Research Methodology

The Primary Source Doctrine

The foundational editorial rule at Daily Driver Advocate is what we call the “Primary Source Doctrine”: every factual claim published on the site must be traceable to an original, authoritative source. In practice, this means:

1. Statutory claims— such as “Pennsylvania mandates a three-month license suspension for a lapse in coverage” — must cite the specific state statute (e.g., PA Title 75) or the official DMV administrative rule.
2. Statistical claims — such as “There were 33,654 fatal crashes in the United States in 2024” — must be derived from the raw federal dataset (e.g., NHTSA FARS accident.csv), not from a news article that reported the number.
3. Regulatory claims— such as “Commercial carriers operating vehicles over 10,001 lbs must maintain $750,000 in liability coverage” — must cite the specific Code of Federal Regulations section (e.g., 49 CFR Part 387).

Secondary sources — news articles, insurance company blog posts, legal commentary — are used for context and background, but they are never the sole citation for a factual claim. When secondary sources are referenced, they are clearly labeled as such.

How We Research State Vehicle Laws

When we produce a state-by-state legal analysis — such as our report on whether it is illegal to drive without insurance — we follow a structured, multi-stage research workflow:

Federal Dataset Acquisition & Processing

A core differentiator of Daily Driver Advocate is that we work directly with raw federal datasets rather than relying on third-party summaries. This section documents exactly which datasets we use, where they come from, and how we process them.

When a federal agency publishes a dataset, we download the original file package directly from the agency's official distribution point. We store the raw source files in our data repository alongside the derived summary files that power our visualizations. This means our entire analytical chain — from raw data to published statistic — is internally reproducible and externally auditable.

The NHTSA FARS Pipeline

The Fatality Analysis Reporting System (FARS) is a census of fatal traffic crashes maintained by the National Highway Traffic Safety Administration (NHTSA), a division of the U.S. Department of Transportation. FARS is the gold standard dataset for fatal crash research in the United States — it is the same dataset used by NHTSA itself, the Insurance Institute for Highway Safety (IIHS), and peer-reviewed transportation safety researchers.

Source Files

How We Process FARS Data

1. Download: The raw ZIP package is downloaded directly from the NHTSA file distribution server.
2. Extract: The ZIP contains multiple CSV files. For crash-level analysis, the primary file is accident.csv, where each row represents one fatal crash. For person-level demographics, we use person.csv.
3. Aggregate: We aggregate crash counts by state using the FIPS STATE column, cross-referencing the FARS codebook to map numeric codes to state names. We derive fatality counts from the FATALS column.
4. Normalize: For per-capita calculations, we divide state-level crash or fatality counts by the corresponding Census Bureau population estimate, then multiply by 100,000 to produce a rate per 100,000 residents.
5. Export: The derived summary data is exported as TypeScript modules that are directly consumed by our React-based visualization components. This eliminates runtime data fetching and ensures the published numbers are locked to the exact dataset version we verified.

Census Population Data

Population denominators are essential for producing meaningful per-capita statistics. We source our population data directly from the U.S. Census Bureau's official state population estimates tables.

Current Coverage

• Vintage 2024 estimates: Downloaded from the Census Bureau's official data distribution at census.gov . This file provides state-level population estimates for the 2020–2024 period.
• Historical series (1975–2024): We maintain a stitched historical population file covering 1975–2024 for all 50 states plus D.C. This file is constructed from the Census Bureau's intercensal and postcensal population estimate series for each decade, combined with the 2024 vintage data for the most recent years. Puerto Rico is excluded from the historical series because the older Census decade files do not extend back to 1975.

Why We Use Census Data

The Census Bureau's Vintage population estimates are the standard denominator used by federal agencies themselves — including NHTSA, the CDC, and the FBI — when computing per-capita rates. By using the same population source as the agency that produces FARS, we ensure our per-capita calculations are directly comparable to official federal statistics.

The Fact-Checking & Verification Cycle

Our editorial process includes a structured verification cycle applied to every report before publication and at regular intervals afterward:

Interactive Visualization Standards

Our interactive maps, charts, and data tables follow a strict set of design and data integrity principles:

• All visualizations are rendered from static, pre-computed data that is locked to a specific dataset version. We do not fetch data at runtime from external APIs, which eliminates the risk of displaying stale or inconsistent numbers.
• Hover states on maps and charts always display the underlying data values, not just visual indicators. A user hovering over Texas on a fatal crash map sees the exact crash count, fatality count, and rank — not just a color shade.
• Every visualization is accompanied by explanatory text that documents what the data represents, which source it comes from, and how derived calculations (such as per-capita rates) were computed.
• Source citations are embedded directly below visualizations, linking to the Primary Source Directory.

Update & Correction Policy

We recognize that laws change, datasets are updated, and errors can occur despite rigorous fact-checking. Our update and correction policy is designed to maintain trust:

• Statutory updates: When a state legislature amends a vehicle code provision covered by one of our reports, we update the report to reflect the new law and reset the “Verified” badge date. Changes to statutory data are not treated as corrections — they are treated as updates.
• Dataset refreshes: When a new vintage of a federal dataset is released (e.g., FARS 2025), we will update our data-driven reports to incorporate the new data and clearly indicate the dataset version used.
• Factual corrections: If we discover or are notified of a factual error in a published report — such as an incorrect statutory limit or a miscalculated statistic — we investigate and issue a correction as quickly as possible. The correction is documented with a permanent note on the page.
• Transparency: We never silently alter published content. All substantive changes are documented and dated.

Limitations & Scope Boundaries

We believe transparency requires acknowledging what we are and what we are not:

• We are not a law firm. Our content is informational research, not legal advice. It does not create an attorney-client relationship and should not be used as a substitute for consultation with a licensed attorney in your jurisdiction.
• Laws change. Despite our best efforts to maintain current information, statutes are subject to amendment, repeal, and reinterpretation by courts. The “Verified” date on each report indicates the last time we confirmed the accuracy of the underlying sources. Between verifications, legislative changes may occur.
• We cover U.S. law only. Our research scope is limited to the 50 U.S. states and the District of Columbia. We do not cover U.S. territories, foreign jurisdictions, or military installations.
• FARS data has inherent limitations. FARS is a census of crashes involving at least one fatality. It does not capture non-fatal crashes, injury-only crashes, or property-damage-only incidents. Per-capita rates based on FARS reflect fatal crash frequency relative to population — not overall driving risk.
• Population estimates are estimates. The Census Bureau's Vintage population series are postcensal estimates revised annually. They are highly accurate for most analytical purposes but are not exact census counts.