Pitfalls in Deception Detection

Eliciting new information from eyewitnesses via
repeated interviews: How does it impact accuracy
and credibility?
Amina Memon
Royal Holloway, University of London
Geralda Odinot
University of Leiden
David LaRooy
University of Abertay
We know that.....


Witnesses may be questioned on multiple
occasions over lengthy delays (LaRooy Katz,
Malloy & Lamb, 2010).
That “different” information is often produced
across repeat interviews raises concerns among
legal professionals
Consistency and Accuracy: Impact on
Witness Credibility
Judges and Legal scholars deem witness consistency to be
one of the most important measures of witness credibility
Witness credibility is often attacked by highlighting
inconsistencies in statements including the reporting of
new previously unreported details.
Self-contradiction is believed to be a result of a defect in
memory or honesty of a witness
Fisher, Brewer & Mitchell, 2009
In an applied setting
We need to look carefully at what is
recalled and not just amount (Koriat and
Goldsmith, 1994)
 New information should not be viewed as
inconsistent or contradictory or indicate
that a witness is not credible or inaccurate
(Gilbert & Fisher, 2006).

Lengthy delays


The delays which occur during an
investigation may compromise the quality
of the evidence that can be obtained from a
witness
We could only sample short delays but we
wanted to see if even a delay of 1 week
would make a difference.
Research Questions

How is eyewitness recall (quality and quantity)
influenced over repeat interviews influenced by:
Delay- forgetting increases over time so should a
repeated interview be conducted as soon as
possible?
Interview (Cognitive or Structured) across repeat
interviews?
Method




107 college students
Video event of stalker who follows victim
and enters her house
One female interviewer fully trained in the
Structured (SI) and Cognitive Interview (CI)
Each witness was interviewed twice
Interviewer Training
The interviewer underwent 2 day
training in the Cognitive and
Structured Interview methods with
practice and feedback.

COGNITIVE
Rapport & Ground
Rules
Report Everything
Context
Reinstatement
Free Recall
Question phase

STRUCTURED
Rapport & Ground
Rules
Free Recall
Question phase
Condition 1
Immediate & 2
days
CI-CI
Condition 2
SI-SI
Delayed for 7 & 9
days
Condition 3
CI-CI
Condition 4
SI-SI
Results


1.
2.
Number of Correct Details in the first interview
The number of NEW unique details across the
two interviews
Interview type (CI or SI)
Interview timing (Early- Delayed for 7 days)
Number of correct details at interview 1
80.000
70.000
60.000
50.000
Early
40.000
Delayed
30.000
20.000
10.000
.000
CI
SI
Number of correct details as a function of interview timing:
Early (immediate and 2 days) versus delayed (2 and 7 days)
80.0000
70.0000
60.0000
50.0000
Interview 1
40.0000
Interview2
30.0000
20.0000
10.0000
.0000
Early
Delayed
New (previously unreported) correct details by interview type and
timing of interview
18.00
16.00
14.00
12.00
10.00
Early
Delayed
8.00
6.00
4.00
2.00
.00
CI
SI
Repeated interviews
Total number of errors across the two interviews
25
20
15
Early
delayed
10
5
0
CI
SI
Repeated interview
Note: most of these errors occurred during the questioning phase.
Results

Next we will look to see if details (correct
and incorrect) were consistently reported
across the two interviews
Number of correct details consistently reported across the two
interviews
60
50
40
Early
30
delayed
20
10
0
CI
SI
Repeated interview
New Incorrect details by interview type and interview timing
(early/delayed)
8.00
7.00
6.00
5.00
Early
4.00
Delayed
3.00
2.00
1.00
.00
CI
SI
Repeated interviews
Information gain from repeated testing


Finally we looked at the cumulative recall
across the two interviews
We counted the repeated details only once
and added any new previously unreported
information.
Total number of unique details* across the two interviews (*total in first
interview plus new items at time 2)
100
90
80
70
60
Early
50
delayed
40
30
20
10
0
CI
SI
Repeated interview
Conclusions 1



Clear effects of early interview and repeated
testing- gain in new correct details with small
increase in errors
CI advantage in correct details and consistent
details- important for credibility
CI small increase in reporting of erroneous detailthere were fewer errors in the free recall phase
most arose in the questioning sequence.
Conclusions 2


Benefits of Repeated interviews-increase in
the number of unique (cumulative) details
Early interview advised but we need to
examine the effects of longer delays
This research was funded by a European Union (FP6) grant
New Meta-analysis of the
Cognitive Interview
http://www.pc.rhul.ac.uk/sites/rheg/