Reefer Exception Source Agreement: A Bounded Evidence Study

Published: September 14, 2026.

This study compares recorded fields across temperature-exception evidence sources in a fixed reefer-load cohort. It does not decide whether cargo was safe, whether equipment was fit, whether a condition caused damage, or whether any party complied with an obligation.

Research question and scope

The question is: when two or more qualifying sources describe the same reported exception, how often do their predeclared time and temperature fields agree within the study's tolerance? The unit is one exception event, not one file.

Define the load cohort, event trigger, expected source types, observation period, timezone rule, and tolerance before inspecting values. Examples of source types include device exports, display photographs, facility records, and messages. These labels describe provenance, not reliability rankings.

Evidence inventory

For each expected source, record whether it was received, its stable identifier, original value, unit, timestamp, timezone, resolution, receipt time, and link. Preserve reported statements separately from device observations. Retain raw sources and perform conversions only in derived fields.

An unreadable image, missing timezone, or rounded value may make a pair non-comparable. Do not force it into agreement or disagreement.

Methodology

Freeze eligible exception events independently of which sources are complete. Pair sources only when load, device or location, and event-window rules match. Convert units and timezones with published rules while retaining originals.

Classify each eligible pair as within tolerance, outside tolerance, or not comparable. Report expected sources, received sources, eligible pairs, comparable pairs, each outcome, and unresolved matches. Show the tolerance and the distribution of absolute differences rather than publishing only a pass rate.

Run sensitivity checks with at least one narrower and one wider tolerance when scientifically appropriate. Label these as sensitivity analyses, not replacements for the registered primary rule.

Quality and reproducibility

A second reviewer should reproduce source classification, event matching, conversion, and pair status for a stratified sample. Include missing timezones, rounded display values, corrected exports, and source conflicts. Keep adjudication decisions in a dated log.

Interpretation boundaries

Agreement may arise because sources share an upstream sensor or copied value. Disagreement can reflect calibration, placement, sampling interval, clock drift, rounding, transcription, or a mismatched event. The study cannot identify which explanation applies without additional validated evidence.

Limitations

The cohort may omit evidence received outside approved systems. Photographs can lack machine-readable context. Device exports may be revised, and clocks can drift. A source-agreement result does not measure actual product temperature across a trailer or throughout transit.

Source treatment

FDA and eCFR materials supply regulatory context. NIST materials supply measurement context. None defines or endorses this local agreement metric. Retrieval dates and direct links remain attached to the study record.

Conclusion

Source agreement can expose record conflicts when the comparison rules and missing population are explicit. Its proper use is evidence-quality review, not a cargo, maintenance, compliance, or liability conclusion.

References

1. https://www.fda.gov/food/food-safety-modernization-act-fsma/fsma-final-rule-sanitary-transportation-human-and-animal-food, retrieved 2026-09-14
2. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-1/subpart-O, retrieved 2026-09-14
3. https://www.nist.gov/pml/owm, retrieved 2026-09-14