The Witness Who Was Certain
During a school incident, a student is knocked to the ground near the canteen. Two hours later, a teacher shows six Year 11 witnesses a photographic lineup. Witness A, who was standing three metres away and saw the incident clearly, identifies a student immediately with high confidence. Witness B, who was twenty metres away and had looked up after hearing shouting, hesitates for ninety seconds, then chooses the same photograph with low confidence. The principal says both identifications 'confirm each other'. A week later, Witness A sees the identified student on the news being questioned and becomes more certain. The identified student denies involvement and has an alibi supported by a timetable record.
1. Compare the reliability of the two identifications using memory research, and evaluate the principal's reasoning.
Read the worked answer
Reliability depends on encoding conditions and confidence calibration, not on agreement between witnesses. Witness A encoded under good conditions: close, clear view, attention directed at the event. Witness B encoded poorly: distance, attention drawn late, brief exposure. B's slow, low-confidence choice is consistent with a weak memory trace and a guess, so B's identification carries little evidential weight.
The principal's claim that the two 'confirm' each other is a reasoning error. Two witnesses can converge because of a shared weak cue, because they discussed the event, or because the lineup contained only one plausible match. Convergence is not independent corroboration unless the witnesses encoded independently and were tested separately.
Witness A's later certainty after media exposure is a post-identification confidence inflation, not new evidence. Confidence after feedback is a poor guide to accuracy; confidence at first test is more informative but still imperfect.
2. Recommend procedural changes for future lineups at the school and justify each in terms of memory.
Read the worked answer
Use a double-blind administrator so no one who knows the suspect can cue the witness. Warn witnesses that the offender may not be present, which reduces relative-judgement guessing. Present members sequentially rather than simultaneously to reduce comparison-based choosing. Record confidence immediately, before any feedback, to capture uncontaminated certainty. Interview witnesses separately and before any discussion to prevent memory conformity, and avoid repeated questioning that can reshape recollection. Each measure targets a known source of distortion: expectancy, relative judgement, feedback effects, and post-event contamination.