Roadpeace Conference: 22 September 2010
Protecting Eyewitness Evidence
Dr Lorraine Hope
Reader in Applied Cognitive Psychology
University of Portsmouth
Incidents, crimes, crashes
• Investigations are an attempt to reconstruct a past event,
incident or episode:
• Physical evidence (fibers, DNA, skid marks, impact analyses)
• protocols for collecting, preserving and interpreting physical
• Optimal ways to avoid contamination
• Physical evidence is often ‘circumstantial’
• Witness evidence (memory & ID)
• Typically collected by non-specialists in human memory
• Protocols for collecting, preserving and interpreting eyewitness
evidence has not always incorporated scientific research
• Can directly link a particular suspect to an incident
Memory as (Re)Construction
 Recall - not an exact replica of original events
 Recall - a construction built and rebuilt from
various sources
 Reconstruction process = vulnerability to error &
 Vulnerability can be reduced by ‘careful handling’
of memory by investigators
When are witness memory errors most likely?
 When original memory trace is weak
 Poor encoding of incident
 Delay
 Forgetting results in ‘gaps’ or a hazy recollection
 Misinformed by a co-witness
 At the scene or subsequently (multiple retellings & elaboration)
 Exposure to misleading press/media
 Misleading questions
 Poor interview technique (formal or informal interviews)
Delay and Decay: Does it matter?
• Empirical evidence for rapid memory decay
– steep forgetting curve
Forgetting Curve
Memory Decay: What does the research show?
• Completeness & accuracy of recall decreases as the delay
between witnessing & recall increases (Penrod et al., 1982; Wixted &
Ebbesen, 1991, 1997)
• Mock witnesses recalled 43% fewer details after a 3-week
delay before being interviewed (Turtle & Yuille, 1994)
• Information less accessible as memory traces (and
associations between traces) are weakened (Ayers & Reder, 1998)
• Memory fades from the specific to the general or gist
(Goldsmith et al., 2005)
Consider the Problem..
Serious incident: multiple witnesses
Limited resources = delay before full interview
During this period, eyewitness memory is:
i) prone to decay
ii) vulnerable to the influence of post
event information
Immediate recall as a solution?
• A (good quality) initial recall attempt has positive
effects on memory:
– Retrieval activates memory traces & strengthens
associations between them
– A recall attempt prior to misleading PEI (post
event information) can help protect the original
Consider the Problem…
Serious incident: multiple witnesses
Limited resources = delay before interview
A system or tool that allows witnesses to record
their own memories shortly after encoding
Our objective?
Develop a ‘Self-Administered Interview’ (SAI)
– A self-explanatory recall tool for witnesses to complete
themselves as soon as possible after witnessing a crime or
incident requiring further investigation
The ‘Self-Administered Interview’
• A recall tool to facilitate witnesses when remembering &
reporting information, comprising;
– Carefully designed information regarding what is expected
from witnesses
– Guidelines and questions that provide retrieval support
– Adapts components of the ‘gold standard’ Cognitive Interview:
• Mental Reinstatement of Context
• Report Everything
– Spatial layout/sketch element
Study 1: Developing & Testing the SAI
1. Identify optimal SAI format
2. Compare with other recall methods
Experimental Conditions
1. Free Recall
2. SAI
3. Cognitive Interview
– General public recruited to act as witnesses (N=55).
– Viewed simulated non-violent video event (2mins 40secs).
10 2 6
7 10
5 8
10 20 30 40 50 60 70 80 90 100
Mean no. of details reported
Overall Correct items: F (2,54) = 24.72, p < .001
No difference in accuracy rates (>90% in all groups)
Study 2: Incorporating a delay
• General public recruited to act as witnesses (N=42).
• Allocated to SAI or Control condition.
