drivers in fatal crashes have no prior DWI conviction on record. - Coverage: 50 states + DC, 2024
- Retrieved: 2026-08-03
IIHS — Maximum Posted Speed Limits
- Source: Insurance Institute for Highway Safety (IIHS)
- URL: https://www.iihs.org/research-areas/speed/speed-limit-laws
- Format: HTML table (transcribed)
- Fields used: Rural interstate maximum posted speed limit per state
- License: IIHS public reference
- Notes: Reflects maximum posted limits including “specified segments of road” where higher limits apply (e.g., Idaho 80 mph on some interstates). Range: 55 mph (DC) to 85 mph (Texas).
- Last verified: August 2026
FHWA — Vehicle Miles Traveled by State (2022)
- Source: Federal Highway Administration — Highway Statistics 2022, Table VM-2
- URL: https://www.fhwa.dot.gov/policyinformation/statistics/2022/vm2.cfm
- Format: Tabular (transcribed from published table)
- Fields used: Annual vehicle miles traveled in millions
- License: Public domain (U.S. government work)
- Notes: 2022 is the latest complete year available. Used to compute fatality rate per 100 million VMT — the standard road safety comparison metric that controls for how much driving occurs in a state.
- Retrieved: 2026-08-03
BAC Testing Rates & Laws
FARS 2024 — BAC Testing Rates by State
- Source: NHTSA FARS 2024 —
person.csv - URL: https://static.nhtsa.gov/nhtsa/downloads/FARS/2024/National/FARS2024NationalCSV.zip
- Computed from:
ALC_STATUSfield (0=Test Not Given, 2=Test Given, 8=Not Reported, 9=Unknown) - License: Public domain (U.S. government work)
- Fields: pct_bac_known_killed, pct_bac_known_all, pct_bac_known_surviving, pct_blood_test (blood vs breath/PBT share), test type counts by state
- DuckDB table:
fars_bac_testing_2024 - Notes: Computed percentage of drivers in fatal crashes with known BAC test results, broken out by killed vs surviving. National avg: 66.8% of killed drivers have known BAC. Range: 9.6% (Mississippi) to 97.7% (Vermont). States with mandatory coroner/ME testing laws average 78.2% vs 51.7% for probable-cause states.
- Retrieved: 2026-08-07
State Mandatory BAC Testing Laws
- Source: Hand-curated from NHTSA Casanova et al. 2012 (DOT HS 811 661) + current state statutes
- Reference URL: https://rosap.ntl.bts.gov/view/dot/1940
- Format: CSV (
data/raw/bac_testing_laws.csv) - Fields: mandatory_testing_law, testing_scope, testing_authority, no_refusal_program, statute_citation, notes
- License: Public domain (legislative references)
- DuckDB table:
bac_testing_laws - Notes: 29 states require BAC testing of fatally injured drivers via coroner/ME statute. 22 states use probable-cause only. The Casanova 2012 NHTSA report identified 25 mandatory states; we updated with 4 additional states that have enacted similar statutes since. Each entry includes the specific statute citation.
- Retrieved: 2026-08-07
DUI Enforcement Procedures (NASID)
NASID — State DUI Enforcement Laws Database
- Source: National Alliance to Stop Impaired Driving (NASID)
- URL: https://nasid.org/state/{state-slug}/
- Format: HTML state pages (scraped 51 pages)
- License: Public reference (data sourced from NHTSA/FARS, May 2024)
- DuckDB tables:
nasid_enforcement(raw text),nasid_enforcement_clean(standardized) - Fields (12 per state):
- Sobriety Checkpoints (permitted/prohibited)
- No Refusal Programs (active/authorized/not authorized)
- Roadside PBT Laws (authorized/not explicit)
- Ignition Interlocks (mandatory_all/high_bac_repeat/repeat/discretionary)
- Felony DUI threshold (2nd/3rd/4th offense)
- DUI Look-back Period (5–lifetime years)
- Enhanced High-BAC Threshold (0.15–0.20)
- Open Container compliance (federal compliant/non-compliant)
- Implied Consent Testing Methods (blood/breath/urine/oral fluid)
- ALS/ALR (administrative license suspension enacted)
- Social Host Laws
- ALR/Hardship license
- Retrieved: 2026-08-07
- Notes: Key distributions: 39 states permit checkpoints; 9 actively use no-refusal programs; 32 have mandatory-all-offender IID; 33 explicitly authorize PBT; 48 have felony DUI laws; 47 have enhanced high-BAC penalties.
Source Provenance in DuckDB
Every table in data/project.duckdb has a corresponding entry in the
_sources metadata table:
SELECT duckdb_table, source_name, methodology, series_breaks FROM _sources;
This table records, for every dataset loaded into the database: duckdb_table,
source_name, url, license, notes, retrieved, and — so methodology travels
with the data — methodology (how the source collects and defines the data) and
series_breaks (boundaries across which the numbers are NOT comparable). The two
most important flags recorded there: FARS raw DRUNK_DR vs the NHTSA imputed BAC≥0.08
series (different measures), and the FBI UCR SRS→NIBRS break at Jan 1, 2021. Keep these
in sync with the per-source sections above.
Body-Worn Camera Laws
NCSL Body-Worn Camera Laws Database
- Source: National Conference of State Legislatures (NCSL)
- URL: https://www.ncsl.org/civil-and-criminal-justice/body-worn-camera-laws-database
- Format: HTML text → hand-curated CSV (
data/raw/body_cam_laws_ncsl.csv) - Fields: state_abbr, state_name, has_bwc_law, mandate_statewide, funding_provided, retention_policy, foia_provisions, year_first_law, notes
- License: Public domain (legislative information)
- Retrieved: 2025-08-07
- DuckDB table:
body_cam_laws - Notes: 34 states + DC have enacted BWC legislation. 8 states mandate statewide use (CO, CT, DE, IL, MD, NJ, NM, SC). Data coded from NCSL narrative descriptions by categorizing: (1) whether a mandate exists, (2) whether state funding was appropriated, (3) whether retention periods are specified, and (4) whether FOIA/public records provisions address BWC footage. Not all nuance is captured — see notes field for details.
Vehicle Impound / Seizure / Forfeiture Laws
Vehicle Sanctions by State
- Source: NHTSA Countermeasures That Work (2023), NHTSA DOT HS 811 028B (2008), CDC MV PICCS, individual state statutes
- Primary URL: https://www.nhtsa.gov/book/countermeasures-that-work/alcohol-impaired-driving/countermeasures/other-strategies-behavior-5
- Secondary URL: https://www.cdc.gov/transportation-safety/calculator/impoundment.html
- Format: Hand-curated CSV stored at
data/raw/vehicle_impound_laws.csv - Fields: state_fips, state_abbr, state_name, vehicle_impound_law, impound_trigger, impound_mandatory, impound_duration_days, vehicle_forfeiture_law, forfeiture_trigger, plate_impound_law, source_note
- License: Public domain (U.S. government work)
- Retrieved: 2026-08-15
- Notes: Compiled from NHTSA’s “Countermeasures That Work” (11th edition, 2023) and “Update of Vehicle Sanction Laws and Their Application, Vol. II” (DOT HS 811 028B, 2008), CDC MV PICCS intervention documentation, and current state statutes verified via official .gov statute databases. Distinguishes between mandatory vehicle impound (automatic upon arrest/conviction) and discretionary (at officer/court discretion). Covers impound, immobilization, forfeiture, and plate impound.