Flagged Exam Score: Statistics, Corroboration, and Due Process
A CAT candidate takes a computer-based exam at a regional test center. Psychometric review flags his record: he answered the first 20 items in 68 minutes with 12 correct, then the last 20 in 11 minutes with 18 correct. His wrong items match, in the same order, the wrong items of a candidate who tested 90 minutes earlier at the same center; the seating log places them in adjacent cubicles. Center video shows the candidate looking toward the adjacent partition four times but is otherwise inconclusive. The candidate states he rushed because of a family emergency and denies any communication. Program policy allows score invalidation for reasonable cause, requires written notice and an opportunity to respond, and gives final decision to the certification committee with appeal to an independent panel. The other candidate’s score is not yet flagged.
1. Does the evidence meet the 'reasonable cause' standard for invalidating the score? Separate the statistical anomaly from the corroborating evidence.
Read the worked answer
The statistical anomaly alone is not enough. A fast, accurate final section can reflect genuine test-taking strategy, prior knowledge, or a candidate who realized he was behind schedule. The anomaly becomes persuasive only when combined with independent corroboration.
Here the corroboration is substantial:
- Identical wrong-item pattern in the same order is a strong signal, but only if the two candidates saw the same item sequence. The auditor must confirm the test form and item order; if the forms were scrambled, matching wrong items is much more probative.
- Adjacency and timing place the candidates in a position and sequence consistent with observation or communication.
- Video showing four looks toward the partition is consistent with copying but is not conclusive by itself.
- The candidate’s explanation (family emergency) explains speed but not the matching error pattern.
The standard is reasonable cause, roughly a preponderance of the evidence, not criminal proof. On these facts, reasonable cause exists to hold the score and investigate, but the committee should not finalize invalidation until it reviews the test-form assignment, proctor logs, and the candidate’s written response. The correct judgement is 'insufficient for final invalidation on statistics alone; sufficient for a formal proceeding when combined with the matching pattern and adjacency.'
2. What interim actions should the program take, and how should it handle the other candidate whose score is not yet flagged?
Read the worked answer
Interim actions should protect the integrity of the record without prematurely branding the candidate.
- Place a hold on the score, not a public finding of misconduct. The candidate should be told in writing that the score is under review and why, at the level of detail needed to respond.
- Preserve evidence: test-form assignment, item-level responses, seat map, proctor notes, video, and the timing log. Do not rely on a summary.
- Interview the proctor on duty and ask whether the candidates interacted, whether dividers were in place, and whether any phone or note was seen.
- Do not disclose the other candidate’s identity to the first candidate. The first candidate does not need it to answer the allegations.
- Open a separate review of the other candidate using the same evidence. The fact that his score is not yet flagged is an administrative gap, not a reason to ignore the match. He must receive his own notice and opportunity to respond.
If the two reviews show a shared item pattern but no communication, consider a test-center environment problem-seat spacing, divider height, or proctor positioning-rather than assuming collusion. The program should avoid treating one candidate as the 'copyist' and the other as innocent without evidence. Both should be handled consistently, and the test center should be asked for a corrective plan.
3. What final outcomes and appeal rights are appropriate if the committee finds misconduct? What if it does not?
Read the worked answer
If the committee finds misconduct by a preponderance, the proportionate outcomes are:
- Invalidate the score for the affected attempt.
- Impose a defined waiting period before retesting-typically one attempt cycle-rather than a permanent bar unless there is a prior finding or a commercial cheating scheme.
- Require a supervised retest at a center with monitored seating and no adjacency to the other candidate.
- Record the finding in the conduct file with the evidence relied on and the sanction rationale.
A permanent revocation of eligibility would be disproportionate for a first, non-commercial incident unless the candidate refused to cooperate or the evidence showed an organized scheme.
If the committee does not find misconduct, it should release the score, remove the hold, and state in writing that no adverse finding was made. The statistical flag should not remain in the candidate’s file as an unexplained suspicion.
Either way, the candidate must receive the committee’s written decision, the evidence relied on (redacted as needed to protect the other candidate), and the appeal route to the independent panel. The appeal should be on the record, with a deadline and a standard of review-whether the panel reviews for procedural error, clear error, or hears the matter de novo. The test center should also receive feedback on seating and proctoring, because the environment contributed to the risk.