Methodology & Data Integrity

What These Numbers Cannot Tell You

Complaint counts are not normalised for sales volume. Best-selling vehicles naturally receive more complaints in absolute numbers.

Complaints are self-selected reports, not verified defects. Older vehicles accumulate more complaints purely by existing longer.

Because of these limits, we never publish a reliability score, a best/worst year claim, or advice to avoid a vehicle. Any site claiming to rank vehicles by raw complaint counts is misusing federal data.

The Variant Merge — Why Our Counts Differ

NHTSA does not store one record per consumer vehicle. It stores records against raw model strings, splitting trucks and SUVs across multiple cab and trim designations. We merge body-style variants deterministically while keeping weight classes distinct.

Core Normalization Rules

• We merge body styles (SUPERCAB, SUPER CREW, REGULAR CAB, etc.) because a buyer researching a model year wants the complete component problem picture.
• We never merge weight classes (e.g. F-150 and F-250, 1500 and 2500 remain strictly separated).
• Where a model string is ambiguous, software suggests a candidate but only a human operator approves the merge.

How Each Figure Is Calculated

Complaint and recall counts reflect distinct records de-duplicated by NHTSA identifier. Component breakdown is the share of reports naming each system, which is mathematically robust because the denominator is the vehicle itself.

AI Disclosure & Site Generation

No page text on this site is written by an AI language model. All page content is deterministically templated from verified structured data. AI is used solely for component taxonomy classification and for suggesting candidate merges that a human reviewer validates.

Raw NHTSA string → YearTell canonical

How often the snapshot refreshes

Complaints and recalls ingest daily. Investigations and EPA ingest weekly (Sunday UTC). vPIC ingest is monthly. Pages stamp the date of the freshest snapshot that passed QA, not the build clock.

Equal-exposure window on compare pages

An adjacent-year comparison counts owner reports in the same number of months after each model year entered the file, so the later year is not punished for having had less time on the road. Pairs are indexed only when that window shows a large delta, a redesign boundary, or a named featured pair.

Component classifier

NHTSA component strings are mapped onto a fixed system list (power train, engine, electrical, and the rest). Ambiguous merges are suggested by software and approved by a person before they affect a published page. Page text is templated from those tables; it is not written by a language model.

Methodology Revision History