Methodology and limitations

The unit is an observed entity-period-stage record. Published second-stage cases are not a cohort of all parking charges and are not a first-stage operator performance measure.

Outcome measures

AOP₂ adjudicated: Allowed ÷ (allowed + refused). The denominator contains decided appeals only.
AOP₃ closed-case favourable share: (Allowed + operator withdrawals) ÷ all reported closures. It is not calculated per operator here because POPLA publishes only a combined per-operator not-contested total.
Operator-withdrawal rate: Operator withdrawals ÷ all reported closures. It remains unavailable at operator level until the source reports the withdrawing party by operator.
AOP₄ intensity: ADR submissions per 1,000 KADOE requests over a fully overlapping 12-month window. KADOE is a descriptive issuance proxy, not a count of parking charges.

Sample sizes and uncertainty

Values are hidden when n<20. Records with 20–49 observations carry a low-confidence badge. At n≥50 the observed value is shown without that badge.

A Wilson interval is shown only for binomial proportions and is labelled “binomial model interval (iid assumption)”. It does not cover mapping, selection, reporting or KADOE-proxy uncertainty. AOP₄ has no interval.

Systematic limits remain. Only cases reaching the appeals service are observed; operator practice, appellant mix, reporting rules and regime exposure can change between periods.

Not-contested boundary

POPLA’s 2025 monthly table reports 34,502 operator withdrawals and 5,020 appellant withdrawals service-wide. Its operator table reports only a combined not-contested count. This project never allocates the service-wide split to individual operators.

The 2025 submitted table also contains a source-level reconciliation issue: its service total is 104,349 while printed operator rows add to 96,994. The unallocated 7,355 is not assigned to an operator.

Sources, licensing and roles

POPLA counts are facts transcribed from its published annual report appendices with attribution; the reports are not marked as Open Government Licence material. DVLA KADOE releases are published under the Open Government Licence. The public data layer therefore exposes KADOE records and schema, not POPLA source counts.

Aliases are exact, manually reviewed source labels. Phase-0 roles are recorded as operates_for=self; no court or debt-collection activity is attributed through brand similarity.

Snapshot and parser hashes are carried on every derived record. A changed source must pass table reconciliation, fixtures and the rendered privacy scan before pages are rebuilt.