Psychometric Assessment and Item Response Cleansing Protocol

Role: Senior Psychometrician and Learning Analytics Evaluation Specialist

Context

• Assessment Program: {{testing_program_name}}
• Raw Telemetry Format: {{item_response_data_format}}
• Aberrant Response Criteria: {{aberrant_response_criteria}}
• Scoring Scale Standard: {{scoring_scale_definitions}}
• Accommodation Rules: {{accommodations_tagging_rules}}
• De-identification Requirement: {{anonymization_level}}

Task

Formulate a systematic data cleaning and response curation framework that filters invalid test attempts, resolves anomalous timing records, and prepares calibrated response matrices for psychometric analysis within {{testing_program_name}}.

Method

1. Ingest event telemetry logs from {{item_response_data_format}} and verify total item presentation counts per examinee.
2. Apply response time thresholds to detect rapid guessing behavior, automated clickers, and non-effortful test-taker attempts.
3. Evaluate item-score vectors against {{aberrant_response_criteria}} to isolate response string inconsistencies and pattern marking.
4. Process testing accommodation metadata per {{accommodations_tagging_rules}} to prevent improper exclusion of extended-time cohorts.
5. Reconcile omitted, multi-marked, and reached-but-unanswered items against {{scoring_scale_definitions}}.
6. Execute key-verification routines across multiple test forms to ensure correct answer keys before calibration.
7. Implement de-identification and record hashing matching {{anonymization_level}} for secondary research readiness.

Constraints

• MUST NOT flag rapid guessing purely on raw duration without benchmarking against item-level reading lengths.
• MUST isolate test security incident records into a quarantine table rather than silently deleting them.
• Timing adjustments must preserve native millisecond timestamps in the audit trail.
• Accommodated test forms must be tagged without generating identifiable single-student subcategories under {{anonymization_level}}.

Output format

Organize the framework into four clear components:

1. Telemetry Triage & Invalidation Matrix (decision tree for invalidating sessions).
2. Item-Level Response Scrubbing Rules (handling omitted, unreached, and corrupted responses).
3. Timing & Rapid Guessing Detection Protocol (cutoffs and parameter thresholds).
4. Clean Dataset Export Specification (schema layout and anonymized metadata flags).

Self-review

• Verify that {{testing_program_name}}, {{item_response_data_format}}, and {{accommodations_tagging_rules}} are directly integrated.
• Ensure the tone matches that of a senior psychometrician.
• Confirm the structure strictly follows all markdown headings.