The Evidence Chain from Observation to Decision

Learning outcomes

After studying this subject, the learner will be able to:

• define the adjudicative evidence chain and identify its links;
• distinguish primary data, derived values, interpretation, and authorized decision;
• trace every final value to its source and transformation history;
• detect breaks, erasure, or incorrect transfer within a record;
• construct an audit history protecting the original state and justifying change; and
• evaluate decision strength through evidence completeness rather than final appearance alone.

Evidence-chain concept

The evidence chain is the documented path through which an observation moves from its creation to its use in a final decision. It is the structural basis of institutional trust. If every link is known, the outcome can be explained. If a link is missing, the final number stands without a clear origin.

The chain begins with primary CJE material: observation, raw score, confidence, time, and context. The value may then pass through verification, entry correction, scale normalization, weighting, or reconciliation. Professional interpretation and authorization follow. Every transformation must be named, timed, and capable of arithmetic reproduction or logical review.

The path is represented as:

E_raw is the original evidence and each τ is a declared operation. The value of the relation lies in requiring every step to be known instead of presenting a final result without history.

Evidence layers

CAMS separates four layers:

• Primary evidence: what the judge directly registered, including score, observation, and confidence.
• Derived evidence: a value produced by a declared calculation, such as a mean, repeat gap, or reliability indicator.
• Interpretive evidence: professional meaning assigned to data and calculations.
• Authorizing evidence: the decision approved by the empowered role under policy.

A derived value does not replace its primary data. If the panel mean is 81, authorized reviewers must still be able to inspect the scores producing it. If judge-reliability weighting is applied, the formula version, inputs, and weights remain available.

Chain-integrity rules

A sound chain follows defined rules:

• Every value has one known source or a specified source set.
• Every transformation records operation name, version, time, and operator.
• Primary evidence is not overwritten by a corrected state; both are retained.
• Interpretation is not inserted into the raw-observation field.
• Manual entry undergoes comparison or dual review.
• Authorized decisions link to the governing rule or policy.
• The system records who viewed, modified, and approved the result.

The goal is not to eliminate every human error. It is to make error visible and correctable before it becomes final harm.

Evidence strength and completeness

Chain completeness can be examined through six questions:

• Do we know who created the data?
• Do we know the sample and event to which it belongs?
• Do we know when it was created and changed?
• Do we know the calculations through which it passed?
• Does the original state remain preserved?
• Do we know who interpreted and authorized the outcome?

When all answers are available, the chain is traceable. Sensory accuracy requires additional indicators. Traceability alone does not guarantee accuracy, but it prevents accuracy from being claimed without a record.

Mind map

Worked numerical case

A judge closed a raw score of 84 in the digital card. During export to an external worksheet, the value appeared as 48 because the digits were reversed. Other panel scores were 82, 85, and 83.

Using the incorrect value produces:

Using the original value produces:

The difference is nine points and may change competitor ranking. The audit history shows 84 in the original event, identifies the export as the transformation point, and proves that the external file timestamp followed closure. The calculation is corrected, but the value 48 is not deleted from the error history. It is retained as a failed transfer with operation name and correction time.

The case demonstrates the difference between correction and concealment. Correction returns to source while retaining history. Concealment prevents the institution from understanding or preventing recurrence.

Applied laboratory: reconstructing a decision

Purpose

Train the learner to trace a final outcome through events, inputs, and transformations and identify the link that introduced error.

Laboratory material

The learner receives raw cards, a data export, a calculation sheet, reconciliation minutes, and an authorization record. Deliberate discrepancies include an incorrectly transferred score, confidence entered as score, an outdated formula version, and an unnamed modification.

Procedure

• Link every final value to its original CJE identifier.
• Compare digital inputs with closed records.
• Reperform calculations under the declared formula version.
• Identify the first appearance of every discrepancy.
• Classify the discrepancy as entry, transformation, interpretation, or authorization.
• Write a correction path preserving every previous state.

Chain-audit card

Practical assessment

The learner receives a final result and its supporting files and prepares an audit report identifying the source of every value, reperforming calculations, locating breaks, and proposing institutional correction.

Assessment rubric

Core terms

• Evidence chain: path linking original data to the authorized decision.
• Primary evidence: event data as registered by its source.
• Derived value: calculated output based on known primary evidence.
• Transformation: operation changing representation, aggregation, or weight.
• Non-erasure: preservation of an earlier state following correction.
• Audit history: sequence showing access, modification, and authorization.
• Reproducibility: capacity to obtain the same derived value from the same inputs and rule.

Scientific conclusion

An outcome is not reliable because it looks precise or originates from a large panel. Institutional trust begins with the ability to move backward from decision to calculation, from calculation to data, and from data to event and observation. The evidence chain makes every transformation visible and turns error from a hidden risk into an occurrence that can be located, corrected, and used for learning.