Session 1
– All pps viewed simulated crime event on video
– SAI participants record recall after event
Delay = 1 week
Session 2
– All participants provide full free recall
Results (Delay = 1 week)
3 7
Mean no. of details reported
Total correct: F (1,40) = 5.41, p = .02
No difference in Incorrect or Confabulated responses
Summary of main findings
• SAI strengthens memory for a witnessed event
• SAI instructions produce significantly more information
than a free recall (‘tell me what you saw’) instruction
• Subsequent studies confirmed that completing an SAI
after witnessing a crime
Minimises forgetting over a delay
Maintains high accuracy rates
Enhances performance in a subsequent CI
Protects against memory distortions caused by exposure to
misleading questions and PEI
Field Trials
Question: Does the SAI deliver in the field?
1. Examine the quantity & quality of information provided by
witnesses using the SAI
2. Gauge ‘usability’ of the SAI by witnesses and officers, to
lead future development of the tool
3. Identify witnessing contexts in which the SAI provides most
benefit to police investigation (e.g. multiple witnesses,
peripheral witnesses, etc.)
Field Studies of the SAI endorsed by ACPO
Case study: Greater Manchester
Case study
• Perpetrators
– 3 males on a motorbike, collide with a car at 60mph
– Passengers thrown from bike and hit a bus
– 1 fatality, 1 seriously injured, rider tries to escape scene
• Investigating officers (x 2)
• Witnesses (x 16)
– 8 witnesses interviewed by police shortly after the collision
– SAI administered to 8 witnesses
Quantity & quality of information obtained
• 88% of selected witnesses completed an SAI.
• Of these:
– All witnesses gave a detailed free recall account of the
incident (average of 54 lines of text / 2.5 A4 pages of info.)
– In addition, most witnesses completed all other relevant
sections of the SAI, providing information about:
• Perpetrators’ actions & descriptions
• The quality of their own view/testimony
• Other potential witnesses who had been present
– 5 witnesses provided a sketch, appended with descriptors
• SAI identified three additional important witnesses
Quantity & Quality of SAI information
• Reports were detailed:
– "Male, white, shaved head, wearing long sleeved jumper and
balaclava, 20 years old, around 11 stone, no facial hair, olive
skin, cut on his head”
– "The motorbike had three riders, they seemed to be sitting in
height order with the smallest at the front. I noticed that they
were riding at speed on the wrong side of the road and
overtaking the slower traffic. None of the riders were wearing
helmets, instead they had their faces covered, in an
intimidating manner with balaclava type clothing"
• High level of corroboration between statements
End user evaluation from investigators
• Usability was evidenced by proper use and understanding
of instructions for use.
• Feedback from the investigating officers:
“Out of all the SAIs issued only one person failed to complete
the form. The forms that were completed were comprehensive,
and a few I would say were more detailed and beyond the
standard that some officers sometimes produce. Many
contained useful detail in proving the case, and in the case of
one witness, crucial evidence in terms of the rider and
passengers’ actions leading up to the event which without doubt
will show the joint venture in this case. The SAI forms have led
to another three crucial statements being taken.”
Use of the SAI within force
• Adopted as force form by Greater Manchester Police.
Summary of main findings
• The SAI is a standardised, empirically tested witness
reporting tool
– Based on sound theoretical principles of memory
– Allows numerous witnesses to give evidence simultaneously and
– Elicits forensically relevant information with high level of detail
from witnesses
– Facilitates the effective prioritisation of witness statements
• Applied Caveats:
- Not suitable for all crimes or incidents
- Not suitable for all witnesses (foreign language, literacy,
social support/vulnerable)
• Witness memory for any incident is vulnerable to both
decay and distortion
• SAI offers clear benefits for quality and quantity of
information provided in witness accounts
– Distinct from previous reporting forms, based on psychological
principles of Cognitive Interviews and empirically tested
• SAI – part of an investigations toolbox
- Preserves and protects witness memory and elicits timecritical information effectively
Dr Lorraine Hope
University of Portsmouth
[email protected]
The British Academy & ESRC