Measuring Completeness in Truck Detention Event Data

Published September 2, 2026.

Research question

Which completeness measures show whether detention records can support a reproducible duration analysis? This brief proposes a carrier-level measurement design. It does not publish an industry benchmark and does not provide legal, safety, maintenance, accounting, employment, or contractual advice.

Evidence scope

FHWA performance-measure material informs the event-based structure, BTS supplies transportation context, FMCSA material helps define the boundary between administrative measurement and safety authority, and current Title 49 text is included as a primary regulatory reference. Public sources do not contain the private records required to calculate a carrier result.

The unit of analysis must be fixed before inspection. Retain the raw identifier, source system, event timestamp, timezone, observer or submitting party, and revision history. Derived fields belong beside the raw evidence rather than replacing it.

Proposed method

Construct five mutually exclusive classes: paired events, arrival-only, release-only, conflicting pairs, and records with neither event. Write the eligible population, opening event, closing event, exclusions, duplicate rule, and unresolved-record treatment before calculating the primary result.

For detention event completeness, the main finding is: Report paired arrival and release events separately from one-sided records, source conflicts, and unverifiable timestamps. A result should remain reproducible when another analyst applies the written rules to the same frozen source extract.

Denominator design

Use each eligible unit once in the primary denominator. Report total eligible units, complete records, incomplete records by missing field, conflicts, exclusions by reason, and units that cannot be matched to a stable identifier. Publish counts beside percentages.

Do not remove difficult records after seeing the result. If a sensitivity analysis uses a narrower population, label it separately and retain the primary denominator. This prevents better documentation coverage from being mistaken for better operational performance.

Source conflicts and missingness

When two sources disagree, preserve both values, their provenance, and their capture times. Apply only a preapproved hierarchy; otherwise classify the record as unresolved. Missingness may cluster by facility, channel, shift, customer process, or system change, so report it by relevant strata when sample size and privacy rules permit.

Never manufacture an event from a later summary. A note that says an action happened is not equivalent to a native event unless the method explicitly defines it as an accepted secondary source.

Quality controls

Sample complete, incomplete, conflicting, and extreme records. Reopen the evidence and verify identifiers, timestamp order, duplicates, exclusions, and derived fields. Record the reviewer role and date. Corrections should create a versioned output.

Test midnight crossings, timezone changes, reopened cases, repeated alerts, superseded documents, late entry, and impossible event order. An impossible interval is a data-quality exception, not a number to average.

Interpretation boundaries

The measure can direct a queue review, documentation improvement, owner follow-up, or narrower study. It cannot on its own prove causation, compliance, negligence, fraud, employee performance, contractual entitlement, equipment condition, or the correct operating response.

TruckingVA may assemble supplied records, apply documented rules, calculate approved fields, and route exceptions. The carrier's qualified or authorized owner decides what the result means and which action follows.

Reporting template

Publish the research question, eligible population, time window, timezone, event definitions, source hierarchy, exclusions, duplicate rule, missingness, conflicts, complete-case count, primary result, strata, sample review, version, and limitations. Retain the query or transformation steps with the frozen input reference.

The conclusion should state both support and limits. For this analysis: Report paired arrival and release events separately from one-sided records, source conflicts, and unverifiable timestamps. A scope or definition change requires a new dated version rather than silent replacement.

Limitations

Public resources describe broad freight measurement and regulatory contexts, not TruckingVA client operations. Private systems differ in labels, event capture, access rules, and retention. Missing records may not be random, and a recorded timestamp may represent data entry rather than occurrence. The design therefore supports transparent local analysis, not a universal benchmark.

Sources

1. https://ops.fhwa.dot.gov/freight/freight_analysis/perform_meas/index.htm, retrieved 2026-09-02
2. https://www.bts.gov/freight-facts-and-figures, retrieved 2026-09-02
3. https://www.fmcsa.dot.gov/safety, retrieved 2026-09-02
4. https://www.ecfr.gov/current/title-49, retrieved 2026-09-02