MOT Risk Index Evidence Capsule: Great Britain vehicle-family and defect-group aggregates, 2024

Version 2024-v5 (26 September 2026) — correction. v4 counted MOT tests from 2019 to 2023 twice in the model-year file (a pipeline error introduced on 21 September 2026): the n column was inflated, the failure share weighted 2019–2023 double and the intervals were too narrow. v5 recounts each test once; 5,665 model-year rows follow the corrected intervals. All other files carry the same numbers as v4. The licence notice for every source (ATTRIBUTION.md) is now inside the package. Version 2024-v4 (25 September 2026). The family and defect-group numbers are unchanged from 2024-v3 (same DVSA 2024 source release); only the release identifier changed. Three files add layers the site already shows: outside sources (DVSA recall campaigns; periodic-inspection failure shares in Norway, Finland and the Netherlands, each against that country's own average, at least 500 inspections; NHTSA complaint and service-bulletin counts only for explicit same-vehicle pairs with a US model-year window), first-time failure by odometer band inside age groups 3–5, 6–9, 10–14 and 15+ (at least 50 tests per cell; under 500 flagged rough estimate with a Wilson 95% interval), and model-year ratios against the same-age norm with 95% intervals. v3 (2026-09-24) corrects v1 and v2. The DVSA 2024 source files repeat many test records (66,857,355 rows hold 42,598,624 distinct tests); v1 and v2 counted the repeats. v3 keeps one record per test ID: class-4 initial tests 33,012,215 (previously 52,245,925), national initial-failure rate 28.28% (previously 26.01%). Coverage grows from 34 to 625 published vehicle families. Method and changelog: methodology, changelog. Aggregated 2024 Great Britain MOT initial-test outcomes for 34 vehicle families, derived from the DVSA Anonymised MOT tests and results bulk dataset (Open Government Licence v3.0). The headline metric is the age-adjusted MOT initial-failure ratio: observed first-time failures (F + PRS) divided by the count expected when frozen national class-4 rates are applied to the family's own age strata, multiplied by 100 (100 = expected for that age profile). A second table gives family-by-defect-group failure-item rates per 100 tests, excess in percentage points and over-representation ratios against the national benchmark, for eleven prespecified top-level defect groups. Disclosure control: cells with n < 1,000 are omitted; rows with n between 1,000 and 5,000 are flagged low_sample; where exactly one cell in a family slice is suppressed, the next-smallest is also removed so the first cannot be recovered by subtraction. No vehicle-level records, registration marks or geographic slices are included. These are observational inspection outcomes, not causal estimates; mileage is not equalised between families. Canonical dataset page: motriskindex.co.uk. Attribution and licences. DVSA MOT tests and results, DVSA vehicle safety recalls and DfT licensing statistics: contains public sector information licensed under the Open Government Licence v3.0. Norway: Statens vegvesen, Periodisk kjøretøykontroll (PKK) 2024, CC BY 4.0, adapted (aggregated by model and matched to UK families). Finland: Traficom, periodic inspections of cars by model 2024, CC BY 4.0, adapted. Netherlands: RDW open data, CC0 1.0 (credited, not an RDW endorsement). United States: NHTSA complaints and manufacturer communications, public domain. MOT Risk Index derivations: CC BY 4.0. Full text: https://motriskindex.co.uk/data/ATTRIBUTION.md